Sie sind auf Seite 1von 32

Psychological Bulletin 1999, Vol. 125, No.

2, 223-254

Copyright 1999 by the American Psychological Association, Inc. 0033-2909/99/$3.00

Enduring and Different: A Meta-Analysis of the Similarity in Parents' Child Rearing


George W. Holden and Pamela C. Miller
University of Texas at Austin
The assessment of child-rearing beliefs and behavior has predominantly focused on qualities and characteristics believed to reflect consistent, enduring qualities of parentingthe similarity in child rearing. This review evaluates the evidence for similarity and differences among 3 types of child-rearing data and includes comparisons across time, children, and situations. Both relative stability and mean level differences were found in all 3 domains. The most similarity was found in the across-time arid across-children domains, although it depended on the child-rearing construct and methodology used. It is argued that attention to the variability and change in child rearing must be incorporated into theoretical models of parenting to better understand the nature of child rearing and, in turn, parental influence on children's development.

Is a mother characterized as sensitive when her son is an infant, equally sensitive when he reaches preschool, elementary school, or high school? Does a father exhibit the same amount of attention, love, and firm discipline to his 5-year-old son as to his 8-year-old daughter? And is a parent's irritability in the supermarket diagnostic of the quality of caregiving provided in the home? Each of these questions represents a different face of the issue concerning the similarity of parental behavior. This question is critical for applied reasons as well as for discovering the role that parents play in their children's development. The rigidity with which individuals are tied to their childrearing beliefs and practices is an important issue for clinical psychologists who work with children exhibiting behavior problems. If parents are to be effective therapists for their children (Kazdin, 1987) or adolescents (Schmidt, Liddle, & Dakof, 1996), then they need to adopt and maintain new behavioral repertoires. Similarly, developmental, social, school, and community psychologists working to prevent a range of social problems including child injury (child maltreatment, poisoning, accidents), youth morbidity (substance abuse, obesity), school failure (low performance, dropouts), teenage sexual promiscuity (STDs, pregnancy), and antisocial behavior (aggression, gang membership) are concerned with the issue of modifying parental behavior. According to human behavioral geneticists, the similarity of parenting is an especially timely theoretical issue. Plomin (1990, 1994) and others (e.g., Hoffman, 1991) have argued that to under-

George W. Holden and Pamela C. Miller, Department of Psychology, University of Texas at Austin. Pamela C. Miller is now at the Department of Psychology, University of Houston. Preparation of this article was supported in part by a National Institute of Child Health and Human Development Grant 1 RO1 HD26574-01A1. We thank Elizabeth Thompson, who served as a second coder, as well as Rebecca Bigler, Anne Cameron, Ted Dix, Judith Langlois, William Swann, Ross Thompson, and John Weisz for their helpful comments. Correspondence concerning this article should be addressed to George W. Holden, Department of Psychology, University of Texas, Austin, Texas 78712. Electronic mail may be sent to holden@psy.utexas.edu.

stand phenotypic differences among siblings, parents' roles in creating shared and especially nonshared environments need to be explicated. The concept of nonshared child-rearing environment recognizes that parents may interact with or structure a child's physical and social world differently from that of his or her sibling. Although such differences may be derived from parental characteristics or elicited by the child's attributes or predispositions (Scarr & McCartney, 1983), siblings nevertheless encounter distinct as well as similar child-rearing experiences (e.g., Anderson, Hetherington, Reiss, & Howe, 1994; Brody, Stoneman, & McCoy, 1994). This particular theoretical emphasis on child-rearing differences comes in stark contrast to the historically prominent view of parents. Theories and popular beliefs about child-rearing effects posit that parents influence their children in a variety of ways (Holden, 1997; Maccoby & Martin, 1983). One way is through habitual patterns of interaction. "In theories of child-rearing, parental behavior is assumed to have effects on children through a history of experiences. There is faith [italics added] that, overtime, parental influences lead to generalized behavior tendencies that have some durability" (Radke-Yarrow, Zahn-Waxier, & Chapman, 1983, p. 502). Social-learning approaches (e.g., Patterson, 1982) as well as family-systems theories (e.g., Minuchin, 1985) both assume that parental effects occur through recurrent interactions. "We can assume that the family system, like any system, has self-stabilizing properties. . .Families stabilize around habitual patterns of interaction; thus there is continuity over time in the familial forces that support the distinctive personality patterns of individual children" (Maccoby, 1984, p. 326). If children's developmental outcomes are affected by recurrent interactions, then stable patterns of child rearing are assumed to be responsible for links between parenting practices and child outcomes. Invariant views of child rearing and families are appealing in that they provide a parsimonious model of parenting. Methodologically, it is also far more convenient for researchers to embrace such a view. If parental behavior were variable or changing, then making assessments of parental behavior would be much more difficult, time consuming, and costly. Even more fundamental is
223

224

HOLDEN AND MILLER

the theoretical havoc that could be associated with an unstable view of child rearing. The assumption of similarity in parenting means that child-rearing behaviors assessed at one point in time can be assumed to reflect prior, as well as future, child-rearing experiences. In turn, one can relate those child-rearing data to indexes of child behavior or outcomes as a way of assessing parental influence (e.g., Baumrind, 1971). These theoretical and empirical positions have contributed to a common assumption of and focus on the unchanging characteristics of child rearing. An example of this assumption made explicit is ". . .parents have fundamental, pervasive, and enduring childrearing orientations" (Roberts, Block, & Block, 1984, p. 595). Similarly, Dunn, Plomin, and Nettles (1985) concluded that maternal behavior was "strikingly consistent" toward two siblings when they were observed at the same age. Another example of the proclivity of researchers to adopt a stable view of parenting is the prominence of the trait approach to parenting. This orientation toward similarity in child rearing has been so central to conceptions of child rearing that it may have precluded reviews on the topic; we were unable to locate any. The fundamental problem with assuming similarity is that it may impede asking questions about variation and change. Important inquiries focusing on the origins of child-rearing behavior, circumstances surrounding change, and efficacious ways of modifying parental behavior have not received the attention they deserve. Given the importance of the assumption of similarity, as well as the long-term debate in personality psychology over the situational specificity of adult behavior (e.g., Bern & Allen, 1974; Epstein, 1979; Mischel, 1979; Mischel & Shoda, 1995), the lack of critical attention devoted to this issue as it pertains to child rearing is somewhat puzzling. This neglect is even more ironic given that the issue of stability and change has long been recognized by developmental psychologists to be the core issue of the discipline (Cairns, 1979; Kagan, 1971). However, that discussion has been limited to only one side of the developing dyadthe children. It is our contention that it is time to broaden the discussion of stability, change, and variation to include parents. Systematic variability and change in parenting has not been adequately appreciated or investigated. In an attempt to redress that omission, the purpose of this article is to review the empirical literature concerning the similarity of parenting and then discuss the implications of those findings. To do that, we begin by clarifying the terminology, then we review the evidence as found in the literature. The final section addresses implications of the analyses and directions for future research. Taxonomies of Similarity in Parenting

across children within a family? Although some child-rearing manuals recommend that parents should act the same way toward each offspring to minimize sibling rivalry (Faber & Mazlish, 1987), to what degree is child rearing affected by a child's age, gender, temperament, and specific behavior? The third aspect of similarity concerns whether child rearing varies systematically across situations or different contexts. More specifically, is child rearing affected by a range of contextual variables, such as the time of day, month of the year, the location of the interaction, the presence of others, or the nature of the interaction? Child-rearing similarity in each of these three domains is discussed further below.

Child Rearing Across Time


Longitudinal analyses focus on the stability of a behavior or psychological construct. Stability is a slippery concept when applied to human development as it embodies multiple forms and terminologies. Most commonly, stability is defined as an individual's tendency to remain behaviorally similar over time (e.g., Baumrind, 1989; Cairns, 1979; Emmerich, 1964; Rutter, 1984). Several types of stability have been identified; for the purposes of this article, discussion is limited primarily to two kinds. Absolute stability refers to behaving in the same way on two or more occasions. For example, do parents hugor hittheir preadolescents as frequently as they did when their children were toddlers? Evidently, given the rapidity of ontogenetic changes in children and their considerable consequences on parenting, such an orientation on surface-level behaviors is too limited to adequately capture what may be continuous about parental behavior. Questions concerning consistency in absolute values focus on the mean level of the variable and therefore utilize analyses of variance (ANOVAs) or t tests to determine whether significant change has occurred. An alternative type of similarity, interindividual consistency, centers on the relative placement of an individual within a group of people. Here the question becomes Does the parent maintain over time his or her relative position compared with other parents? Such an analysis takes into account developmental change in the child. The appropriate statistical index of this type of similarity is the Pearson correlation coefficient, as its computation relies heavily on the relative ranking between individuals. Assessments of interindividual consistency can focus on surface-level behaviors but also may take into account developmental change in the parent and child. Displaying different behaviors (e.g., kissing an infant, talking affectionately to a preschooler) that reflect the same underlying construct (e.g., positive affect) is an example of this second type of stability. If the time span between two assessments is long enough, it is unrealistic to expect the same behavioral manifestation of the construct. Indeed, no parent continues to provide the same caregiving behavior with normally developing children. Good parents cannot continue to engage in the same behavior and maintain the appellation of effective parents. Rather, such parents must modify their behavior in response to contextual circumstances and ontogenetic changes in their children. Consequently, it is likely that measures of interindividual consistency represent a more appropriate approach for assessing similarity in child rearing. A third type of continuity is that of functional equivalence,

Domains of Similarity
The question How similar is parenting? actually entails three components. The most commonly examined approach to the question concerns the stability of parental behavior across time: Do parents rear their children similarly week to week, month after month, and across the years? In addition to that longitudinal question, there are two other important questions inherent in the issue of child-rearing similarity. To what extent do parents rear their offspring in the same way? Given that 80% of U.S. families have two or more children (Dunn, 1992), how similar is parenting

SIMILARITY IN PARENTING

225

whereby the functional properties of behaviors or parent-child relationships do not change (Bern & Funder, 1978). Although the particular behavioral expression may differ, the underlying function of the behavior remains the same. Examples of functional relationships include well-attuned versus disharmonious dyadic states (Isabella & Belsky, 1991; Leyendecker, Lamb, Fracasso, Scholmerich, & Larson, 1997), cooperative relationships (Parpal & Maccoby, 1985), seductive interaction patterns (Sroufe, Jacobvitz, Mangelsdorf, DeAngelo, & Ward, 1985), or discordant or rejecting relations (Rohner, 1986; Stattin & Klackenberg, 1992). The most frequently investigated example of functional equivalence comes from research examining parent-child attachment relationships. Waters (1978) found that discrete attachment behaviors (e.g., smiling, approaching, holding) showed little stability. However, when the functional relationship was analyzed (i.e., secure vs. insecure attachments), there was impressive stability of attachment classifications across a 6-month period. Unfortunately, subsequent studies have not found the same degree of continuity in attachment relationships. Thompson (1998) reported that out of 15 studies addressing this question, a broad range of short-term stability estimates was reported. It appears that a major reason for variations in the attachment relationship stems from changes in life circumstances, such as alterations in employment status or stress levels (e.g., Thompson, Lamb, & Estes, 1982; Vaughn, Egeland, Sroufe, & Waters, 1979).

Evidence for Similarities and Differences in Child Rearing


Inspection of the child-rearing literature provides evidence for both similarities and differences. At a theoretical level, some approaches toward parents assume a largely static and unchanging view, whereas others highlight the mutable nature of child rearing. Conflicting information about similarity and variation in parenting can also be found in empirical research concerning the determinants of parental behavior.

Two Contrasting Models of Child Rearing


Different underlying approaches to the study of parenting are likely to contain discrepant views about the similarity in child rearing. For example, the trait and child-effects approaches represent two sharply contrasting models (Holden, 1997; Maccoby, 1992). The trait approach is the oldest and most prominent approach to the study of parents. It depicts the essential ingredient of parenting as being found in recurrent patterns of behavior; the particular pattern embodied by a parent therefore represents the essence of that parent's child rearing. A series of different parenting traits has been postulated as early as 1931 and variously labeled as syndromes, types, styles, or patterns. Examples of these typologies and their creators are listed in Table 1. Several of the typologies, most notably Baumrind's triarchy of child-rearing styles and those assessing acceptance and rejection, continue to be actively investigated (e.g., Rohner, 1986; Steinberg, Lamborn, Dornbusch, & Darling, 1992). Each new categorization of parents has served to promote the focus on the trait-like and stable characteristics of parenting. A trait is intended to represent the essence of the parent culled across time and situations. Like any personality trait, a parental style is thought to provide a better representation of that parent than any single or group of particular behavior(s) because it is a summary variable. As such, the trait approach is intended to reflect not one interaction but rather the "season's average." One hazard with the trait approach is that it can lead to erroneous conclusions of similarityacross time, children, and situations. As Darling and Steinberg (1993) recognized in their discussion of traits, little is known about the similarity of parenting styles across time. Thomas and Chess (1977) were sharper in their concern:
Our quarrel with the [trait] formulations arises when they assume an "all or nothing" character, when they are credited with exclusive significance in determining the child's psychological development. Similarly, these concepts can be criticized when the parental attributes are reined and given a global dimension. It is insufficient and inaccurate to characterize a parent in an overall, diffuse way as "rejecting," "overprotective," "insecure," etc. A parent may be unsympathetic and antagonistic to certain of the child's characteristics and accepting and approving of others; overprotective and restrictive of some of the child's activities but not of others; insecure and unsure in specific areas of child-care responsibilities and self-confident and assured in others, (pp. 78-79)

Child Rearing Across Children and Across Situations


Change is the term commonly used as the antonym to stability or consistency. However, change implies a permanent modification or transformation. For some questions about parenting, the question is not Did parenting change? but rather Are parents capable of modifying their behavior under the circumstances? Thus, variability more accurately reflects the focus of concern. To what extent do parents vary their child-rearing behavior or thinking when interacting with their different children or under differing situations? This capacity to modulate or modify child-rearing thinking or behavior has recently begun to attract more attention from researchers, as is discussed below.

Methodological Issues
It is now well recognized that results from investigations concerning the components of stability and change are closely linked to methodology (e.g., Radke-Yarrow, 1989). The level of analysis (e.g., global attitudes or specific behavioral responses to a child) is likely to affect the degree of similarity found. Single acts of specific, discrete behavior are often highly unstable. Molar behavioral ratings or aggregate scores are more apt to capture interindividual similarity than molecular codes or single observations (Cairns & Green, 1979; Epstein, 1980). Documentation for parental similarity and difference is not just closely aligned to methodology, it is also strongly affiliated with the theoretical approach taken and the particular questions addressed. The next section illustrates how both similarity and difference in child rearing can be found in the empirical research literature.

In stark contrast to the trait conceptualization of the parentchild relationship lies the child-effects approach (Bell, 1968; Bell & Chapman, 1986). Instead of highlighting parental similarity, investigators in this genre recognized the behavioral adjustments

226
Table 1

HOLDEN AND MILLER

A Chronological Listing of Some of the Parenting Trait Schemes Published Since 1931
Author Levy (1931, 1943) Watson (1934) Hattick & Stowell (1936) Symonds (1938) Baldwin, Kalhorn, & Breese (1945) Lafore (1945) Crandall & Preston (1955) Brody (1956) Sears, Maccoby, & Levin (1957) Schaefer (1959) Spitz (1965) Moulton, Burnstein, Liberty, & Altucher (1966) Ainsworth, Bell, & Stayton (1971) Baumrind (1971) Martin (1981) Koch, Chandler, Harder, & Paget (1982) Pulkkinen (1982) Abelman (1985) Sameroff & Feil (1985) Raphael-Leff (1986) Metcalf &Gaier(1987) McCord (1988) Trait Over-protection Lax vs. strict home training Baby the child, push the child, or appropriate behavior Acceptance, rejection Rejectant, casual, or acceptant (varying on dimensions of democratic to autocratic and indulgent to nonchalant) Dictators, cooperators, temporizers, or appeasers Affection, protection, coactive control, coercive control Sensitive, less sensitive, insufficiently sensitive, or hypersensitive Aggressive, &/or punitive, permissive/strict, warm, responsible Accepting, overindulgent, protective-indulgent, overprotective, possessive, authoritariandictatorial, demanding antagonistic, rejecting, neglecting, indifferent, detached, freedom, democratic, cooperative Rejecting, overpermissive, hostility in guise of anxiety, oscillation, cyclical mood swings, hostility consciously compensated Mother vs. father disciplinary dominance & high vs. low affection Sensitive vs. insensitive Authoritarian, authoritative, permissive (and other subtypes) Involved vs. autonomous Repression, denial, intellecrualization, displacement, or projection as parental defense styles Child vs. parent centered, guidance vs. selfish treatment Inductive vs. sensitizing discipline Symbiotic, categorical, compensating, perspectivistic levels of parental thinking Facilitator vs. regulator Upward striving, overprotective, indifferent, conflicted Aggressive, nonaggressive, or punitive

that parents are capable of making. Studies adopting this perspective have shown that parenting is affected by such characteristics as the children's age and gender, behavior, appearance, temperament, and activity level (e.g., Anderson, Lytton, & Romney, 1986; Fagot & Kavanaugh, 1993; Maccoby, 1984). More sophisticated investigations along these lines have illustrated how child-rearing practices reflect the interaction of child and parent characteristics (Bugental & Shennum, 1984; Dix, Ruble, & Zambarano, 1989). The source of much parental behavior is not an internal trait-like orientation but rather reactions to children's characteristics and behavior. Hence, according to this perspective, what similarity parents display would be a reflection of the stability in their children's behavior and characteristics. Child-effect researchers do not maintain that this approach portrays the full picture of parent-child interactions. Rather, it represents one aspect of a more general model of reciprocal interactions or transactional models (Bell & Chapman, 1986). The central theme of such models is that parental behavior reflects a joint product of parent and child characteristics. It is based on the history of interactions and may become transformed over time. From this perspective, parenting should be thought of as a relational rather than an individual difference construct, and therefore child rearing will vary when interacting with different children in a family. The trait and child-effect approaches are just two of several different conceptualizations of parents that have guided development of the empirical literature (for other approaches, see Holden, 1997). However, these two models of parents contrast most dramatically in their conceptual orientations concerning the similarity of child rearing. Conflict over this same issue can also be found in

the research literature focusing on the determinants of parental behavior.

Determinants of Parenting
At an empirical level, investigations into parental behavior reveal evidence for both similarity and difference. In fact, the question of similarity or difference in behavior is closely related to the issue of what determines or influences behavior (Mischel, 1977). If parental behavior is determined by only a few variables that are likely to be unchanging or have a pervasive influence on parenting, then child rearing should be characterized by more, rather than less, similarity. However, if child rearing is susceptible to a multitude of external influences, or even just one variable that undergoes considerable change (i.e., child behavior), then a view of similarity must be carefully reconsidered.

Sources of Child-Rearing Similarity


A number of categories of determinants of parental behavior have been identified. For example, Holden (1997) found more than 30 variables that have been shown empirically to influence parenting. The variables can be most readily organized into parent characteristics, child characteristics, and contextual variables (e.g., Belsky, 1984). Depending on the variable and the quality of that variable, a particular determinant could either serve to promote similarity or difference. Determinants that promote similarity exert constraints on parents to maintain similar behaviors. The most general of these is culture. Culture can play the function of prescribing, guiding, and

SIMILARITY IN PARENTING limiting child rearing (Bornstein, 1991, 1995; Whiting & Child, 1953). These constraints typically work through parental beliefs and attitudessuch as defining when and how to care for infants, what child characteristics are desirable, and specifying when particular parenting practices (e.g., nurturance, encouraging independence, physical punishment) are accepted, expected, or perhaps taboo (Bornstein, 1995; Valsiner, 1989). More generally, cultural orientations toward family versus work, maternal employment, and child care provide important guides for the degree of parental involvement. These perceived cultural mandates as well as the specific adaptations made within a particular ecocultural niche of the family have been labeled parental ethnotheories (Harkness & Super, 1995). At the same time, culture can serve to prescribe variation and change in parenting. These cultural recommendations are likely to come in the form of widely held instrumental and descriptive beliefs about children and parenting (Stolz, 1967). A number of cultural prescriptions are related to the child's age. For instance, such culturally guided child-rearing practices may concern when children should be weaned from the breast or sleep separately from parents (Meldrum, 1982; Morelli, Rogoff, Oppenheim, & Goldsmith, 1992). Or the prescriptions in some cultures may involve accelerating toilet training or attempting to suppress early motor development (Valsiner, 1989). Consequently, ethnotheories represent a homogenizing influence on parents within a particular society or group. In a similar fashion, membership in a social class, ethnic, or religious group can provide implicit or explicit models of child rearing and constraints on individual variation in parenting; it is likely that these subcultural divisions can be as potent or even stronger than more generalized cultural influences. The impact of social class on child-rearing values has been best revealed by Kohn (e.g., 1979), who found that working-class parents value obedience and conformity in their children in contrast to the autonomy and initiative valued by parents who occupy a higher socioeconomic status. An example of an explicit model of child rearing can be found in conservative Protestants' subscription to the belief that corporal punishment of young children is a disciplinary practice beneficial for their socialization (Ellison, Bartowski, & Segal, 1996; Thompson & Miller, 1997). At the individual level, a number of variables can promote similarity in child rearing. Most important are child-rearing beliefsthose cognitions that represent values, attitudes, perceptions, expectations, or ideas about children or child-rearing practices (Goodnow, 1995; Goodnow & Collins, 1990; Holden, 1995; McGillicuddy-DeLisi & Sigel, 1995; Sigel, McGillicuddy-DeLisi, & Goodnow, 1992). If the belief is sufficiently strong such that it becomes a guiding principle for that parent or it colors the quality of interactions with the child, then that cognition provides a potent determinant for similarity. For example, if a parent believes that a high level of involvement, constant monitoring, or power-assertive discipline is an essential parenting ingredient for child socialization, then similarity along those dimensions would be expected. Parents' perceptions can also promote similarity. If a father regularly attributes his child's noncompliance to an internal disposition, or if a mother's belief that she has little ability to control her children resulted in a self-fulfilling prophecy, then those perceptions would likely promote similarity across time, children, and situations. The work by Bugental and her colleagues has best

227

documented the role that maternal attributions of perceived control play when interacting with children. In several different experimental demonstrations, they have shown that mothers with low perceived control feel threatened by unresponsive children and act accordingly (e.g., Bugental, Blue, & Lewis, 1990; Bugental & Shennum, 1984). The cognition that has most recently been linked to the idea of similarity in child rearing is an individual's internal representation of self and others. These ideas are believed to stem initially from early attachment relationships but reflect the parents' current mental representation of their childhood relationships with their own parents. In turn, the representations serve to prime parents with expectations about their children that then influence how they approach their own children (e.g., Main, Kaplan, & Cassidy, 1985). Although these internal ideations of the self and child are potentially malleable, at least one study has found that these representations were stable in women from pregnancy through the time when their children reached 11 months of age (Benoit & Parker, 1994). This is not to say that parental beliefs and perceptions function only to promote similarity. Just as culture can play dual roles in prescribing similarity or difference, so too can parents' beliefs and attitudes. Favoritism for one child over another, rigid sex-role attitudes, beliefs about how parenting should change with the child's age, or beliefs that child rearing should be modified in public settings versus private ones are examples of how individual beliefs could promote differential behavior. However, we suspect that for the most part, beliefs serve to promote similarity. Certain adult personality characteristics are also likely to result in greater similarity. It is not difficult to picture how continuity in parenting would be observed in a father with a chronic shorttemper, in a highly empathic mother, or in a self-absorbed teenage parent. Behavioral similarity also appears from the consolidation of habitual forms of interacting. Individuals are likely to repeat a behavior if the conditions for its prior elicitation recur (Cairns, 1979). These and other likely sources of similarity are listed in the top half of Table 2.

Sources of Child-Rearing Differences


In the midst of these sources of similarity, there are also sources and pressures to modify child-rearing beliefs and behavior. Given that the nature of parenting is multiply determined and dynamic (e.g., Belsky, 1984; Pettit & Bates, 1984), the similarity can be undermined by a change in one or more determinants. Parents must modify their behavior in important ways to respond to changing child behavior and characteristics (e.g., Maccoby, 1984). Development in children has long been appreciated as a catalyst for transformations in parenting behavior (Baldwin, 1946). For instance, the onset of infant locomotion is accompanied by a dramatic increase in parental expectations of compliance and the use of angry prohibitions but also new expressions of warmth, such as verbal affection (Campos, Kermonian, & Zumbahlen, 1992). Children who were early walkers showed more conflict as well as positive exchanges with their parents than later walkers (Biringen, Emde, Campos, & Appelbaum, 1995). Not surprisingly, new capabilities in the child can be associated with changes in parental satisfaction. Parents report less child-rearing enjoyment with their more mobile 18-month-old children than they did when their

228

HOLDEN AND MILLER

Table 2 Influences That May Promote Child-Rearing Similarities or Differences


Parental variables Child characteristics Influences promoting similarities Experience in family of origin Personality Child-rearing beliefs/attitudes Habits, routines Friends, social support Formal education Occupation Gender Temperament Birth order Attractiveness Biological/adopted Handicap Influences promoting differences Aging Change in marital status/satisfaction Shifts in stressors Different immediate goals Changes at work Mood New child-rearing information Note. SES = Socioeconomic status. Growth/age Behavior Mood Time of year, day Day of week Presence of others Competing demands Nature of specific context Culture SES Religion Neighborhood Context

infants were 12 months old (Fagot & Kavanagh, 1993). The focus of parental concerns may also change with age. Mothers of toddlers are primarily concerned with their safety, but those concerns soon shift to moral and conventional issues (Gralinski & Kopp, 1993; Smetana, 1997). As children attain puberty, adolescents and their parents exhibit an increased amount of conflict and decreased warmth compared with before or after puberty (Paikoff & BrooksGunn, 1991; Steinberg, 1981). Differences in child rearing are not only instigated by a child's development; the need to modify parenting behavior can come from many sources. The family systems perspective (e.g., Minuchin, 1985) reminds us that parent-child dyadic interactions do not occur in a vacuum but can be influenced by the presence of other individuals. Support for this has been borne out with the research on second-order effects (Bronfenbrenner, 1979) and triadic family interactions. For example, fathers became more demanding of their sons when observed with their wives compared with father-son dyadic interactions (Buhrmester, Camparo, Christensen, Gonzalez, & Hinshaw, 1992). Part of the impetus for change in the family context may come from competing goals and the motivation to balance the needs of different individuals (Dix, 1992). In the case of marital violence, battered women report that they often modify their child-rearing behavior to avoid inciting the wrath of their abusive partner (Holden & Ritchie, 1991). Every family unit experiences changes over time, such as the arrival of another child, a change in the quality of the marital relationship or employment, health problems, divorce, or remarriage. Many of these changes in the family structure or functioning have been shown to impact the quality of parenting (e.g., Dunn & Kendrick, 1980; Hetherington & Stanley-Hagan, 1995; Taylor & Kogan, 1973). The effects of marital conflict and divorce on child-rearing practices have been frequently investigated. For example, the two best documented findings are that parents in maritally turbulent homes are likely to express negative emotionality to their children and experience problems in child management (Cummings & Davies, 1994).

Change in the family is often associated with increases in stress. Stress has also been shown to be a potent agent for changes in child-rearing practices, whether it be transient or chronic (Mash & Johnson, 1990; Webster-Stratton, 1990). In the case of unemployment, economic need commonly results in stress and the onset of negative parenting (Conger et al., 1992; McLoyd, 1990). Similarly, the stress associated with living with a violent husband appears to result in increased maternal aggression directed toward the child (Holden, Stein, Harris, Ritchie, & Jouriles, 1998). Other sources of variation in child-rearing behavior include such mundane determinants as the time of year and day. Crouter and McHale (1993) found differences in parenting during the summer and the winter months, at least for those parents whose work schedule changed. Parent behavior may also vary across the course of a day. It is not difficult to predict that most parents would say the quality of their interactions with their rested children is better during the mornings than in the early evening, a time colloquially referred to in some families as "hell hour." In fact, there is some evidence that parents are more likely to spank their children in the evening than other times of the day (Clifford, 1959; Holden, Coleman, & Schmidt, 1995). Although there are competing explanations for this observation (e.g., decrease in parental patience vs. increase in base rate of child misbehaviors), the time of day is nonetheless correlated with differing rates of child-rearing behavior. A number of other characteristics can also account for differences in parenting. As parents age, their child-rearing practices may be transformed because of one or more of a variety of variables. Fatigue or physical ailments, new information learned, competing demands, or previous experience with children are a few of the sources of change that have been suggested in the literature (Clarke-Stewart, 1978; Holden, 1988; Ragozin, Basham, Crnic, Greenberg, & Robinson, 1982; Zussman, 1980). Transient variables such as parental mood, shifting goals, or the immediate context have also been shown to impact parenting (Dix, 1991; Dix

SIMILARITY IN PARENTING

229

& Reinhold, 1991; Kuczynski, 1984; Schaffer & Liddell, 1984; Zussman, 1980). The variability associated with parental behavior has been best documented in studies investigating parental disciplinary practices. In a pioneering study, Grusec and Kuczynski (1980) showed that mothers' disciplinary responses were far more dependent on the nature of the transgression rather than a belief about how to discipline a child. That work has since been extended in several different directions (e.g., Grusec & Goodnow, 1994; Smetana, 1989; Trickett & Kuczynski, 1986; Zahn-Waxler & Chapman, 1982). These and other potential sources of child-rearing variability are listed in the lower half of Table 2. The likelihood that parent-child relationships can be characterized by considerable difference should not be surprising; there is ample evidence that adult intimate relationships undergo significant changes over relatively short periods of time, such as during honeymoons or after the birth of a child (Belsky & Rovine, 1990; Cowan & Cowan, 1992). Certain close observers of parental behavior are keenly aware of variability in child-rearing practices: Children are quick to protest that parents act more favorably toward their siblings than to them (e.g., Daniels, Dunn, Furstenberg, & Plomin, 1985).

Summary
The above examples of sources of child-rearing similarities and differences illustrate the competing evidence for both continuity and change in parenting. How the determinants interrelateand impact child-rearing similarityis likely to be abstruse. Certain determinants may compete between each other (e.g., current beliefs vs. experience in family of origin), others may be additive (e.g., difficult temperament and poverty), and still others may interact (e.g., personality and stress). Some variables mediate child-rearing behavior and its stability (e.g., attributions); others serve as moderators (e.g., poverty). However the variables relate, they represent a multitude of potential influences that contribute to both similarities and differences across the three child-rearing domains of consideration. Through this discussion of influences on parenting, it becomes apparent that which variables or influences are held constant and which variables are manipulated will impact the outcome. It is likely that an investigator could design a study to maximize either child-rearing similarities or differences. To some degree, we suspect that this has occurred in the literature. Longitudinal studies of parent-child relations are intended to reveal stability, in contrast to cross-situational studies that are more likely to emphasize the situational specificity of child rearing. Research examining parenting across children falls somewhere in the middle, whereby each child's temperament and behavior are the variables allowed to fluctuate.

similarity to be found in the across-time studies rather than the across-children or across-situations studies. Foremost, given that the same dyadic relationships were assessed, those relationships are based on a common history of interactions and that longitudinal studies were intended to maximize the likelihood of finding stability, we expected to find the most evidence of continuity in that domain. Another reason why we expected the most similarity across time was that such studies were unlikely to contrast the parenting of an infant with the parenting of an adolescent; rather, it was expected that most studies would utilize relatively short longitudinal time frames by making comparisons across months rather than across years. Given the potential power of proximate contextual influences on both the child's and the parent's behavior, we expected the acrosssituations studies would show the least amount of parenting similarity for several reasons. First, we recognized that parental behavior can be highly situation specific, as much of the childrearing literature reviewed above has shown. Second, in line with theorists such as Mischel (1984), we expected the person-situation interaction would result in differences across contexts. In addition, we reasoned that studies examining behavior across situations would be designed to maximize the ways in which the context elicited different behavior. We anticipated that studies comparing parenting across children to fall in between the other two domains in terms of the extent of similarities or differences found. Several specific variables were hypothesized to modify the level of similarity for the across-time domain. We expected studies containing longer samples of child-rearing behavior (thereby capturing more representative samples of behavior) to show more similarity than studies with briefer observations. Also, studies with shorter time intervals between assessments were predicted to show more similarity, as has been found with children's intelligence test scores (e.g., McCall, Appelbaum, & Hogarty, 1973). We reasoned that investigations across time were more likely to use global and molar-type variables, which in turn should show greater interindividual stability than behavioral variables commonly assessed in studies across children or across situations. Similarly, longitudinal studies also are likely to rely on child-rearing attitudes, which are likely to be more stable than child-rearing behavior (Holden, 1995). Finally, we anticipated that similarity would be more apparent in studies that began with older children, where parentchild behavioral relationships would be more established and would undergo fewer dramatic ontogenetic changes than in studies commencing in infancy. Method

Identification of the Studies


Three strategies were used to identify potential studies for inclusion. The primary search method involved conducting a computer search of the Psychological Abstracts database (PsyclNFO, 1888-1997). In addition, to identify unpublished conference presentations, a computer search of the Educational Resources Information Center (ERIC, 1901-1997) database was also conducted. Both searches involved the use of the descriptors parents) and child-rearing, cross-referenced with stability, instability, consistency, variation, longitudinal, context, situation, and siblings. Second, a manual search was conducted of articles published in Child Development and Developmental Psychology from the years 1980 through 1997. The ancestry method was also used, whereby references listed in review

Questions Addressed in the Meta-Analysis


Theoretical orientations and empirical reports have presented credible reasons concerning why parenting behavior might be characterized by either similarity or difference. This meta-analysis is designed to systematically evaluate the available evidence. On the basis of the existing literature, we generated several hypotheses. For several reasons, we expected the greatest amount of

230
Table 3

HOLDEN AND MILLER

Characteristics of the Studies in the Three Child-Rearing Domains


Child-rearing domain Characteristic Number of studies found Number of studies used Assessment method Questionnaire Interview Observation Multimethod/other Time period when study appeared 1960-1972 1973-1982 1983-1997 Across time
78

Across children

Across situations

56 9 3 37 7
4

23 13 0 1 9 3 1 1 11

31 20 0 0 20 0 1 5
14

8
44

Assessment First Age of child 0-12 months 13-35 months 3-5 years 6-10 years 1 1 years or older Sample size (mother-child dyads) M Range
30 17 2 4 3

Child First
3 3 2 2 3

Final
10 23 10 2 11

Second
3 3 0 2 5 4

6 9 1 0 139
31-516

75
20-359

41
12-120

articles (Maccoby & Martin, 1983; Martin, 1975; Rollins & Thomas, 1979) or empirical studies provided a source for other articles. Those searches led to a total of 132 studies that were examined for possible inclusion. To be included in the data set, the study had to meet five criteria. First, the study had to include a direct assessment of parents, rather than the children's perception of their parents' child-rearing behaviors or attitudes. The assessment method could involve the use of questionnaires, interviews, observations, or some combination. Second, because we wanted to investigate the similarity of normal parental behavior, research into clinical samples of parents or children (e.g., depressed parents, noncompliant children), family transitions (e.g., modifications in family composition such as divorce, change in parents' work status, or adjustments to natural disasters or trauma), or studies involving parents of disabled children were omitted. Although these criteria reduced the sample size, it was done in order that the results be generalizable to normal child-rearing practices. For example, it has been shown that certain clinical populations of mothers (e.g., depressed, schizophrenic) not only exhibit different rearing practices than comparison mothers (Goodman & Brumley, 1990; Kochanska, Kuczynski, & Maguire, 1989), but they show diminished ability to adapt and modulate their behavior (Dix, 1992). Inclusion of such groups could have inflated measures of the similarity of child rearing. However, wherever possible, data from comparison groups used in those clinical studies were included (e.g., Kochanska, 1990; McHale & Pawletko, 1992). The third criterion for inclusion was that the study needed to contain sufficient statistical information (correlations, F or t values, means, and standard deviations, or at a minimum, some information regarding significance) to allow the use of meta-analytic techniques. Fourth, we limited our focus to mothers' child rearing because of an insufficient number of studies focusing on fathers. (We did identify a total of 21 studies that sampled fathers; all but four also included mothers.) Finally, because we wanted to examine actual child-rearing behaviors or beliefs, studies that focused on

parental behavioral intentions were excluded. Most of these studies used vignettes asking parents to predict how they think they would behave in response to a particular scenario. A total of 87 empirical studies met those criteria. The investigations fell into three categories: (a) 56 studies examined parental behavior or attitudes over time; (b) 13 studies investigated parental behavior across different children within a family; and (c) 20 studies assessed parental behavior across situations.1 Those studies included 66 published articles (69% from Child Development or Developmental Psychology), nine conference presentations, six chapters, three monographs, two dissertations, and one manual. A summary of the characteristics of the studies is found in Table 3. Sixty-three percent of the studies contributed to the across-time domain, 15% to the across-children domain, and 22% to the across-situations domain. Most (84%) of the studies that used a single assessment method relied on systematic observations, followed by questionnaires (11%), and interviews (5%). Although the data set spans almost 35 years, more than three fourths of the studies (78%) appeared in the past 15 years. With regard to the age of the children, infants or toddlers were the focus in 71% of the studies; only 18% of the studies included children older than age 10. A final characteristic of the studies that varied across child-rearing domain was the sample size. The mean number of mother-child dyads in the across-time studies was 75. The across-children studies had the largest number of mothers per study (n = 139), whereas the across-situations studies had, on average, the fewest number of mother-child dyads (n = 41). Another characteristic of the three sets of studies was the presence of confounds. Studies in each of the three domains had confounds. Of the 56
1 Two studies contained data for two domains (Conger & Conger, 1994; Dunn & Plomin, 1986).

SIMILARITY IN PARENTING across-time studies, nine introduced a confound by changing the setting (k = 6) or switching the assessment method from observation to self-report (k = 3). All of the across-children studies had one of two types of limitations: The observations inevitably confound either the age of a child or the age of the parent. Investigations must either have had parents interact with their own children, who were at different ages, or the researchers must have waited until the younger child reached the same age that the older child was before conducting the second parent-child observation. Out of the thirteen across-children studies included in the meta-analysis, seven confounded age of child, and six confounded time (along with the age of the parent). Another confound with the data concerns the across-situations studies. The design of the studies involved changes in location, tasks, or both. Three of these studies compared parental behavior in the home with behavior in the laboratory or another location (e.g., park). The other 17 studies occurred only in one setting (10 in the laboratory and 7 in the home). Each of these single-setting studies created two or three different situations by modifying activities (k = 13), introducing other individuals into the interactions (k = 2), switching rooms (k = 1), or modifying activities and introducing another individual (k = 1).

231

In several instances, two or more variables used in a study were collapsed and classified into the same construct. For example, in a study by Clarke-Stewart (1973), appropriateness and responsiveness were both included under the construct responsiveness. In such cases, the effect sizes from the two (or more) variables were averaged to form one measure per study as recommended by Rosenthal (1991) to maintain independent samples in the meta-analysis. Tables 4, 5, and 6 list the studies analyzed in each of the three domains and provide a summary of the key characteristics of the studies, including the sample size, presence of moderator variables, and the particular constructs examined.

The Meta-Analytic Method Used


Meta-analyses typically examine the effects of an independent variable on a dependent variable. In the case of parent-child relationships, this technique has been used to address such questions as whether parents differentially socialize their sons and daughters (Lytton & Romney, 1991), whether children's well being is adversely affected by being raised in divorced, single-parent families compared with two-parent families (Amato & Keith, 1991), and whether parental caregiving is related to child externalizing behavior (Rothbaum & Weisz, 1994). Usually, the independent variable is manipulated at two or more levels, and the relation between the two variables is then expressed as an effect size (either with an r or d). The central analyses in this article, however, do not examine the effect of an independent variable on a dependent variable. Rather, we examine whether parents are relatively similar over time, different children, or contexts. Thus, a particular variable was assessed at two points in time, with two children, or in two or more situations. Because the issue of the degree of similarity is determined from correlational analyses, the effect size (ES) estimates were based on Pearson product-moment correlations (e.g., Rosenthal, 1984). For studies examining two (or more) variables classified into the same construct, the ES was the average correlation. For the computation of the average correlation, Fischer's r-to-z transformation was used. If the results indicated only that the correlation was nonsignificant, the variable was given an effect size of 0. A number of studies, rather than assessing degree of similarity, examined the issue of whether child rearing differed significantly in one of the domains. Analyses from these studies were most often /-tests or ANOVAs and were translated into effect sizes with d (Hedges & Olkin, 1985). Effect sizes were also computed in studies that provided means and standard deviations but did not report any statistical test. Studies reporting only significance levels were also included in the analyses. For unspecified significant results, p values along with sample sizes were used to determine effect sizes (Johnson, 1993). As with the correlational data, nonsignificant results were assigned an effect size of 0. Across both types of data, the only information provided for 29 (5.4%) of the 498 variables was that the analyses were nonsignificant.2 Correlational data as well as data concerning the difference in mean levels of scores both yield information about similarity and change. However, for convenience, correlations are referred to as an index of similarity, whereas tests of mean levels of scores are ascribed to as an index of difference or change. Because the effect sizes obtained from t or F values denoting differences across time, children, or situations could not be combined with the correlations expressing similarities, separate analyses were conducted for similarity and difference tests. It should be pointed out that the similarity and difference tests are not conceptually opposite. For example, the frequency of hugging or kissing children will likely show a

Selection of Child-Rearing Variables


Each study measured an average of 6.1 child-rearing variables, resulting in a total of 533 variables of interest. Two coders then independently grouped those variables into child-rearing constructs. Considerable attention was devoted to determining the appropriate construct categories. Collapsing variables into too few categories might have obscured any systematic patterns. Alternatively, if too many categories were created, there might not have been enough data for analyses. The coders arrived at 11 independent child-rearing constructs, ranging from caregive to verbalization. Four of the constructs reflected different aspects of the quality of interactions. Caregive included observational variables that reflected daily custodial activities with young children, such as feeding, holding, and soothing. Behavioral or attitudinal measures that reflected support, praise, help, rewards, or use of positive reinforcement were coded into the construct encouragement. Interaction was a behavioral construct that indicated the degree of positive involvement between mother and child, including engagement, play, and proximity. .Behaviors that involved demonstrating to the child, teaching, or providing some type of cognitive or social stimulation were coded as stimulation. Maternal vocalizations directed to the child that may have been qualified as rate of speech, syntactic complexity, or richness of vocabulary were classified as verbalization. A construct that captured the quality of interaction was responsiveness. The behavioral variables subsumed by this construct included assessments of sensitivity or responsivity. Attitudinal or observational assessments of child management or disciplinary practices or orientations that may have appeared as restrictiveness, power assertion, or punishment were collapsed into the construct control. Another behavioral construct was monitor, which consisted of parental supervision (watching) or maintaining an awareness of the child's activities. Two affect-based constructs were included. Negative affect consisted of the observed or self-reported attitudes of hostility, rejection, criticism, or negative emotionality. Positive affect encompassed behavioral or attitudinal measures of maternal pleasure or enjoyment with the child, operationalized as laugh, smile, warmth, and affection. Finally, the construct noninvolvement included those behaviors that reflected a lack of interest or involvement with the child, such as ignoring or nonengagement. To assess the reliability of the coding of the variables, two coders independently categorized the variables into the 11 constructs. They agreed on category placement in 82% of the cases. All discrepancies were discussed and resolved by the coders. A total of 35 of the 533 variables (6.6%) could not be classified into one of the constructs and thus were not included in the analyses (e.g., achievement pressure, chores, feelings of competence, independence, and over-protection or fostering dependency).

The meta-analyses were recomputed without these nonsignificant variables. Removing those variables resulted, for the most part, in minimal changes in the overall effect sizes. This set of meta-analyses is available from either author.

232
c/f
M ,

HOLDEN AND MILLER

Og

g|

>?g

1s Is
>g
"3 t/i

'H li" ! 'f g


i ^ | 1 S m Q > E 5 o
'-C t/5 G ^ /-^ Q ^Z,

g^sl
^ I S Q 5 ' 3 - S -
"3 * W) *^^ td
2

^
*
^

s
1 oo" "^^ .2 Q

^s
^ 5 co" ^ '' O
2 t/J

>

> g g *o eft *tt


C u

T3

^ S*\ ss 1 1 1|^:|
r-v

Ql | | | | g .
^^ C
-T^ ^ - '" ' i t Q . 3 '" y ^ ^" J2 ^ i Q 2 ^ .2 <S

1 iJll
'"^ v5 ^"^

S
t> ^ *tt

If
C O '"" '*^ cfl

1 1 1
D P , . .. cd '-"v ^ Q ^G

3 U

3 SA

c/a ^

'~N G Pi Q Q, o . fl -^ ^^. c/5" H 0 ^

*H

^ -^

"H

^ ' -

w"

^ Q *3 1) ^ ^ G ^ ' " C/5 .S .

Q" ^"^ ^ 55"

S" *^

r^

"^

o ^

.IS
rt

a
en

to G

|s
g

S S

crt

T*

| ss
'a 'a o o Q, CX
at &
4J O

"8. -a r: u
/^T
1

fflsl i
U U co O <3 U S
3
O

alsS-s

IslI 1 I^|^|ll |s^ | i&1&^'B.^^ s


^ |5a|l l

- 1 l-igs"

i I s-fss i ||gi

Jliji 1 i l j | | | | J | 1 i llUfi if | Ifllill


Z (2 6 S U U a UW W -o U U U

ii
i
c

-S

1 si g |> ^ 1 is^ss'is
M "1 M "Is

ii i
0 0

a-o

-a 3 O
G
D
CA

|
3
n U

>-. o3

3 3 O O
CA

3
U

3
U

3
U

C"^

CJ

<

3
O

G C
D D

CA

1
S "S
V3
t f )

1 1 1
D D "S 5
C
t/l

1-3
P H 3 "S
d
J ^
, f i

11
D D
CA
JQ J2 J ^ yj J J

l a1 ^ , ^ ^
D DE H
C A
J2

1 1 1
D (2 D
X
VJ J ^

a.

C
Wl

V
^3

V
J ^

f]

VI

J ^

V)

G
S

-S
P

P
c o

B o p

H ' S ' S f i ' S O C C O C S o o p o B B S


N O O

O O - ^ f

f i t s O O 5 p e e

t s ' S e S ' S g g O C . 2 r t 2 P C o _ S - . E B
I C M

G G ' S c O O C O p ap o p e E
C M C M

"S C o E

C ^ O O

S c O O p p c c

oc

ii 5 E; g? <U (5
.> 0

G C

'

ss s

.S

9
S
P

.S

l^s-is
(A

ES

.S

.S

.S

S P
SA

f r

^r^'3

.S e o . p '3

P 'I

l s
M

s ^ p
J 3 J 3 S S
O O
WJ W5

c <n

^"^cl^'p

ia 2 "e

c S - S o

1 ^
C A 6
P

^ ?

it_

'3 E 6
"8

ill!
cW

J S w c c 2 ^ G S ' S ' S - | p
O P P p S O S

O S

JH S

CA

S g.

J2

^ o o S

E2

s l l l 3

a a

f l

l | | 2

s a! a !

S ^

Assessment n

8> G
^ ^ ^>

1
K | g g 1
2

-P 'Z3 v, '&

cd cd u cd

XI JS ^^ ^3

8<^ S

"*3 w > I-H 4i

1 S

oo

E o
j* "

H . 3 . 3 , 3 a o o cxo
g a o o c a j . v o

T3 '-3 S "S = < d j r . >4 . > j _ ; i i; _ > U > - 55 t U c A { U 4 i c n 4 J

g 1

"-3 '-3 c a e f l r * rU * U U U

' C O S

1 g1 g'& g '.C

^ ^ o o

^ ' g j ^ o o1 o o
~ Jj! j -

e d ^ S c o e > 3 *j d . i_ s > i_ O f c J C A Q J

d > U U

'5 ' i *i c d c d c > U i>U>i U < 1 > U D

*5
d W U > i ( U

^ ^ ^ ^ o oo o

'S ed > U ( U

J o

"^ O '3 ed'^Jed >c > tqjo E4)31 1)

gig

^ o^ o o
^

^5 p
c
o.'o

p o x

o r ^ o m

5?

00

'-.-O

JJ

C. ^ 1

51 1 r? 1

c^e ^ I ^ &

S t . " ^

2 d>(
3

a,

1 ^ 1 . ^

r^ C

cn

^J-

1 ^

o '2 u l c8 oi

S H ^

3 "

<L> -^

1 s||>g
03 09 03

js js ^ a ^ .tS .t! CM . <*

i?l |lil(2ll"
f2^
^^ ^
u

I i^l ?|2| I f 1 i ^ 1 I ?i|l I S|!


"1 1^1?*^
G c " ^ - r t " ^

"2 1 JI2
"^ ** 7j^-S

1^^

c^
C

^
C

* 1^1^
C S3 G

E | * I o
o

p-s "IS
GQ\

1||

gcn^'

03

03

03

UU

(3 (3

6^

ffi

35

SIMILARITY IN PARENTING

233
+-.... ^

^
sfl

C/5

CO

-~.-

Q
On

CJ
<
^-s

<U

co
O

^r^
W5

<u
jj

Q
3 ' '

.c
^

monitor (S, D); positive affect ; stimulation (D)

<5

1 a
Constructs mt
.2

s I I a |n
&g ,
K |
g|g

Si

cd D >

g4_<

' '3 t\ ' a> O

OD

CL

1 Jg
Q g o g

S
D

r is -s s I S 2

^ o

c^ c:

lg

|I

g Ji

' S ^f

1 11
'- "^
--

C S, 0 _ &

52_ ^ 5 Q |
S '-C

r/C

tS

-3 g c S
"*' *-

'"

1|
A;> Q -3 "c M |
S

1
2 & o ^
-t-* s

|^i s~! 3
f l g f c S 2 , | ^ -

^"S-^^-E Q S c j S u t i t S ^ 2 s - > < - . 2 ^g ' | g


n'-3c c co 4)

S
^ ^
8>

i g is i |i
e^>
-l S
0& o bo

IK

1 1 1 fa o ,-> .. ca u

E Q,

I I I O C3

a 1 O

'1 S
c
.

S"o SS ,-. 8 U
"^

" ^ I ^ S2 5^.|g
is .2 ^ w

S,Q g'w g | |
Q > co

?-S ?^

g
o

c | g1 g_
G ^,

M -S^ S ^ MS C - s t s l -Sgs 1 Isal-^agK

C a> >c 'S? t3

U ^ OO V) ^

1
8

g^ g

& g g
1 1 1 |

pa
c !
n. 3 ^ .S

uz u

o u o

C M C

a g g^a 1 l^i'||
U

g ^ g ^Q

>

is * <u co 2

uS u

1 "a
UU

g'lgl? ||Su 1 1 ll l^l'l '1


U U U

g'i|g | s| fla|g|g|
u
S o S c O ^ i O

,gc S> O U

s^^s
u

oz

a:

&

= a '5 l-a a^lls'lal e a52,c'gegSS


u
O

co

ctf

a -a
V 0}

o
0 o

3 j. ^ 1 1J 3
3 MS M

"2
S
to

T)
3

-a
3

3 3 " I I -SDD
eo.

I'S H"g
tc .<u to ("U

cj 5 o 5

1
O &0 00

S
DO

H
C3 cd

2
C

I -S
C g M

.S
M g

.3 &
u^M

-aSP

ID

D u H - s a c u D - ^ - a - a H a c u

g O. <D

co

"O

"B

-S *

i
0

en oo <!

-ll s | s, >> . E 4 jq o ^
^o CM <n

|g |J
m u

s
g

c H

" S

fi

"S

en

"S

oo

cs

S oooCM

g ^ E - ^ ^ ^ V O - H
CM r-CN CO (N

g ^
OO

^ 2 > % C | ) r y
ON ' ON

c
OO

'i |k

1"
OX)

<u

t~

c
ON

c E o

' t
js .g .3 >,

i s g <=
C IO VO

o
G\

7
-3<i
C ^O ^ C M C

^
J3 00

^ 's
To rovo -S c

t_ c;

^ -^
-^ ^ ro m

"s
m <
- c -5 c -5 c

"i
^ C ^ c o

5b
IS 0

C O

e E s,i
OO < <C

GCo.ro O O I

i i

^ \O & ro en

^ssS P^ -S S >^ > J


ID -rt5 4i

"f

1
T3

2_g
ti c9

& Z

S c 22 c co j^oos;-- i B >,
CO g T)

S c <><
O^

1 -c w c ^ o oo

i/> ^

(/)

^ t

c/i

-^

O S

E_

>>

E ^ ^ E

^ 0 o

>,

E>>E

|1

-^

(N

--^ r-

cJ

r-

<

v o - r o v o - v o c o ^ - c o o l - ^

1
"S

#
Interview Observation Observation
c .2 ^
tl> 05 JD

Assessment

- a c e

*1
C! O

4
'3 G 2 O

-S e t

c .2 -2 S as ?5

n o -a *i3 ca
O 3 4M ) aa ETXi ^^43

C CS C O

o
vo Tf

cxo o
r-- CM ON in ro oo

co qj U 41 co co 3 X) 43

c S

'+ 3 D
<U crt 3 43

S -s g (

1 i 1 > S .0 |
J o
(^ - 5 C O -' 3

2
.S | '1 .1 .2 1 lg o SI f CC 'O C C C e
flJW W3W5 X>^

o =3
>

bb

'a
U 3

0
O

OO o> cs
VO O

o
O

U tA ^

oo

a
CM

t/3 U 3

U U Wlcc .C_O

oo

4> in 4^

a
in

S ?i i

o '> d

co

qj -^

o '.^ 3

a
in

WSl>^i4> C ( C v i , o . a . c

| u| u | 1'fef'fcS Ca>Ca)C

o
'

o
o

oo r- O O CO ^* <N

cs

rs

CN

CM

ri in

oo ON

^"

o\

<u

&

~
S
X-s

*S" OO o\

5 H J3

^ 0 "^" "c
T M

^ "O

3~

i ~

*&

oo
in" O --r
ON OO ^^ O\ ON a

1
o
cd

y> Illsl S c .S ON g
- u H-J r"^
~&^a*i

x^TS _r co o

% ri

3 _ |
O N a\

C g
S

33 -*)
bb aJ

Jg-

^ oM

"O

l"i | || 1 ||2 5 |S
^ ^ jr ^ ^ u B S- d' ^ ^ ^ s sf >; =3 B| s S J 3 y m ^"-u
MJ2 S COOQ S <%i >-W S ^ 'C S

s? ^^ |

ON

2&

*"

^"^

S. 2

^^

r^

=J 13 Q
Mo ""^ O =<)

2 C^ 1 1

J-S ** C-ON p
jf 5 ON g '3 S

>
g

^i *
38^

<J O ~ c S <$j "^ oo


"7-1
pri

s >-. c

^ v>

8 ^

8sI 1 x s2 j

&^

1 1^1

IHi 11 ill jsi 1 ! PI


O a-

ou

0.0.

o-

a,

ci

oo

on

on

234

HOLDEN AND MILLER reliable decrease across time, but parents could maintain their relative position on the construct of positive affect. The meta-analyses were conducted at two levels, as recommended by Light and Pillemer (1984). The most general approach involved analyses at the study level. At that level, all constructs assessed in a particular study, irrespective of their content, were averaged to obtain one estimate (or entry) per study.3 The study-level analyses involved a total of six metaanalytic tests: similarity and difference tests across time, across children, and across situations. The purpose of this level of analysis was to discover the extent of similarity independent of the particular type of parental behavior; it provides a conservative estimate of similarity of child rearing (Rosenthal, 1991).4 The primary analyses at the study level were computed with the software package DSTAT (Johnson, 1993), a program that uses the procedures and formulas developed by Hedges and Olkin (1985). Composite mean effect sizes were computed by averaging the effect sizes with each effect size weighted by variance estimates. Using this technique, the greatest weight is given to the most reliably estimated studies and thus takes into account sample-size variation. The reported results include composite mean effect sizes, 95% confidence interval limits (CI), and homogeneity estimates. To interpret composite effect sizes, we followed Cohen's (1988) recommendations. A small effect size has an r of at least .10 or a d value of a minimum of .20. An effect size is considered medium with an r of .24 or better and a d of at least .50. A large effect size would have a minimum r of .37 or a d of .80 or greater. The homogeneity estimates (Qh) tested the extent to which results across studies differed. The statistic, which follows a chi-square distribution, compares the actual variation in outcomes to what would be expected simply because of sampling error (Hedges & Olkin, 1985). Large Qb values indicate significant heterogeneity. Thus, if the Qb statistic is significant, heterogeneous rather than homogeneous results are present and tests of moderator variables are warranted to reveal those factors contributing to the variance among the studies. The second type of analyses consisted of computations at the individual construct level. At this level, each study could only provide one datum per construct; within one study, two or more variables that fell under the same construct were averaged. The DSTAT (Johnson, 1993) program was again used to determine effect sizes for the individual constructs within each study as well as to compute the composite effect sizes. If there were enough data, a total of six (similarity and difference tests for the time, children, and situations domains) composite effect sizes was calculated on each construct, although the mean number of tests per construct was 4.8 (SD = 1.47). At the individual construct level, results include composite effect sizes, 95% confidence interval limits, and homogeneity estimates. Once effect sizes were determined, follow-up analyses were conducted to assess whether the across-time, across-children, and across-situations results were quantitatively different at the study and individual-construct levels. These linear contrasts of the average effect sizes were computed following the method recommended by Hedges and Olkin (1985) and are comparable to contrasts used in ANOVAs. A test statistic was computed for each contrast using the two effect sizes, estimates of the variance, and the pooled variance. The standardized contrast value obtained was then

1 'S S
CS

g
5? .2 ^
ta

6
O g

5?
1
g 55- g .S ^ .2
5? M ^ rt

1
tj

J 5
5

>

c o U

1
S 1
CD

m@
o
C3

05

Q
CJ
<4-j

' '

Interactic

Negative Stimulati

"aTa

>

rt

S S ^ u s 1/5 >

cs
fll g

S.

Oil en

cd bfl

11
o ^

H II H 1 H II
^4 [

<4-.

o -^
D. "*

o H

"S
3 o

-S

a 1 a S

J2 1 P 1 1
-S 1 n 0 o
rcs

!s S
s 1
E

E c: ,o

s- s o
E
^

1
-H

r^ S

u; o
00

^.fi

&
0)

m cs

^*

f_ ^ ^ I.
rt O II M rt

c .S o 04
l/-^
^f <N

.s

> OO

o
<u

sJ
M> CO

"S H

i^
V)

c ^

a c 1 'E o

'g o 2

o .. c H B P

Crt

> CT

jr

1
S
VI

11
^> o

6 g H

TB

1 s

II
^ v-i

! !
c c 0 0 a jo > > J a i J 0 0
0

1 1

a 1 1
8 1
'3
V

Assessme

Observati

Observati

Observati

Interview Observati

Observati

5 c

c o

c o

c: o

c:

"O

2 CS

2
4
en1

vO

cs

Sigman, Neumann, Carter, Cattle (1988) Stattin & Klackenberg (195 Tamis-LeMonda & Bornste (1989)

Tamis-LeMonda & Bornste (1991)

=a

CS

G S

^
M ^ |

1
'en

known; '

3 O

1 II

.c
u en
II

Q
Crt

il c
ea

Eiden (1996) Vuchinich, Angelelli, & Gatherum (1996)

Table 4 (continued)

1 u
2
u

fc <

o 2

yc
1 ^

s II
00

> c. >a ^ H

3 Twenty-six studies reported both correlations and difference tests, so each study contributed one measure for each type of analysis. 4 A second type of analysis was conducted in which the constructs were not averaged. Rather, each study could contribute multiple pieces of data to the meta-analyses if the study included multiple constructs. For example, if three constructs were reported in one study, each construct was treated as an independent measure. As with the study-level analyses, a total of six meta-analytic tests were performed at this level. These analyses provided results that were very close to the study-level analyses. They are available from either author.

SIMILARITY IN PARENTING

235
C

i verbalization (S) ol (S); negative affei (S); positiv ; responsiveness (S); emulation (

S
o

1 'I - ^^ ^

>

>

u .1 > ra

10) C/3
>

<L> -2

oQ
S

xx

'i 1 0.-C
C en D <A

g" S
X * o

o> " ' tl si


00 C <d O

r^ ^
O

Q W ^T
o

' |c u '-C a N
" *e3

ol (D); negative affe set (S, D); verbalizat ol (D); encourageme x;t (D); positive affe sonsiveness (D) ive affect (S, D)

3! (D); encourageme ; negative affect (D)


c Q O ^^ U
&

Constructs m

23 n ID

ive affect (S); positi'

ive affect (D); positi

*-" O ti U . C *-> u

1 1 gf
Q c 1

S 5

C S* "
S

^5* ^ S ^^

o o
tO LO

g 5 Cd Cd

a.
z

! f i fi c2 u
"a s
c
cd

|ll 11 1 .H gS'S 1 gg

g|g g 1 1 O O U

>

|fl 1 g|g
0

oo 1> Z

oo CJ Z

u3

<~ ^ c

S'e & l

cd
C/l V3

fl
CO 60

cd 3,

_*
^Hcq H
cfl

_cd

>,
en
60

JS S H

to ed

8 ft

cd

< H

** ft,

a H

cd H

ob ;

I 3 o o

7
<N

'

1 2
"^
W

'
<r^

c 'i o

C
cC
O

E |
m
^

<s

V-)

*
z
e o
Cfl

s 0
5
0

II

o
(U C C u a> tt3 *i "O
"P

f
ffi
qt

a-
R]

i
7
c!j

M
U

<3-

u >^

D c o

>-.

3 c o >^

<u c o Z

z
to

c o

^f

u >^

<N

1
o c
11 03

00

oo

oo

S M U U

1
2 U

1 1
o QQ o CO

f ^ CQ o ao fO CQ

"o

"o "*

li
'3 'S? cd "^ S S-

to

t/}

=1 s
o oq

i s
oa
o

iS
<y
JJ

3.3| O O S
J= -O <S

i 8 O J3
f . <->

"o , i

J/T
>s

g o 13 o CQ

o a e " u '" f S?f


CQ

2 "> a, u u 00 J= c rt o O PQ

rt C

ca 0) o

1
=3 S
e

in
Sj

(U

bservation (dyads)
O
in

bservation (dyads)

c 11
qj U

c
'*3

c
'3

c o

g ?2 ? C .^ C .2
O o

If
O
Os

S D 1>

c o

a S' s 1? | 1 0 ~"

3 O '3 1? 9- o

s s^l-s -a -a S 3
S "2S 8 =1 S 3=3 c
W
_

'53

3 *s; G S o ^
i f?

6 o

00

0
OO

i| 0
m

11
O
^_

_Q

O
VO

O
\O

O
OV

<
VO

o I II Q
uT

iS

^H

r*~i

in

13

>,

_
00
OO c^ *

1 1

r? ON

80 S 1

=y
55

_g
u

4>

^
1

S
u>

c'

&>

o\
_J

1
S^
Q"

x> o

j|I

I
CO

1/1

1-^" OO *- OO
-^* ON S ON " r-* ffl i

i <

aS

.a |

.s"
o

i !
|f
*

c
T3 0)

E I

3 a "1 &
H =0

l^m^

ii

t 1

tj 2 o
M

c- S
- ^" 3

el-

oo 3
> So - oo

D
W5

1^1 S S ~f cs

236

HOLDEN AND MILLER

1 | r. c S sConstructs meas
o 3

- t 1 ll |
i ^'

Q
v$
2
S3

S
3

3 B OS: ., c - 8
gg ^ (-.
^ U

-S

1
^.
^

s
C/5

ac
I'S 33 N S'a "e ^ u

gg

Q
I .2 1 1 2

rff
.2
cs >-5
t- W

3 S 1 '?! en ,,
Control (S, D); responsiveness (S, DJ Negative affect (D); responsiveness (

Caregive (D); control (D); interaction (D); positive affect (D); stimulatioi Interaction (D); responsiveness (S) Caregive (D); interaction (D); positiv (D); verbalization (D)

Encouragement (D); interaction (D); noninvolvement (D); stimulation (1 Control (S, D); positive affect (S, D)

Control (S); encouragement (S); negE affect (S); stimulation (S) Interaction (D); stimulation (D); verb

Caregive (D); positive affect (D); res verbalization (D)

Encouragement (D); stimulation (D);

:ncouragement (S, D ); noninvolvement (S S, D)

i."

1 1 1 s s -
s ^ S.

o S

^-^ r-i

<u c

So

1
>

5? g 5
a
c: ^o S c

sfIL
o S g -
c r

's .5 Q, S > " -I "- 1


U Q 3 j_) i_i O

A^

"

jjtlff

Responsiveness (S)

:s)

Verbalization (S)

Responsiveness 1

Encouragement (

5?

J | | | | |
VI

i . s a | g
u v "o QQ &0 o > bfi > O '-3

OS

U3
60

PJ

U
^

wi

3 ca

* O
^ wi

W C
$ > Q^ >> g >

BO C

1 O
&.

1 li
o.

>^

Situation

"i t*_
c
O

-o tt-

cture draw:

^ 'c
0

>

3 ^r S *c
W5
O

3 >

li J^

"1
en

wi

bO _G

1 ll
0

> !U 5tI| I ! III | | l fii s 111 a s s


E E

s 1 s * ? *
guse
^ fll

I.PI

> J

2 ^

"In cd

ects/tasks

fe 1

S "S. 1 g

^ O

'3)
e3 U

'3 ^

* GO

oT '3 l-i Q, <S C

-^ en 2

-^ fit 2

1
/-v ^^

O fl>

> ^ CT "^ j_, v>


if
^0

r7 B

i ^ l S) ** c c -j3 ^

a. u 1 1 f
^ i-J
crt C

o S

J^o-l
, | 5

a| 1"
.C ' -W ^ yi

^as-So-S

fl

r"

(U

-i-S

c -o o O U M

^ > '

j .-S

r\ M

x
u"
CO.

3
U

| c1
tj ;

e s. <| ^-?
OH hS PH

-rt S ^ ^

a.

8>JS I1
c ~ .2
CB U

I II
C C

i ji
-0 fl

o =3

1 s 11
'
c/3 bl S S

1 3 3 3
c c
O

3 3 3 3 3
g c
VO

..g
c S c
O

b"

en

_ c
.

c
<N

cs o -S
<N

3
"3 ^ e 1 g i g

^ Tt
V3

oo m \o
1/3

1 b
in
n 3 . 2 =
r* *3 C 0

m ts

o r4

'

^ oo

c
'B in
en

-a
0 i

'g
o m 2

^: ^ <n rt ^

~! ^
.. O >

^3 00

1 in

1-1 (S
(M

>>

-H

J^

e l? ^1* 4 2 2

e M ^i > j S 0\ S U

-e
V

1
E
in

in
en

^ I
<

="
m
C

i D
3
_o
"cd
C

x S

ua S

in S O

>

r-

ll

c o 3 1 1
O *

c o c o a a II II
(S OO -H Tf

11 |
| 2:
ao

^
n

6 o 8 S 8 S
O4 ^

e s e e 8 8 8 8
'iD
^

a -ja

c c ^ 1 I 1 1 | .2 | "ea j* ^
ed

c .2 I </l

c .2 I W

c .2 I C/J

&

&

g e

1
Leyendecker, Lamb, Fracasso, et al.

> O
OO

js O
^f

> O
O

s
<J
O

8 6 6 S 2 ?

Belsky (1979)

-5-

& ^P ii^ ll^tllllf iSs^^c^^tg


3 toQ

- i'i
M

3 Ox 5fl*

, 1 _r f-^, -f? ON

1^
ai j

=*_ jj
jj c-=y ^j S
C r*-T O OO u? ^*

~-< "ob *^ nJ u. Z

ra ra o aa as a

P-.O S ^ c & js o "p, JS 'C C g * <

S b "b

<

ills

Studies Contributir

Belsky (1980) Cain, Rabinovich, Pederson, Zaslow, & Suwalsky (1985) Crockenberg & Litman (1990) Fiese (1990)

situation.
II

lilt

2 a 2 c -a t3 g 1 1I I U 0, U G c3 to

c' S o 5

II
o f

a i" ~ 11 1 1
S

II

'3 <u
c
O

S3
a
t 1

S
^G O

SIMILARITY IN PARENTING compared with the critical value of the standard normal distribution at a set significance level (1.96 for one contrast and 2.24 for three contrasts). If the contrast value exceeded the critical value, the difference between the two effect sizes was considered significant. Because of the risk of Type 1 errors resulting from multiple comparisons for each construct, we used a simultaneous test procedure (Bonferroni inequalities) and set the significance level as well as critical values according to the number of contrasts conducted (Hedges & Olkin, 1985). An inspection for studies that may have contributed outliers was conducted for the major analyses; no clear outliers were found. The final set of analyses involved testing the moderator variables. These analyses were computed at the study level. When available, a total of six moderator variables were tested for. Two types of moderator variables were present in two domains: (a) age of child (infant, toddler, school age) and (b) length of observation (less than 30 min, 30-59 min, 60 or more min). In addition, four other moderators could be assessed in only one of the domains. Across-time data was used to test whether the method of assessment (observation, self-report from questionnaire or interview) affected the levels of similarity. In addition, the impact of the time interval between assessments (6 months or less, more than 6 but less than or equal to 12 months, over 12 months) was examined in this domain. For the across-children studies, the age difference between the children at the time of the assessment (same age vs. different ages) was tested. Type of situation (different settings vs. different tasks) was assessed in the acrosssituations analyses. Analyses resulted in a Qb, which follows a chi-square distribution and has pl degrees of freedom where p is the number of levels. If there was a significant Qb, the means of the two or more levels of the moderator variable were significantly different from each other (Hedges & Olkin, 1985). Follow-up linear contrasts were conducted among the mean weighted effect sizes in the case of a significant Qh when there were more than two levels of the moderator variable.

237

Results at the Study Level


Results from the study level revealed that high levels of childrearing similarity were found in the across-children (r = .50, p < .001, CI = .487.53, k = 10, n = 1,670; Q = 95.35, p < .001) and the across-time (r = .45, p < .001, CI = .437.46, k = 47, n = 3,786; Q = 497.40, p < .001) studies. The mean effect size from the across-situations studies was significant as well but showed less similarity than the other domains (r = .26, p < .001, CI = .207.32, k=l2,n = 509, Q = 47.14, p < .001). Contrast tests revealed that the effect size for the across-children similarity was significantly stronger than the across-time and the acrosssituations effect sizes (ZCT = 2.47, p < .05, zcs = 5.93, p < .05, respectively). The across-time effect size was also significantly higher than the across-situations effect size (ZTS = 4.80, p < .05). However, the results at the study level were qualified by the finding of heterogeneity for all three domains. Nevertheless, the fail-safe Ns for the three domains are substantial. It would take 64,553 studies with null results to make the across-time results nonsignificant. The comparable number of studies for the across-children data is 6,441 and for the across-situations data is 976.

Results at the Individual Construct Level


Meta-analyses at the individual construct level provide the most precise information about similarity of child rearing. However, that set of analyses was limited by sample size; in several cases, there were an insufficient number of studies to provide a useful estimate. Nevertheless, meta-analyses could be computed to assess the child-rearing similarity of seven or more of the 11 constructs in each of the three domains. Similarity across time. A total of 47 longitudinal studies contributed to these analyses; stability could be assessed in all 11 child-rearing constructs. The results, listed in Table 7, indicated that significant levels of stability were found on all the constructs, with a median effect size of .38 (range = .20 to .55). Except for the constructs noninvolvement and monitor that were assessed in only three and five studies, respectively, the effect sizes for the remaining constructs were based on an average of 14.7 studies. Heterogeneous results were present with all 11 constructs. However, the heterogeneity was a result of the variation in the magnitude of the positive effect sizes. For example, the construct that showed the most variation across studies was control. The heterogeneous result was due to correlations that ranged from .20 to .82, with an overall effect size of .53. Similarity across children. Seven of the 11 child-rearing constructs could be tested in this domain, with information from a total of 10 studies. All effect sizes were significant and in the moderate range, with a median r of .38 (range = .33 to .59). Of the 7 constructs, 5 manifested heterogeneous results (see Table 8). The 2 constructs with homogeneous results, indicating similar results, were positive affect (r = .44) and responsiveness (r = .38).
5 These studies were Bates et al. (1982); Belsky (1980); Dunn et al. (1985); Frankel et al. (1980); Green et al. (1980); Holden (1983); Leyendecker, Lamb, Scholmerich, and Fricke (1997); Lytton and Zwirner (1975); Merrill (1946); Tamis-LeMonda et al. (1998); and Wachs (1987).

Results Similarities in Child Rearing Test-Retest Reliability


Interpretation of the results from individual studies is dependent on the reliability of the assessment instruments; the similarity observed cannot exceed the reliability of the method (Alder & Scher, 1994; Nunnally, 1978). Test-retest reliability of measures of child rearing has been examined with two methodsbehavioral observations and attitude questionnaires. Correlations of maternal behavior based on repeated observations of the same mother-child dyad in the same activity at the same location (generally free play in the home or laboratory) over a short period of time (from 3 days to 1 month apart) were reported in 11 studies.5 The median correlation was .59, ranging from a low of .35 to a high of .78. These results indicate that repeated observations of maternal behavior over a short period of time do provide a moderate degree of reliability. The test-retest reliability of global attitude questionnaires is typically higher than the retest reliability of observational data. Although retest data were not included in the present articles with enough frequency to be summarized, a review of attitude questionnaires found that the mean test-retest reliability for 12 instruments that used samples of parents was .74 (range = .61 to .89; Holden & Edwards, 1989). These reliability limits should be kept in mind when interpreting the following findings.

238

HOLDEN AND MILLER Table 7 Meta-Analytic Results for Child-Rearing Constructs: Similarities Across Time Construct Caregive Control Encouragement Interaction Monitor Negative affect Noninvolvement Positive affect Responsiveness Stimulation Verbalization

r
.20** * .53** * .55** * .38** .35** .48** .39** .45** .38** .34** .29**

95% confidence interval: lower/upper .141.21 .507.55 .507.58


.347.41

k
8 22

G
19.93** 202.50*** 95.13*** 104.79*** 109.80*** 234.80*** 8.33* 179.50*** 200.17*** 49.30*** 59.34***

Fail-safe N

8
18

.307.40
.457.51 .307.47 .427.48 .347.41

5
14 3 18 17 16 11

454 1,570 451 1,081 514 1,377

273 21,141
1,178

111
1,296

1,281

.307.38 .247.34

853 586

20,072 671 8,930 162 8,216 5,658 2,734


1,089

* / > < .05. **p<.01.

***;?<.001.

Similarity across situations. Twelve studies involving observations of the same parent-child dyads in two or more settings or situations had sufficient data to be analyzed. Cross-situational data were available on eight of the child-rearing constructs (Table 9). Significant cross-situational similarity in parenting behavior was found for all but one construct. The effect sizes ranged from a high of r = .38 (interaction) to a low of r = .06 (verbalization), with a median ES of .22. Besides interaction, the constructs that showed the strongest effect sizes were control, responsiveness, and positive affect; all were in the medium range. Despite the small number of studies that have assessed cross-situation similarity in parenting, half of the constructs had heterogeneous results.

Comparison of Effect Sizes Across Domains


Quantitative comparisons were conducted to test the significance of differences between the effect sizes for each construct across the three domains. Contrasts could be computed on eight of the child-rearing constructs. Of the 22 contrasts, 13 were significant at the .05 level (see Table 10). Only two of the eight constructs (responsiveness and interaction) did not show significant differences in the levels of similarity across domains. For the remaining constructs, across-time similarity effect sizes were sig-

nificantly larger than across situations in six contrasts (control, encouragement, negative affect, positive affect, stimulation, and verbalization) and larger than across children in two contrasts (control and encouragement). The across-children effect size was larger than the across-situations data for four constructs (negative affect, positive affect, stimulation, and verbalization), and for one construct (verbalization), the across-children effect size was significantly stronger than the across-time ES. For example, parental control had a significantly larger effect size across time, indicating more stability in that domain than in the other two domains (zts = 3.94; ztc = 2.83, both exceeding the critical value of 2.24). However, the effect sizes for parental control did not differ between situations and children (zsc = .44). The construct verbalization showed a different pattern. Here, the effect size for across children was significantly larger than across time, indicating verbalization is more stable across different children than for the same children over time (zct = 5.65).

Moderator Variables
Homogeneity tests at the study level indicated the moderator variables warranted examination in all three child-rearing domains. Sufficient data were available to analyze four moderator variables

Table 8 Meta-Analytic Results for Child-Rearing Constructs: Similarities Across Children Construct Caregive Control Encouragement Interaction Monitor Negative affect Noninvolvement Positive affect Responsiveness Stimulation Verbalization r .34*** .33*** .51*** .44*** .38*** .36*** .59*** 95% confidence interval: lower/upper .237.44 .267.39 .487.54 .417.47 .307.44 .297.42 .547.63 k
3 4 6 8 3 4 5

N
135 324 1,050 1,246 257 303 343

Q
18.68*** 8.60*

Fail-safe N

_ 96 250
2,360 2,740 169 177 1,267

106.69*** 5.76 5.37 23.44*** 27.38***

Note. Dashes indicate not enough data were available for an analysis. * p < . 0 5 . ***/>< .001.

SIMILARITY IN PARENTING

239

Table 9 Meta-Analytic Results for Child-Rearing Constructs: Similarities Across Situations


Construct Caregive Control Encouragement Interaction Monitor Negative affect Noninvolvement Positive affect Responsiveness Stimulation Verbalization

r
.30*** .20*** .38***

95% confidence interval: lower/upper .217.38 .107.28 .297.46

k
4
4 4

N
187
202 174

Q
2.62 0.87 11.50** 0.08 7.13* 46.30*** 0.18 8.56*

Fail-safe N

90
39 189

.18*** .24*** .26*** .15***

.077.28 .147.34 .177.34 .077.23 -.057.16

160 194
251 297

13 70
350 44

3
6
5

.06

160

Note. Dashes indicate not enough data were available for an analysis. * p < .05. **p<. 01. ***/>< .001.

for the across-time studies (child age, observation length, assessment method, and time interval between assessments), two moderator variables for the across-children studies (child age and age discrepancy between children at time of assessment), and three moderator variables for the across-situations studies (child age, observation length, and type of the situations). Partial support was found for the hypothesis that child rearing would be more stable with older children than with younger children. Results indicated that maternal behavior was most stable with school-aged children and the least stable with infants when assessed across time. Studies that began with school-aged children had an average r of .53. Effect sizes for studies that started when children were toddlers averaged .44, and studies beginning in infancy had an average r of .40. The difference among the three levels was significant, Qb = 17.84, p < .001, and the contrast between studies with infants and studies with school-aged children was also significant. Similarly, mothers were more stable across children with older children (see Table 11 for the statistics). The

Table 10 Comparison of Similarity Effect Sizes Across Domains at the Construct Level
Domains compared Time/child critical value Construct Caregive Control Encouragement Interaction Monitor Negative affect Noninvolvement Positive affect Responsiveness Stimulation Verbalization Time/situation critical value Child/situation critical value

(Z)
2.83* 4.00*

(Z)
3.94* 5.19* 0

(Z) .44
1.61

4.65* 3.06* 1.52 2.73* 6.70*

0.90
0.30 0.04 0.25 5.65*

4.27* 3.23* 1.93 3.08* 2.79*

Note. Dashes indicate not enough data were available for an analysis. *p < .05.

largest average effect size was found with studies when both children were at least school-aged (r = .52). The average effect size was lower when both children were toddlers (r = .40) or when both children were infants (r = .44). Follow-up contrasts indicated that studies with school-aged children had significantly greater ES than studies with toddlers. With regard to age of child and maternal behavior in the across-situations domain, the effect sizes unexpectedly revealed that parenting was more stable with infants than with toddlers (only one study included children in the school-age range). Studies conducted with infants had a larger effect size (r = .40) than those conducted with toddlers (r = .19). The second moderator variable examined was the length of observation. It was expected that studies with longer observational periods would result in more similar maternal behavior than studies with shorter observational periods. Analyses provided some support for that hypothesis. In the across-time domain, maternal behavior was more stable for observations that lasted 30 min to 59 min (r = .52) compared with shorter observations (r = .41). However, the longest observations (r = .41) were less stable than the moderate length observations as well. The three levels were significantly different, Qb = 6.16, p < .05. Observation length as a moderator variable was also examined in the across-situations data for short and moderate observations (only one study had an observation length of one hour or more). The results were in the predicted direction, but no significant difference was found between the short and moderate observation lengths, Qb = .48, p = .5. Comparable analyses could not be computed in the acrosschildren domain because all but one of the eight observational studies assessed parental behavior in less than 30 min. The remaining four moderator variables were not able to be tested in more than one domain. Two moderator variables (assessment method & duration of time between assessments) were unique to the across-time studies. It was predicted that studies using questionnaire or interview methodologies would reveal more similarity than studies using an observational methodology. The results were consistent with this hypothesis. Effect sizes for studies using self-report data averaged .50 compared with an average effect size of .41 for behavioral measures. It was also predicted that shorter time spans between visits would result in greater stability of child-rearing behavior. The moderator, when divided

240

HOLDEN AND MILLER

Table 11 Moderator Variables Associated With Similarity Analyses at the Study Level Across time Variable Child age Infant (<1 year) Toddler (1-5 year) School-age (>5 years) Observation length Short (<30 min) Moderate (30-59 min) Long (a60 min) Assessment method Observation Self-reports Assessment interval Short (<6 mos) Medium (>6 < 12 mos) Long (>12 mos) Child age discrepancy Same age Different age Type of situations Different settings Different tasks
k
43 23

Across children within 227.79*** 126.70*** 57.77*** 43.29*** 173.90*** 22.90** 22.60*** 128.50*** 221.70*** 157.80*** 64.90*** 245.29*** 10.58 78.26*** 156.46***

Across situations
Q within k
11

r
.40 .44 .53

Q between

k 10 3 3 4

r
.44 .40 .52

Q between

r
.40 .19

Q between
4.15*

Q within
3.59 0.79 2.80

17.84***

5.89*

15 5 31
8 5

41.8*** 1.64 1.36 38.80***

3
8

6.16*
.41 .52 .41

10
7

0.48 .25 .33

20.70** 18.8** 1.90

18
43 33 10 47 7 17 23

7.20**
.41 .50

13.41**
.52 .38 .46

10 6 4

5.56*
.43 .52

42.12*** 3.32 38.80***


12 3

0.22
.22 .27

2.38 20.60**

*/?<.05. **p<.01. ***/><.001. into three levels, was significant, Qb = 13.41, p < .01. Follow-up contrasts indicated that the average effect size for studies with the shortest time spans (r = .52) was larger than the ES for moderate time spans (r = .38) but not significantly different than the ES for time spans over 12 months (r = .46). The moderator variable unique to the across-children domain concerned the age differences between children at the time of assessment. Some studies included assessments of the children when they were the same age (k = 6) in contrast to other investigations that assessed the children at the same point in time and with the children being two or more years in age apart (k = 4). The effect sizes for the two levels of the moderator variable were significantly different, Qb = 5.56, p < .05, with studies assessing maternal behavior with children when they were different ages at the time of assessment being more stable (r = .52) than in studies where the mothers were observed longitudinally and the children were the same age (r = .43). The final moderator variable that was examined occurred in the across-situations domain: those studies that manipulated the settings (k = 3) in contrast to those that explored different tasks within the same setting (k = 9). No difference was found between the two situation types, Qb = .22, p = .65. ers were behaving differently. The largest ES was found in the across-situations studies (d = .52, CI = .417.64, k = 13, n = 605; Q = 47.14, p < .7), followed by the across-time results (d = .40, CI = .347.47, k = 27,n = 2,014; Q = 37.48, p < .2) and then the across-children studies (d = .23, CI = .127.34, k = 7, n = 640; Q = 16.71, p < .01). The across-situations and across-time analyses were homogeneous in contrast to the significant Q statistic found in the across-children data. The fail-safe Ns for the acrosstime and across-situations domains were substantial (3,124 & 809, respectively). In the case of the across-children data, the fail-safe number still represents a considerable number of studies necessary to refute the conclusions found here (n = 150). Contrasts revealed that the across-situations ES was significantly larger than that of the across-children ES (zsc = 2.54). The across-time effect size did not differ significantly from either domain.

Results at the Individual Construct Level


Each of the 11 child-rearing constructs had at least three studies that reported difference tests in one or more domains. The most data were found for the across-time analyses. Twenty-seven longitudinal studies provided difference data; an average of 9 studies (range = 3 to 16) contributed to each of the 11 child-rearing constructs. The results, listed in Table 12, indicate that significant differences were found on all the constructs, with a median effect size of .36 (range = .27 to .71). However, only two constructs provided homogeneous results. Longitudinal child-rearing differences were, not surprisingly, largest with the construct caregive (d = .71), although that finding was qualified by its heterogeneity. The two other variables showing the greatest difference across time were verbalization (d = .57) and stimulation (d = .52). Monitor and noninvolvement, though limited by few studies, showed the least amount of difference (rfs = .27, .29, respectively).

Differences in Child Rearing


Just over half (55%) of the studies provided data for tests examining differences in child rearing. The difference results are reported at the study level and the individual constructs level.

Results at the Study Level


Tests of differences at the study level revealed statistically significant effect sizes for all three domains, indicating that moth-

SIMILARITY IN PARENTING

241

Table 12 Meta-Analytic Results for Child-Rearing Constructs: Differences Across Time


Construct Caregive Control Encouragement Interaction Monitor Negative affect Noninvolvement Positive affect Responsiveness Stimulation Verbalization < .05.

d
71*** .36*** .40*** .34*** 27*** .38*** 29** .32*** 34*** .52*** .57***

k
9 16 5 13

N
388 1,153 218 652

Q
66.72*** 26.92* 9.54* 19.24 3.01 17.83** 5.88 33.28** 1.74 19.08* 30.14***

95% confidence interval: lower/upper .567.86 .287.44 .21/.59 .237.45 .147.40 .287.48 .097.50 .247.41 .217.47 .387.66 .397.74

Fail-safe N
683 763 77 371

4
7 3

448
777 185 1,053 482 400

37
273 15

16
8 10

756
157 379

273

278

**/?<.01.

* * * / > < . 001.

In the across-children domain, only six constructs could be tested (see Table 13). Difference tests were significant for all six, with effect sizes in the small to moderate range. The three constructs manifesting the most difference across children were positive affect (d = .75), control (d = .50), and interaction (d = .49). However, some of the effect sizes were heterogeneous and were based on an average of only four studies. In contrast, all but four of the 10 constructs tested in the across-situations domain showed homogeneous results in the small to moderate range (.23 to .72) as can be seen in Table 14. Differences in parenting across contexts were most pronounced for interaction (d = .72), verbalization (d = .70), and responsiveness (d .65); the smallest difference effect sizes occurred with negative affect (d = .23), positive affect (d = .30), and noninvolvement (d = .31). Comparison of Effect Sizes Across Domains Contrasts examining differences among the effect sizes for each construct across the domains revealed that only one of the 22 contrasts was significant. For the construct interaction, the acrosssituations effect size was larger than the across-time effect size (z = 2.45), indicating that more difference was found in studies

assessing levels of maternal interaction across different contexts than over time. Moderator Variables Homogeneity tests for the studies examining differences in child-rearing behavior revealed that at the study level, only the across-children results were heterogeneous. Because of the limited number of studies that contributed to the difference analyses in the across-children domain (k = 7), moderator variables were not examined. Direction of Change The direction of change in individual constructs could only be examined in the difference data for the across-time domain. We assessed patterns of change for the 11 constructs by examining the number of studies in which a particular construct showed a significant increase, decrease, mixed result, or nonsignificant change (see Table 15). If there were multiple measures of the same construct in a particular study, they were averaged. The clearest trend was with caregive. The mean level decreased in a majority (67%) of the studies (k = 6), showed mixed results in one study,

Table 13 Meta-Analytic Results for Child-Rearing Constructs: Differences Across Children


Construct Caregive Control Encouragement Interaction Monitor Negative affect Noninvolvement Positive affect Responsiveness Stimulation Verbalization

d
.50*** .32** 49**

k
4 4 2 5 3 3

19.13*** 10.61* 0.93

95% confidence interval: lower/upper

Fail-safe

.22***

.75***


33**

177 202 87 544 129 127

.287.71 .127.51 .187.79

70
35 12 92 85 15

.107.34

19.11***
4.31

.49/1.00


.097.58

1.61

Note, Dashes indicate not enough data were available for an analysis. *p < .05. **/?<.01. * * * / > < . 001.

242

HOLDEN AND MILLER

Table 14 Meta-Analytic Results for Child-Rearing Constructs: Differences Across Situation


Construct Caregive Control Encouragement Interaction Monitor Negative affect Noninvolvement Positive affect Responsiveness Stimulation Verbalization

d
49*** .48*** .46*** .72***

k
3 5 5 7 4 4

N
128 227

Q
2.27 10.88* 2.21 3.92

95% confidence interval: lower/upper .247.73

Fail-safe
N

.297.67
.197.73 .547.91

108
239 127 88 214 232 531 367

.23 .31*
.30** .65*** 52*** .70***

1.59 21.70*** 46.76*** 5.29 13.40 15.68*

6
4 11 8

-.017.48 .007.62 .117.50


.467.84 .407.65

31 127 34 298 5 30 47 143

686
544

.557.85

Note. Dashes indicate not enough data were available for an analysis. *p<.05. ** p < . 01. ***p<.001. and was not significant in two other studies. The two other apparent patterns of change were limited by the small number of studies: Monitor decreased over time in two of the four studies that assessed it but had nonsignificant results in other two studies. In contrast, noninvolvement increased in two of the three studies in which it was included but it was nonsignificant in the remaining study. The other eight constructs showed differing degrees of mixed, nonsignificant, or conflicting results. Summary of the Similarity and Difference Results difference, and the across-situations effect revealed the most difference. As predicted, the across-situations data, compared with the two other domains, revealed the greatest level of variation in child rearing. The findings at the individual construct level paint a slightly different picture of the degree of similarity and difference across the three domains. Once again, it was clearly the across-situations analyses that yielded the greatest differences in parental behavior. Out of the 15 similarity contrasts across domains involving acrosssituations studies, two thirds of the contrasts indicated that crosssituation effect sizes were significantly less similar than the comparable effect sizes in one of the other two domains. The remaining similarity contrasts indicated no significant differences across domains. For the difference comparisons, only one contrast was significant, but once again the result indicated greater variability across different contexts than over time. In line with the predictions, but in contrast to the finding at the study level, more similarity was found in individual constructs across time compared with across children. Out of the seven similarity contrasts computed, three were significant. In each case, the construct when assessed across time showed more similarity than when assessed across children. The two domains did not differ significantly in effect size when difference scores were examined. With regard to the size of the effects, the strongest evidence for similarity in parenting was found in the across-time domain. Sixtyfour percent of the child-rearing constructs showed a large effect size, and another 27% showed moderate levels of similarity. Similarly, 57% of the across-children analyses showed large effect sizes, with the remaining ones indicating moderate levels. However, in the case of the constructs in the across-situations domain, only one (13%) had a large similarity ES, and 50% had medium effect sizes. The difference analyses showed the converse: Acrosssituations effect sizes were the largest. Seventy percent of the individual constructs in this domain had effect sizes at .46 or over. In contrast, only 50% of the across-children and 27% of the across-time effect sizes were of this magnitude or greater. Although the constructs in general showed the pattern of more stability across time and the least across situations, there were some differing patterns of variability on particular constructs. For

This meta-analysis has found evidence for both similarities in child rearing as well as differences. Results at the study level indicate that the across-children data showed the highest correlations, followed closely by the across-time and then the acrosssituations data. The results from the analyses of means were complementary: Across-children data showed the least amount of Table 15 Summary of Direction of Change by Construct at the Study Level
Percentage of studies Nonsignificant

Increased Construct Caregive Control Encouragement

Decreased

Mixed

% 25 20 7 43 67

k
4 1 1

% 67 6 20 23 50 14

k
6 1 1 3 2

% 11 19 20 46 25 25 50 57

k
1 3 1

% 22 44 20 15 50 14 33 25 30 14

k 1 7 1

Interaction
Monitor Negative affect Noninvolvement Positive affect Responsiveness Stimulation Verbalization

3
2 1
4

6
50 20 29

2 2

38 13

2 6 1

6
4 2 5 4

2
2 1 1 4

3 1

Note. Some rows do not add up to 100% because of studies that could not be classified because of insufficient information. Dashes indicate not enough data were available for an analysis.

SIMILARITY IN PARENTING

243

example, as one would expect, caregive (assessed only across time) had the lowest similarity ES but the highest ES of the difference scores. In contrast, negative affect showed large effect sizes for similarity across time and children, along with small difference effect sizes in all three domains. The pattern for control was different. Although considerable child-rearing stability over time was found, there were also moderate levels of variability present in parenting across offspring and situations.

Discussion
The results of this review reveal a different picture of child rearing than is commonly depicted in the research literature. When the data are derived from a moving picture of parent-child interactionspanning time, different children, or multiple contexts they show that child rearing can indeed be characterized by variation and change. At the same time, evidence was found for the utility of the snapshot: Child-rearing practices assessed at one time can reflect an enduring characteristic that persists over time, across different offspring, or to a less extent, across settings or tasks. Ironically, several child-rearing variables, such as responsiveness, positive affect, and control, recognized as fundamentally important in parenting (Holden, 1997), were found to be exemplars of both similarity and difference. Although the two sets of analyses computed here assessing change (through relative rankings and mean levels) are independent, the analyses provided converging results. Consequently, we conclude that the nature of child rearing is simultaneously enduring and different. However, the meaning of differences in child rearing deserves some clarification. In fact, two very different types of difference are implied. When assessed in the across-children or across-situations studies, the nature of the difference reflected a capacity to adjust or modify child-rearing behavior in response to the child or to the immediate context. In the across-time studies, the assumption is that the difference in child rearing reflects a more durable type of transformation. To observers of parental behavior, the verdict that child rearing is both enduring and different is not surprising. Perhaps what is remarkable is the relative lack of research attention that has been devoted to parental variability and change. Below we discuss why these two qualities of child rearing coexist, the limitations of our data base, and the theoretical and empirical implications of the coexisting child-rearing characteristics.

of analysis for understanding patterns of behavior than personality characteristics alone. However, the determinants of child rearing are further complicated by the three-way interaction of the parent- child-situation. Indeed, child-rearing stability can be maximizedor minimizedby the extent to which the measurements are kept within particular child-rearing domains. This review has systematically analyzed three types of domains: time, children, and situations. At the same time, there are many other ways in which domain specificity of social development can be conceptualized (see Bugental & Goodnow, 1998). For example, Bornstein and his colleagues (e.g., Bornstein, Vibbert, Tal, & O'Donnell, 1992) have identified two different modes that caregiver interaction takes with infants: social and didactic. These interaction modes are orthogonal, and consequently one can expect to find greater stability within than across domains. Fiske (1992) has proposed another type of parent-child domain. He posited that there are four elementary and universal forms of interactions that can characterize social relationships (communal sharing, authority ranking, equality matching, and market pricing). Children progress through these modes in a fixed ontogenetic order, according to Fiske. If this theory is accurate, it follows then that child rearing should be more stable while within a particular relationship. Indeed, developmental transitions, such as the onset of walking or puberty, can be thought of as representing a domain change and are thus likely to evoke new patterns of child rearing (e.g., Campos et ah, 1992; Steinberg, 1981). One other example of a domain-specific model of child rearing is especially pertinent to issues of stability and change. Smetana (1997) recently argued that child rearing is strongly affected by whether the issue at hand concerns a moral, social-conventional, prudential (safety), or personal topic. Given that parents differentiate among and respond differently to the nature of the concern, child-rearing similarity is likely affected. A second reason why it is important to recognize what the particular domain is is that across time, parents' focus of concern will shift from prudential concerns to moral and social-conventional transgressions. Once again, the clear prediction is that within a domain, greater similarity in child-rearing practices will be found, compared with across domain. The three examples above of child-rearing domains illustrate that the degree of similarity found can be a function of the domain specificity and child's age.

The Coexistence of Differing and Enduring Child Rearing


Why does this paradox exist? The explanation, we believe, lies in the interplay of the methodology and the phenomenon of child rearing. Four considerations help to reveal the nature of the paradox: the domain of analyses, the particular construct assessed, the level of analysis, and the methodological design.

The Child-Rearing Construct


Even more specific to the nature of the child-rearing interaction than the domain is the particular construct under examination. Constructs vary on multiple dimensions, but the most important distinctions here are the degree to which the construct is linked to the child's developmental level and the extent to which it is centered in the child, the parent, or the dyadic relationship. Three constructs that are closely tied to a child's developmental level are caregive, stimulation, and verbalization. In contrast, a variable that is less closely aligned with the child's developmental level is the parent's emotionality (positive or negative affect). In the acrosstime analysis of differences, the first three constructs had moderate effect sizes, in contrast to the small effect sizes of the two affect variables. Consequently, maternal emotionality showed greater

The Domain of Analysis


Perhaps the clearest result from this meta-analysis, after the finding of both stability and change, comes at the domain level: The least amount of similarity and greatest difference was found in the across-situations analyses. This finding is reminiscent of Mischel's (e.g., 1984) and others' (e.g., Magnusson & Stattin, 1998) argument that the person-situation interaction is a better unit

244

HOLDEN AND MILLER they have been studied in the literature, would also fit at this level of specificity. The final class of variables, and the one most susceptible to contextual considerations, is observed behavior. Such behavior may reflect enduring patterns, but it is also at the mercy of the child's behavior or any of the influences listed in Table 2. Partial support for this scheme can be found across domains. Child-rearing practices across situationswhich relied exclusively on observational datashowed the most difference among the three domains. The cross-situational correlations (ranging from .06 to .38) indicate that parenting is considerably less consistent than what has been found in the person-situation literature (e.g., Funder & Colvin, 1991). This inconsistency can be accounted for in several ways. One might expect similarity across situations if the parent was exposed to similar contextual variables or held similar goals for the situation. However, the methodology used likely accentuated differences by comparing child rearing across divergent situations. For example, all 10 studies that conducted laboratory observations of mother-child behavior presented different tasks for each situation, such as comparing parental behavior in block construction or story retelling tasks (e.g., Pratt, Kerig, Cowan, & Cowan, 1988). Similarly, home and laboratory comparisons typically involved very different situations. For example, Crockenberg and Litman (1990) contrasted observations of a laboratory clean-up task with dinner preparation time.

similarity over time in terms of mean levels than constructs that are more child-centered. The variable that appears to be most parent-centered of all the 11 constructs, and thus reflects individual parental differences, is monitoring. Although it has not been assessed in many of the investigations included in this meta-analysis, more similarity was found in this construct than many of the others. The explanation for this lies in that monitoring likely depends more on such determinants as enduring interest in a child and conscientiousness than on the child's actions or developmental level (Crouter, MacDermid, McHale, & Perry-Jenkins, 1990). Finally, the constructs responsiveness and interaction are dyadic in nature, and, depending on how they are operationalized, may manifest either similarity or difference.

Level of Analysis
Another key to the enduring-different paradox lies in the level of analysis. As was revealed in the tests of moderators, similarity varied as a function of the level of analysis. This proposition could not be fully tested but was partially supported by the finding that verbal reports (most often assessing global attitudes) were more stable across time than behavioral data. That observation leads to another: Child-rearing variables are conceptually nested within each other, with lower-level ones reflecting more context-specificity and therefore being multiply determined. Although there are insufficient data to test this proposition, there appears to be a hierarchy of child-rearing variables. At the most general level are the overarching parental traits or styles. Despite the lack of data concerning the stability of styles (Darling & Steinberg, 1993), it is likely that such traits represent the superordinate level of assessment as the other child-rearing variables may be a function of the particular parenting trait. Below traits are global child-rearing values, reflecting parents' enduring principles that by definition transcend specific attitudes, intentions, or typical reactions to a situation or particular child. Child-rearing values concerning morality, respecting authority, individualism, and caring for others are prominent examples (e.g., Kohn, 1979). One step down from values are child-rearing attitudes concerning evaluations of attitude objects. These evaluations are more specific than values as they may take into account some specificity, such as the child's developmental level or gender (e.g., the appropriateness of spanking young children, tolerance of emotional expression in boys vs. girls). Consequently, attitudes are more likely than values to show change over time, be childspecific, or to manifest situational variability. Nevertheless, some attitudes endure. Attitudes toward maternal investment in the motherhood role have been found to be relatively stable over 30 months (r = .65) and even across 6 years (r = .48; Hock, 1988, cited in Hock & DeMeis, 1990). A fourth level of variables that is more responsive to situational considerations is behavioral intentions or self-reported behavior preferences (e.g., Fishbein & Azjen, 1974). Intentions to behave in particular ways take into account some of the situational exigencies but ignore others. Consequently, intentions are likely to be more similar than actual behavior. Child-rearing behavioral intentions are often assessed as responses to hypothetical vignettes (e.g., "How would you behave i f . . . "). Other social cognition variables involved in child rearing (e.g., attributions, perceptions), insofar as

Methodological Design
The moderator analyses computed on the across-time data revealed the impact of several features of the methodological design on the effect sizes. Besides self-reports having a larger similarity effect size than observations, studies using older children showed more similarity than studies using infants (except for across situations); longitudinal assessments less than 6 months apart resulted in greater stability than longer intervals; and studies that used medium-length observations (30 to 59 min) revealed more parental stability than either shorter or longer observations. Part of this last finding was not predicted but may be accounted for by the nature of the observations. Observations that lasted an hour or longer most likely included behavior samples from different activities or contexts and consequently were more varied. Instead, more circumscribed observations, but those long enough to provide a good behavioral sample, resulted in greater stability. As the moderator analyses make clear, a study can be designed to optimize stability or change.

Appraisal of the Database


The conclusions of this meta-analysis are tempered by the number and quality of studies found in the literature. Given the popularity and longevity of research into parent-child relationships, it is likely that we inadvertently omitted some studies that should have been included. However, even if this data set is not exhaustive, we believe it is representative of the extant published and unpublished research. At the same time, there are several limitations to the data set. One consequence of parsing the studies into the framework we used was that 17% of the individual construct effect-size analyses were based on only two or three studies. Such computations warrant caution.

SIMILARITY IN PARENTING

245

A more complete and balanced analysis of child-rearing similarities requires not just a greater number of studies but investigations that include data from fathers; parents from different cultural, ethnic, and racial backgrounds; and studies that more systematically test the moderator variables. For instance, in two studies that included both mothers and fathers, both found lower levels of similarity in fathers' than mothers' behavior (Belsky, Taylor, & Rovine, 1984; Lytton & Zwirner, 1975). Socioeconomic status can also affect similarity in parenting. Thompson (1998) found in his review that attachment relationships in middle-class mother-infant were more stable than in lower socioeconomic status families. More broadly, the data are parochial: Only 3% of the studies included data from a nonwestern country. The database contains other limitations. A full analysis of childrearing similarities and differences requires examination of both correlational data and difference tests. However, both types of analyses were reported in less than one-third of the studies (n = 27). Consequently, in the cross-domain analyses, we were left with comparing rs and ds derived from mostly different sets of studies. Those two sets of studies differed on various dimensions, and particular variables may have been confounded (e.g., age of child, assessment length). Another limitation was that very few studies computed similarity or difference scores separately for boys versus girls, although such a division could be revealing. For example, Kagan and Moss (1962) found a median across-time correlation of .77 for maternal control of girls, but for boys, the comparable correlation was .24. It is worth considering whether this meta-analysis provides an overestimation or underestimation of parental similarity. Overestimation may result from the bias to report at conferences or publish studies that find statistically significant levels of similarity. Unsuccessful investigations are far less likely to be reported in the literature (Roggman, Langlois, Hubbs-Tait, & Rieser-Danner, 1994). However, because significant results could be derived from difference scores as well as similarity scores, we do not think the meta-analytic results overestimate child-rearing similarity. Alternatively, it can be argued that parental similarity has been underestimated for at least two methodological reasons. Most (82%) of the studies included used observational procedures. However, one problem with all of the observational studies reported here is the failure to heed Epstein's (1979) warning: Attempts to characterize typical or stable behavior cannot be effectively accomplished with only one observation. Assessments of the shortterm similarity of child-rearing behavior are generally low; the median test-retest reliability of observations was only r = .58. A more reliable index of typical behavior can be achieved through aggregation, as Wachs (1987) and Leyendecker et al. (1997) have demonstrated with observations of mother-infant interactions. However, if data are to be aggregated, it should only be done within the same situation and child so as not to lose valuable information about how parents adjust their behavior across situations and children. Another reliability issue concerns the issue of interobserver reliabilities. Low reliability attenuates results. If researchers corrected for the reliability of the measurement instruments, then the effect sizes may have increased (Alder & Scher, 1994). Observational studies may also have underestimated parental similarity because of the lack of behavioral ratings (used in only about 10% of the across-time studies). Molecular variables that are typically

coded in observations capture only a modest amount of stable individual-difference variance because they are influenced by interactional and contextual factors (Cairns & Green, 1979). A fuller picture of the extent of similarity in parenting should consist of a comprehensive multimethod approach using observer ratings, observation codings, and self-reports. A final limitation of the data set lies in the type of analyses reported. This meta-analysis was limited to studies that reported correlational or difference tests. That is not to say that there are no other approaches to addressing the question of similarity in child rearing. As Alder and Scher (1994) argued, growth-curve analyses provide an alternative method for capturing both consistency in absolute values as well as individual differences. To date, this approach has rarely been used in the study of parent-child relations (cf. van den Boom & Hoeksma, 1994). Another analytic approach to assessing continuity and change that occurred too infrequently to be summarized was trend analyses, a procedure that allows for a more complex analyses of change using three or more assessment times. For example, McNally, Eisenberg, and Harris (1991) found a quadratic trend in attitudes toward control, whereby after relative stability, there was an increase in views about control during midadolescence. Complex patterns of nonlinear change have also been observed with Egyptian and Kenyan caregivers. Sigman and Wachs (1991) found linear, cubic, and quadratic trends in child-rearing behaviors when they observed interactions between toddlers aged 18 to 29 months and their caregivers. Despite these limitations with the data set, the fail-safe n computations give us confidence in the findings. Those results hold several important theoretical and empirical ramifications for studying and understanding the significance of child rearing.

The Implications of Variability and Change in Child Rearing


If researchers into socialization are to achieve a comprehensive understanding of child rearingits origins, nature, and effects then parents must be studied more intensively. The fundamental implication from this study is the need to recognize both similarity and differences within parents. Recognition of this idea has implications for theory, the understanding of how child rearing impacts ontogeny, and future empirical work.

Theoretical Implications
Much like the relations between heredity and environment or assimilation and accommodation, attempts to dichotomize child rearing as being either similar or different are clearly misguided. Given that child rearing is an interpersonal activity that reflects the constant interplay and coordination of goals between at least two individuals (Maccoby, 1992), such dynamics cannot be reduced into one category or another. On the basis of this review, we would refine the definition of parenting to be an individual's adaptation to three sets of variables: (a) his or her own contemporaneous internal cognitive and affective factors, (b) a particular child, and (c) the context. Such an adaptation may rely on old solutions or elicit novel responses. These dynamics have not yet been adequately recognized or explicated in the theoretical conceptualizations about parenting. Perhaps the most unequivocal theoretical implication of this

246

HOLDEN AND MILLER behavior and child outcome, it is time to incorporate more refined views of the nature of child rearing into our models of parentchild relationships. Greater clarification of the nature of child rearing may result in new insights into how parenting is associated with child outcomes. Variability in parenting across time. The search for main effects (or even interactions) of stable child-rearing characteristics with child outcomes is too limited. As Baumrind herself has recognized, "... parents who are highly effective at one stage in the child's life are not necessarily as effective at another; . . . similar practices do not necessarily produce the same effects at successive stages in child's life" (Baumrind, 1989, p. 189). Some parents may thrive with infant caregiving tasks but are overwhelmed when dealing with a defiant toddler. Other parents may feel lost when relating to their pubescent child but may be adept at providing guidance a few years later. Empirical support for the suggestion that a parent may function more or less effectively at different developmental levels can be inferred from some of the infant-attachment research. Although maternal sensitivity during the first 6 months of a child's life has repeatedly been shown to be associated with a secure attachment at 12 months of age, five different studies have been unsuccessful at finding differences during the second half of the first year between the mothers of infants who developed an insecure attachment from mothers of future-secure infants (see review by Isabella, 1995). There are several explanations for this finding (e.g., maternal sensitivity may be most influential early in ontogeny; it may be more difficult to make valid assessments of sensitivity in the second half of the first year), but it is possible that mothers of future-insecure infants were able to parent more effectively during that later time period. Another implication for children's outcomes that stems from this review is that parents' capacity for change over time might relate to the quality of their parent-child relationship. Parents who are able to modify their views and interests, as well as adjust their child rearing to better match their children's needs, may be rewarded with better communication, more influence, and a closer relationship with their offspring. Variability in parenting across children. Variability in parental effectiveness may also be at work across siblings. For a variety of reasons, ranging from life circumstances such as divorce to child characteristics (e.g., gender, temperament, or wantedness), a parent may feel closer and more effective with one child than another. One ramification from this observation is that it is likely that parents do not have a blanket effect on all of their children. As behavioral geneticists have argued, if researchers are to understand how social development affects children, then much more attention needs to be devoted to within-family, nonshared environment (e.g., Plomin, 1994). The most important place to observe these nonshared differences lies in children's understanding of, and reactions to, differential parental treatment. Indeed, it may be that children's perceptions of parental behaviornot the behavior itselfrepresents the active ingredient of child rearing. Investigations by Dunn and her colleagues involving siblings provide some hints of what a revised emphasis on variation in parents' behavior may reveal. Parents who manifest differential behavior to their children, at least as perceived by the children, promote greater sibling hostility (Boer & Dunn, 1992). In a similar vein, Kowal and Kramer (1997)

review is that it has highlighted the context-dependent nature of child rearing. The person-situation interaction in child rearing has only sometimes been appreciated (e.g., Luster & Okagaki, 1993). New models of parents must recognize the ecological contexts within which parenting is embedded (e.g., Belsky & Vondra, 1985) but must also allow for variability in response as a consequence of the interactional context (Bronfenbrenner & Crouter, 1983; Lerner, 1989). Two recent theoretical conceptualizations of parenting that do recognize the variable nature of parenting along with the interaction context have been articulated by Grusec and Goodnow (1994) and Dix (1992). Both models focus on microprocesses associated with parents' capacity for adjustment and change. In their reconceptualization of the relations between parental discipline and child internalization, Grusec and Goodnow explicitly recognized the variation inherent in parents' disciplinary responses to children's misdeeds. Although their analyses focused on betweenparent differences, many of their points hold for within-parent variability as well. For instance, they argued that parental disciplinary reactions reflect internal dispositions, perceptions of the child (e.g., temperament, mood), and the type of child transgression (e.g., moral vs. social conventional). These three factors are likely to interact differently under different conditions. A second theoretical model of child rearing that embraces parental variation in the form of moment-to-moment adjustments has been proposed by Dix (1992). Centered around how emotions provide an organizing framework for parenting, Dix argued that parental emotionand behavioris activated by a complex relation between goals, concerns, and parental appraisals of ongoing events. Child rearing then represents the outcome of multiple transactional processesbetween parental social cognition, the child, and the context. Recently, Kuczynski and Lollis (in press) have articulated a model of parenting that nicely captures the fluid nature of child rearing. They took issue with the all-too-common unidirectional model of parenting that views the parents as powerful and active, and the children as passive and largely powerless. Instead, parenting is the outcome of a bidirectional process that involves interdependent power relationships as well as behavioral and cognitive agency of both parent and child. Child rearing, according to Kuczynski and Lollis, is a dynamic process that undergoes frequent adjustment as a consequence of variables residing in the parent, child, parent-child relationship, or context. Viewing parenting from such a perspective allows for a better appreciation for how child-rearing similarities and differences may coexist and comingle.

The Relation of Child Outcomes to Child-Rearing Variation and Change


To date, the trait approach to parenting has been the most effective way to examine associations between parenting and child outcome (Holden, 1997; Maccoby, 1992). However, such an approach does not invite development-related questions, nor does it allow for bidirectionality or the fact that a child's characteristics may affect or interact with a particular parent trait (Darling & Steinberg, 1993; Lewis, 1981). Although trait views of parents have served an important role in focusing attention on particular characteristics of parents and establishing links between parental

SIMILARITY IN PARENTING

247

recently reported that children who perceived their parents' differential sibling treatment to be fair had more positive appraisals of their sibling relationships than other children. Variability in parenting across situations. This review raises questions about the underlying reasons for and constraints on variations in child-rearing across situations. This is most conspicuous with the limited understanding or parental sensitivity (or responsivity). Sensitivity is a complex construct but is commonly defined as "contingent, appropriate [italics added], and consistent responses to an infant's signals or needs" (Lamb & Easterbrooks, 1981, p. 127). Surprisingly, the context-specific nature of "appropriate" has not been carefully examined. Little is known about why some parents are able to navigate through successive and sometimes challenging contexts (such as the supermarket, Holden, 1983) with deftness and propriety, whereas other parents may lose tempers or control. Furthermore, as De Wolff and van Ijzendoorn (1997) pointed out, a more contextualized view of sensitivity is needed to account for parent-infant attachment patterns in home environments characterized by stress and instability in attachment relationships. A concept closely related to sensitivity is flexibility. Variability in parental behavior, an expression of flexibility, is thought by some investigators to be central for effective parenting. As Mischel (1984) recognized some time ago, greater consistency in behavior is displayed by individuals who are functioning poorly. Along those lines, when discussing dysfunctional mothers, Wahler and Dumas (1989) observed "Mothers who attend to the complex patterns of child-care stimuli will also perform a set of highly relevant parenting behaviors . . . mothers who are deficient in observational processes are prone to develop response-response linkages that permit stimuli in one setting to influence her [sic] behavior in other settings. This is tantamount to saying that attention-deficit mothers are marked by trait-like behavior patterns [italics added]..." (p. 123). Similarly, Grusec and Goodnow (1994) proposed that flexibility in discipline may be more important in terms of teaching the child than the particular method used: " . . . parents [must] be flexible in their disciplinary reactions, matching them to the child's perceptions of and reactions to the conflict situation: Effective parenting involves sensitivity to the child's emotional state and cognitions" (p. 17). Several other researchers have also called for the study of flexibility in parents (Hoffman, 1970) or have recognized how effective parents must be flexible to balance child-like and adult perceptions (Maccoby, 1992), competing needs (Dix, 1992), or the dialectical process of resolving conflicting considerations (Holden & Ritchie, 1988). Thus, it appears that a key parenting characteristic that has gone ignored empirically is the ability to exhibit flexibility. Parental flexibility is not synonymous with inconsistent discipline (e.g., Patterson, 1982). Rather, like sensitivity, effective use of the principle of consistency may necessitate taking into account the immediate situational and child factors, such as intentionality and recent interactional history. The implication of Grusec and Goodnow's (1994) analysis is that the child's perception of parental equity and justification rather than consistency per se is what is most important. Presumably, children of flexible parents feel more respected as individuals, develop better relationships with their parents, and acquire a greater sense of self-efficacy than children of parents who are unjustly rigid or easily manipulated.

Empirical Investigations Into Parental Variation and Change


This new emphasis on parental variation and change must be supported with empirical evidence into the types, sources, and consequences of parental variation and change. Some of this work has begun, as was indicated in the introduction. However, a new conceptualization invites many other novel research questions. In addition to the research suggestions mentioned above, we identify three areas as most pressing: inquiries into the nature of parental variation, systematic research into multiple levels of parental variables, and investigations into the psychological processes associated with variation and change. The nature of parental variation. Systematic analyses into parental variation and change across time, children, and situations are needed to begin to reveal the nature of the shifting landscape. Just as substantial differences in the quality of marital relations are now widely recognized to occur over relatively short periods of time (e.g., Belsky & Rovine, 1990; Cowan & Cowan, 1992), variation and change in parenting also warrant examination. As this review has revealed, there is substantially more information about how child rearing changes over time than across children or situations. Consequently, there is a greater need for investigations into the latter two domains. For example, research is needed to address why parents interact differently with different children. Similarly, much of parents' situationally induced behavioral variation is not random; rather we believe that it is lawful and that lawfulness deserves study. Empirical work should strive to reveal the ways in which parents modify their behavior in response to the characteristics of situations. For example, Valsiner (1984) took a Vygotskian approach when he proposed that within particular contexts, parents have three different zones of child behavior: those that are encouraged, discouraged, and tolerated. In an approach intended to account for the parent-situation interaction between parents and setting, Miller, Shim, and Holden (1999) adopted a Gibsonian perspective of the environment. They used observational data from the home, park, and laboratory to show how settings provided different affordances and demands on parent-child interactions. Multilevel investigations. Empirical work is also needed to explicate the interplay of continuity, variability, and change at each of the five levels of variables distinguished above (trait, value, attitude, behavioral intention, and behavior). To what extent are parenting traits maintained over time, different offspring, and situations? Are child-rearing values maintained over time, or do many parents experience "value stretching" as the sociologist Rodman suggested (cited in Goodnow, 1997). Despite the popularity of studying child-rearing attitudes, scant attention has been devoted to parental attitude change (Holden, 1995). At the level of behavioral intentions, several studies have shown that parents believe they do indeed modify their actions because of the specifics of the situations (e.g., Catron & Masters, 1993; Dix & Reinhold, 1991; Grusec & Kuczynski, 1980). Finally, as this review has shown, there is ample evidence that observed child-rearing behavior can be modified. However, what is less clear is the systematic determinants of those modifications. The psychological processes. Far more important than simply categorizing whether variation and change occur is the need to understand the psychological processes at work. To date there is

248

HOLDEN AND MILLER

some work in this area. The change agent that has been best documented is the child developmental level. As reviewed in the introduction, there is already some evidence for how parenting changes as a consequence of a child's development or as a consequence of family structural changes, such as divorce (Hetherington & Stanley-Hagan, 1995). However, other influences on change in child rearing have received scant research attention. Some of these sources include learning from experiences with infants and children (Holden, 1988), changing practices because of a child's behavioral response (Holden, Thompson, Zambarano, & Marshall, 1997), being influenced by a spouse (McHale, 1995), reading new child-rearing information (Clarke-Stewart, 1978), or profiting from a parent education program (Wierson & Forehand, 1994). The most basic research initiative needed involves a focus on how parenting is organized and what influences the coherence of the individual's organization at given moments. Mischel and Shoda's (1995) theory of personality provides a useful model. They assumed that personality is composed not of isolated tendencies but of a psychologically meaningful organization of relations among cognitions and affect. The researcher's task is to explicate or map the particular domain of interest. A key feature is identification of the conditional probabilities associated with an individual's behavior. In this construct, the momentary activation among cognitions and affect composes the individual's personality state. In the terms of their model, child rearing reflects the dynamic interplay of the parent's personality system and the specific cognitive-affective processes activated at the current momentin response to the child and immediate context. Research into parenting would also profit by explicating how cognitions and affect interact to contribute to parental behavior. On the one hand, it is likely that parenting, like children's behavior, changes over time to become more hierarchical, differentiated, and complex (e.g., Sroufe & Jacobvitz, 1989). However, it is also possible that there are ways in which parental behavior is constrained by earlier experiences or adaptations, such as internal representations of attachment relationships (e.g., Main et al., 1985). Other examples of potential constraints on variation or change include ethnotheories (Harkness & Super, 1995), deeply held values and strong convictions (Abelson, 1988), personality dispositions and orientations toward child-rearing efficacy (Bugental et al., 1990), and emotional or mental health problems, such as depression (Kochanska et al., 1989).

References
References marked with an asterisk indicate studies included in the meta-analysis. Abelman, R. (1985). Styles of parental disciplinary practices as a mediator of children's learning from prosocial television portrayals. Child Study Journal, 15, 131-146. Abelson, R. P. (1988). Conviction. American Psychologist, 43, 267-275. Ainsworth, M. S. D., Bell, S. M. V., & Stayton, D. J. (1971). Individual differences in strange situation behavior of one-year-olds. In H. R. Schaffer (Ed.), The origins of human social relations (pp. 17-52). New York: Academic. Alder, A. G., & Scher, S. J. (1994). Using growth curve analysis to assess personality change and stability in adulthood. In T. F. Heatherton & J. L. Weinberger (Eds.), Can personality change? (pp. 149-173). Washington, DC: American Psychological Association. Amato, P. R., & Keith, B. (1991). Parental divorce and the well being of children: A meta-analysis. Psychological Bulletin, 110, 26-46. Anderson, E. R., Hetherington, E. M., Reiss, D., & Howe, G. (1994). Parents' nonshared treatment of siblings and the development of social competence during adolescence. Journal of Family Psychology, 8, 303320. Anderson, K. E., Lytton, H., & Romney, D. M. (1986). Mothers' interactions with normal and conduct-disordered boys: Who affects whom? Developmental Psychology, 22, 604-609. Baldwin, A. L. (1946). Differences in parent behavior toward three- and nine-year-old children. Journal of Personality, 15, 143-165. Baldwin, A. L., Kalhorn, J., & Breese, F. (1945). Patterns of parent behavior. Psychological Monographs, 58 (3, Serial No. 268). *Bates, J. E., Olson, S. L., Pettit, G. S., & Bayles, K. (1982). Dimensions of individuality in the mother-infant relationship at six months of age. Child Development, 53, 446-461. Baumrind, D. (1971). Current patterns of parental authority. Developmental Psychology Monographs, 4 (1, Pt. 2). Baumrind, D. (1989). The permanence of change and the impermanence of stability. Human Development, 32, 187-195. *Beckwith, L., & Cohen, S. E. (1989). Maternal responsiveness with preterm infants and later competency. In W. Damon (Series Ed.) & M. H. Bornstein (Vol. Ed.), New directions for child development. No. 43. Maternal responsiveness: Characteristics and consequences (pp. 75-87). San Francisco: Jossey-Bass. *Beckwith, L., Rodning, C., & Cohen, S. (1992). Preterm children at early adolescence and continuity and discontinuity in maternal responsiveness from infancy. Child Development, 63, 1198-1208. Bell, R. Q. (1968). A reinterpretation of the direction of effects in studies of socialization. Psychological Review, 75, 81-95. Bell, R. Q., & Chapman, M. (1986). Child effects in studies using experimental or brief longitudinal approaches to socialization. Developmental Psychology, 22, 595-603. *Belsky, J. (1979). Mother-father-infant interaction: A naturalistic observational study. Developmental Psychology, 15, 601-607. *Belsky, J. (1980). Mother-infant interaction at home and in the laboratory: A comparative study. Journal of Genetic Psychology, 137, 37-47. Belsky, J. (1984). The determinants of parenting: A process model. Child Development, 55, 83-96. *Belsky, J., Gilstrap, B., & Rovine, M. (1984). The Pennsylvania infant and family development project, I: Stability and change in mother-infant and father-infant interaction in a family setting at one, three, and nine months. Child Development, 55, 692-705. Belsky, J., & Rovine, M. (1990). Patterns of marital change across the transition to parenthood: Pregnancy to three years postpartum. Journal of Marriage and the Family, 52, 5-19. *Belsky, J., Taylor, D. G., & Rovine, M. (1984). The Pennsylvania infant

Conclusion
Acceptance of the assertion that child rearing is simultaneously enduring and different implies a fundamental change in the way that parents are thought of and studied. Thus, in addition to recognition of the enduring characteristics of parents, the question of paramount importance is under what conditions, in what ways, to what extent, and why does child rearing vary or change? In turn, how do children perceive that variation and change? It is likely that only with such information will we be able to fully understand parental behavior and assess its influence on children. Such information will also lead to better education programs for parents as well as more effective treatment for those who are experiencing problems in raising their children. It is time to bid farewell to simplehowever parsimoniousviews of child rearing.

SIMILARITY IN PARENTING and family development project, II: The development of reciprocal interaction in the mother-infant dyad. Child Development, 55, 706-717. Belsky, J., & Vondra, J. (1985). Characteristics, consequences, and determinants of parenting. In L. L'Abate (Ed.), The handbook of family psychology and therapy (Vol. 1, pp. 523-556). Homewood, IL: Dorsey. Bern, D. J., & Allen, A. (1974). On predicting some of the people some of the time: The search for cross-situational consistencies in behavior. Psychological Review, 81, 506-520. Bern, D. J., & Funder, D. C. (1978). Predicting more of the people more of the time: Assessing the personality of situations. Psychological Review, 85, 485-501. Benoit, D., & Parker, K. C. H. (1994). Stability and transmission of attachment across three generations. Child Development, 65, 14441456. Biringen, Z., Emde, R. N., Campos, J. J., & Appelbaum, M. I. (1995). Affective reorganization in the infant, the mother, and the dyad: The role of upright locomotion and its timing. Child Development, 66, 499-514. Boer, F., & Dunn, J. (1992). Sibling relationships: Developmental and clinical issues. Hillsdale, NJ: Erlbaum. *Bohman, T., Hazen, N., Burton, H., & DeSantis, R. (1991, April). Change in parental attitudes toward child-rearing over time. Paper presented at the biennial meeting of the Society for Research in Child Development, Seattle, WA. Bornstein, M. H. (Ed.). (1991). Cultural approaches to parenting. Hillsdale, NJ: Erlbaum. Bomstein, M. H. (Ed.). (1995). Parenting infants. In M. H. Bornstein (Ed.), Handbook of parenting: Vol. I. Children and parenting (pp. 339). Hillsdale, NJ: Erlbaum. *Bornstein, M. H., & Tamis-LeManda, C. S. (1990). Activities and interactions of mothers and their firstborn infants in the first six months of life: Covariation, stability, continuity, correspondence, and prediction. Child Development, 61, 1206-1217. Bornstein, M. H., Vibbert, M., Tal, J., & O'Donnell, D. (1992). Toddler language and play in the second year: Stability, covariation and influences of parenting. First Language, 12, 323-338. *Brody, G. H., Stoneman, Z., & Burke, M. (1987). Child temperaments, maternal differential behavior, and sibling relationships. Developmental Psychology, 23, 354-362. Brody, G. H., Stoneman, Z., & McCoy, J. K. (1994). Contributions of family relationships and child temperaments to longitudinal variations in sibling relationship quality and sibling relationship styles. Journal of Family Psychology, 8, 274-286. Brody, S. (1956). Patterns of mothering: Maternal influence during infancy. New York: International Universities. Bronfenbrenner, U. (1979). The ecology of human development: Experiments by nature and design. Cambridge, MA: Harvard University Press. Bronfenbrenner, U., & Crouter, A. C. (1983). The evolution of environmental models in developmental research. In P. H. Mussen (Series Ed.) & W. Kessen (Vol. Ed.), Handbook of child psychology: Vol. 1. History, theory, and methods (4th ed., pp. 357-414). New York: Wiley. *Bryant, B. K., & Crockenberg, S. B. (1980). Correlates and dimensions of prosocial behavior: A study of female siblings with their mothers. Child Development, 51, 529-544. Bugental, D. B., Blue, J., & Lewis, J. (1990). Caregiver beliefs and dysphoric affect directed to difficult children. Developmental Psychology, 26, 631-638. Bugental, D. B., & Goodnow, J. J. (1998). Socialization processes. In W. Damon (Series Ed.) & N. Eisenberg (Vol. Ed.), Handbook of child psychology: Vol. 3. Social, emotional, and personality development (5th ed., pp. 389-462). New York: Wiley. Bugental, D. B., & Shennum, W. A. (1984). "Difficult" children as elicitors and targets of adult communication patterns: An attributional-behavioral transactional analysis. Monographs of the Society for Research in Child Development, 49 (1, Serial No. 205).

249

Buhrmester, D., Camparo, L., Christensen, A., Gonzalez, L. S., & Hinshaw, S. P. (1992). Mothers and fathers interacting in dyads and triads with normal and hyperactive sons. Developmental Psychology, 28, 500509. *Cain, R., Rabinovich, B., Pedersen, F., Zaslow, M., & Suwalsky, J. (1985, August). Mother-infant interaction in two social contexts at two ages. Poster presented at the annual meeting of the American Psychological Association, Los Angeles, CA. Cairns, R. B. (1979). Social development: The origins and plasticity of interchanges. San Francisco: Freeman. Cairns, R. B., & Green, J. (1979). How to assess personality and social patterns: Observations or ratings? In R. B. Cairns (Ed.), The analysis of social interactions (pp. 37-65). Hillsdale, NJ: Erlbaum. *Caldwell, B., & Bradley, R. (1984). HOME observation for measurement of the environment. Little Rock, AR: University of Arkansas. Campos, J. J., Kermonian, R., & Zumbahlen, M. R. (1992). Socioemotional transformations in the family system following infant crawling onset. In W. Damon (Series Ed.), & N. Eisenberg & R. Fabes (Vol. Eds.), New directions for child development: No. 55. Emotion and its regulation in early development (pp. 25-40). San Francisco: Jossey-Bass. Catron, T. R., & Masters, J. C. (1993). Mothers' and children's conceptualizations of corporal punishment. Child Development, 64, 1815-1828. *Clarke-Stewart, K. A. (1973). Interactions between mothers and their young children: Characteristics and consequences. Monographs of the Society for Research in Child Development, 38 (153, Serial No. 6-7). Clarke-Stewart, K. A. (1978). Popular primers for parents. American Psychologist, 33, 359-369. *Clarke-Stewart, K. A., & Hevey, C. M. (1981). Longitudinal relations in repeated observations of mother-child interaction from 1 to 2 1/2 years. Developmental Psychology, 17, 127-145. Clifford, E. (1959). Discipline in the home: A controlled observational study of parental practices. Journal of Genetic Psychology, 95, 45-82. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (Rev. ed.). New York: Academic Press. *Conger, K. J., & Conger, R. D. (1994). Differential parenting and change in sibling differences in delinquency. Journal of Family Psychology, 8, 287-302. Conger, R. D., Conger, K. J., Elder, G. H., Lorenz, F. O., Simons, R. L., & Whitbeck, L. B. (1992). A family process model of economic hardship and adjustment of early adolescent boys. Child Development, 63, 526-541. Cowan, C. P., & Cowan, P. A. (1992). When partners become parents: The big life change for couples. New York: Basic Books. Crandall, V. J., & Preston, A. (1955). Patterns and levels of maternal behavior. Child Development, 26, 267-277. *Crockenberg, S., & Litman, C. (1990). Autonomy as competence in 2-year-olds: Maternal correlates of child compliance, noncompliance, and self-assertion. Developmental Psychology, 26, 961-971. *Crockenberg, S. B., & McCIuskey, K. (1986). Change in maternal behavior during the baby's first year of life. Child Development, 57, 746-753. Crouter, A. C., MacDermid, S. M., McHale, S. M., & Perry-Jenkins, M. (1990). Parental monitoring and perceptions of children's school performance and conduct in dual- and single-earner families. Developmental Psychology, 26, 649-657. Crouter, A. C., & McHale, S. M. (1993). Temporal rhythms in family life: Seasonal variation in the relation between parental work and family processes. Developmental Psychology, 29, 198-205. Cummings, E. M., & Davies, P. (1994). Children and marital conflict: The impact of family dispute and resolution. New York: Guilford Press. *Daniels, D., Dunn, J., Furstenberg, F., & Plomin, R. (1985). Environmental differences within the family and adjustment differences within pairs of adolescent siblings. Child Development, 56, 764-774.

250

HOLDEN AND MILLER Effects of children's age, sex, and attachment classification. Child Development, 64, 258-271. *Fiese, B. H. (1990). Playful relationships: A contextual analysis of mother-toddler interactions and symbolic play. Child Development, 61, 1648-1656. Fishbein, M., & Ajzen, I. (1974). Attitudes towards objects as predictors of single and multiple behavioral criteria. Psychological Review, 81, 5974. Fiske, A. P. (1992). The four elementary forms of sociality: Framework for a unified theory of social relations. Psychological Review, 99, 689-723. *Frankel, D., Arbel, T., & Mendrovsky, L. (1980). Stability and distinctiveness in interaction of mother and neonate. Psychological Reports, 47, 1103-1108. Funder, D. C., & Colvin, C. R. (1991). Explorations in behavioral consistency: Properties of persons, situations, and behaviors. Journal of Personality and Social Psychology, 60, 773-794. Goodman, S. H., & Brumley, H. E. (1990). Schizophrenic and depressed mothers: Relational deficits in parenting. Child Development, 26, 31-39. Goodnow, J. J. (1995). Parents' knowledge and expectations. In M. H. Bornstein (Ed.), Handbook of parenting: Vol. 3. Status and social conditions of parenting (pp. 305-332). Mahwah, NJ: Erlbaum. Goodnow, J. J. (1997). Parenting and the transmission and internalization of values: From social-cultural perspective to within-family analyses. In J. E. Grusec & L. Kuczynski (Eds.), Parenting and children's internalization of values: A handbook of contemporary theory (pp. 333-361). New York: Wiley. Goodnow, J. J., & Collins, W. A. (1990). Development according to parents: The nature, sources, and consequences of parents' ideas. Hillsdale, NJ: Erlbaum. Gralinski, H. H., & Kopp, C. B. (1993). Everyday rules for behavior: Mothers' requests to young children. Developmental Psychology, 29, 573-584. *Green, J. A., Gustafson, G. E., & West, M. J. (1980). Effects of infant development of mother-infant interactions. Child Development, 51, 199-207. Grusec, J. E., & Goodnow, J. J. (1994). Impact of parental discipline methods on the child's internalization of values: A reconceptualization of current points of view. Developmental Psychology, 30, 4-19. Grusec, J. E., & Kuczynski, L. (1980). Direction of effect in socialization: A comparison of the parent vs. the child's behavior as determinants of disciplinary techniques. Developmental Psychology, 16, 1-9. Grusec, J. E., & Kuczynski, L. (Eds.) (1997). Parenting and children's internalization of values: A handbook of contemporary theory. New York: Wiley. *Haden, C. A., & Fivush, R. (1996). Contextual variation in maternal conversational styles. Merrill-Palmer Quarterly, 42, 200-227. *Hagan, S. M., Hollier, E. A., O'Conner, T. G., & Eisenberg, M. (1992). Parent-child relationships in nondivorced, divorced single-mother, and remarried families. In E. M. Hetherington & W. G. Clingempeel (Eds.), Coping with marital transitions. Monographs of the Society for Research in Child Development, 57, (2-3, Serial No. 227). Harkness, S., & Super, C. (1995). Parental ethnotheories. In M. H. (Ed.), Handbook of of parenting: Vol. 2. Biology and ecology of parenting (pp. 211-234). Hillsdale, NJ: Erlbaum. *Hart, B., & Risley, T. R. (1992). American parenting of language-learning children: Persisting differences in family-child interactions observed in natural home environments. Developmental Psychology, 28, 10961105. *Hatfield, J. S., Ferguson, L. R., & Alpert, R. (1967). Mother-child interaction and the socialization process. Child Development, 38, 365414. Hattick, B. W., & Stowell, M. (1936). The relation of parental overattentiveness to children's work habits and social adjustments in kinder-

Darling, N., & Steinberg, L. (1993). Parenting style as context: An integrative model. Psychological Bulletin, 113, 487-496. *David, A., & Holden, R. H. (1970). Consistency of maternal attitudes and personality from pregnancy to eight months following childbirth. Developmental Psychology, 2, 364-366. De Wolff, M. S., & van Ijzendoorn, M. H. (1997). Sensitivity and attachment: A meta-analysis on parental antecedents of infant attachment. Child Development, 68, 571-591. DiLalla, L. F., & Bishop, E. G. (1996). Differential maternal treatment of infant twins: Effects on infant behavior. Behavior Genetics, 26, 535542. Dix, T. (1991). The affective organization of parenting: Adaptive and maladaptive processes. Psychological Bulletin, 110, 3-25. Dix, T. (1992). Parenting on behalf of the child: Empathic goals in the regulation of responsive parenting. In I. E. Sigel, A. V. McGillicuddyDeLisi, & J. J. Goodnow (Eds.), Parental belief systems: The psychological consequences for children (pp. 319-346). Hillsdale, NJ: Erlbaum. Dix, T., & Reinhold, D. P. (1991). Chronic and temporary influences on mothers' attributions for children's disobedience. Merrill-Palmer Quarterly, 37, 251-271. Dix, T., Ruble, D. N., & Zambarano, R. J. (1989). Mothers' implicit theories of discipline: Child effects, parent effects, and the attribution process. Child Development, 60, 1373-1391. *Dunn, J. F. (1977). Patterns of early interaction: Continuities and consequences. In Schaffer, H. R. (Ed.), Studies in mother-infant interaction (pp. 457-474). London: Academic Press. Dunn, J. F. (1992). Siblings and development. Current Directions in Psychological Science, 1, 6-9. *Dunn, J. F., & Kendrick, C. (1980). The arrival of a sibling: Changes in patterns of interaction between mother and first-born child. Journal of Child Psychology and Psychiatry, 21, 119-132. *Dunn, J. F., & Plomin, R. (1986). Determinants of maternal behavior toward 3-year-old siblings. British Journal of Developmental Psychology, 4, 127-137. *Dunn, J. F., Plomin, R., & Daniels, D. (1986). Consistency and change in mothers' behavior toward young siblings. Child Development, 57, 348356. *Dunn, J. F., Plomin, R., & Nettles, M. (1985). Consistency of mothers' behavior toward infant siblings. Developmental Psychology, 21, 11881195. *Easterbrooks, M. A., & Emde, R. N. (1985, March). When mommy and daddy say no: A longitudinal study of toddler compliance. Poster presented at the biennial meeting of the Society for Research in Child Development, Toronto, Canada. *Eisenberg, N., Wolchik, S., Hernandez, R., & Pasternack, J. (1985). Paternal socialization of young children's play: A short-term longitudinal study. Child Development, 56, 1506-1513. Ellison, C. G., Bartowski, J. P., & Segal, M. L. (1996). Conservative Protestantism and the parental use of corporal punishment. Social Forces, 74, 1003-1028. Emmerich, W. (1964). Continuity and stability in early social development. Child Development, 35, 311-332. Epstein, S. (1979). The stability of behavior: I. On predicting most of the people much of the time. Journal of Personality and Social Psychology, 37, 1097-1282. Epstein, S. (1980). The stability of behavior: II. Implications for psychological research. American Psychologist, 35, 790-806. Faber, A., & Mazlish, E. (1987). Siblings without rivalry. New York: Avon. *Fagot, B., & Gauvain, M. (1997). Mother-child problem solving: Continuity through the early childhood years. Developmental Psychology, 33, 480-488. Fagot, B. I., & Kavanaugh, K. (1993). Parenting during the second year:

SIMILARITY IN PARENTING garten and the first six grades of school. Journal of Educational Research, 30, 169-176. Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta-analysis. Orlando, FL: Academic Press. *Heinicke, C. M, Diskin, S. D., Ramsey-Klee, D. M., & Given, K. (1983). Pre-birth parent characteristics and family development in the first year of life. Child Development, 54, 194-208. Hetherington, E. M., & Stanley-Hagan, M. M. (1995). Parenting in divorced and remarried families. In M. H. Bornstein (Ed.), Handbook of parenting: Vol. 3. Status and social conditions of parenting (pp. 233254). Hillsdale, NJ: Erlbaum. Hock, E., & Lindamood, E. (1981). Continuity of child-rearing attitudes in mothers of young children. Journal of Genetic Psychology, 138, 305306. Hock, E., & DeMeis, D. K. (1990). Depression in mothers of infants: The role of maternal employment. Developmental Psychology, 26, 285-291. Hoff-Ginsberg, E. (1991). Mother-child conversation in different social classes and communicative settings. Child Development, 62, 782-796. Hoffman, L. W. (1991). The influence of the family environment on personality: Accounting for sibling differences. Psychological Bulletin, 110, 187-203. Hoffman, M. L. (1970). Conscience, personality, and socialization techniques. Human Development, 13, 90-126. *Holden, G. W. (1983). Avoiding conflict: Mothers as tacticians in the supermarket. Child Development, 54, 233-240. Holden, G. W. (1988). Adults' thinking about a child-rearing problem: Effects of experience, parental status, and gender. Child Development, 59, 1623-1632. Holden, G. W. (1995). Parental attitudes toward childrearing. In M. H. Bornstein (Ed.) Handbook of parenting: Vol. 3. The status and social conditions of parenting (pp. 359-392). Mahwah, NJ: Erlbaum. Holden, G. W. (1997). Parents and the dynamics of child rearing. Boulder, CO: Westview Press. Holden, G. W., Coleman, S., & Schmidt, K. L. (1995). Why 3-year-old children get spanked: Parent and child determinants in a sample of college-educated mothers. Merrill-Palmer Quarterly, 41, 431-452. Holden, G. W., & Edwards, J. (1989). Parental attitudes toward child rearing: Instruments, issues, and implications. Psychological Bulletin, 106, 29-58. Holden, G. W., & Ritchie, K. L. (1988). Child rearing and the dialectics of parental intelligence. In J. Valsiner (Ed.), Children's development within socio-culturally structured environments (pp. 30-59). Norwood, NJ: Ablex. Holden, G. W., & Ritchie, K. L. (1991). Linking extreme marital discord, child rearing, and child behavior problems: Evidence from battered women. Child Development, 62, 311-327. Holden, G. W., Stein, J., Harris, S., Ritchie, K. L., & Jouriles, E. N. (1998). The parenting beliefs and behaviors of battered women. In G. W. Holden, R. Geffner, & E. N. Jouriles (Eds.), Children exposed to marital violence: Theory, research, and applied issues (pp. 289-334). Washington, DC: American Psychological Association. Holden, G. W., Thompson, L., Zambarano, R., & Marshall, L. (1997). Child effects as a source of change in maternal attitudes toward corporal punishment. Journal of Social and Personal Relationships, 14, 481-490. Isabella, R. (1993). Origins of attachment: Maternal interactive behavior across the first year. Child Development, 64, 605-621. Isabella, R. A. (1995). The origins of infant-mother attachment: Maternal behavior and infant development. In R. Vasta (Ed.), Annals of child .development (Vol. 10, pp. 57-81). Bristol, PA: Kingsley. Isabella, R. A., & Belsky, J. (1991). Interactional synchrony and the origins of infant-mother attachment: A replication study. Child Development, 62, 373-384. Johnson, B. T. (1993). DSTAT 1.10: Software for the meta-analytic review of research literatures. Hillsdale, NJ: Erlbaum.

251

Kagan, J. (1971). Change and continuity in infancy. New York: Wiley. Kagan, J., & Moss, H. A. (1962). Birth to maturity: A study in psychological development. New York: Wiley. Kandel, D. B., & Wu, P. (1995). Disentangling mother-child effects in the development of antisocial behavior. In J. McCord (Ed.), Coercion and punishment in long-term perspectives (pp. 106-123). New York: Cambridge University Press. Kaye, K., & Fogel, A. (1980). The temporal structure of face-to-face communication between mothers and infants. Developmental Psychology, 16, 454-464. Kazdin, A. E. (1987). Treatment of antisocial behavior in children: Current status and future directions. Psychological Bulletin, 102, 187-203. Kendrick, C., & Dunn, J. (1983). Sibling quarrels and maternal responses. Developmental Psychology, 19, 62-70. Koch, D. A., Chandler, M. J., Harder, D. W., & Paget, K. F. (1982). Parental defense style and child competence: A match-mismatch hypothesis. Journal of Applied Developmental Psychology, 3, 11-21. Kochanska, G. (1990). Maternal beliefs as long-term predictors of motherchild interaction and report. Child Development, 61, 1934-1943. Kochanska, G., Kuczynski, L., & Maguire, M. (1989). Impact of diagnosed depression and self-reported mood on mothers' control strategies: A longitudinal study. Journal of Abnormal Child Psychology, 17, 493-511. Kohn, M. L. (1979). The effects of social class on parental values and practices. In D. Reiss & H. Hoffman (Eds.), The American family: Dying or developing (pp. 45-68). New York: Plenum. Kowal, A., & Kramer, L. (1997), Children's understanding of parental differential treatment. Child Development, 68, 113-126. Krampen, G. (1989). Perceived childrearing practices and the development of locus of control in early adolescence. International Journal of Behavioral Development, 12, 177-193. Kuczynski, L. (1984). Socialization goals and mother-child interaction: Strategies for long-term and short-term compliance. Developmental Psychology, 20, 1061-1073. Kuczynski, L., & Lollis, S. (in press). Four foundations for a dynamic model of parenting. In J. R. M. Gerris (Ed.), Dynamics of parenting. Hillsdale, NJ: Erlbaum. Kuczynski, L., Marshall, S., & Schell, K. (1997). Value socialization in a bidirectional context. In J. E. Grusec & L. Kuczynski (Eds.), Parenting and children's internalization of values: A handbook of contemporary theory (pp. 23-52). New York: Wiley. Lafore, G. G. (1945). Practices of parents in dealing with preschool children. Child Development Monographs, 31. New York: Teacher's College, Columbia University. *Lamb, M. (1977). Father-infant and mother-infant interaction in the first year of life. Child Development, 48, 167-181. Lamb, M. E., & Easterbrooks, M. A. (1981). Individual differences in parental sensitivity: Origins, components, and consequences. In M. E. Lamb & L. R. Sherrod (Eds.), Infant social cognition: Empirical and theoretical considerations (pp. 127-153). Hillsdale, NJ: Erlbaum. Lerner, R. M. (1989). Developmental contextualism and the life-span view of person-context interaction. In M. H. Bornstein & J. S. Bruner (Eds.), Interaction in human development (pp. 217-239). Hillsdale, NJ: Erlbaum. Levy, D. M. (1931). Maternal over-protection and rejection. Journal of Nervous and Mental Diseases, 73, 65-77. Levy, D. M. (1943). Maternal overprotection. New York: Columbia University. Lewis, C. C. (1981). The effects of parental firm control: A reinterpretation of the findings. Psychological Bulletin, 90, 547-563. *Leyendecker, B., Lamb, M., Fracasso, M., Scholmerich, A., & Larson, C. (1997). Playful interaction and the antecedents of attachment: A longitudinal study of Central American and Euro-American mothers and infants. Merrill-Palmer Quarterly, 43, 24-47. *Leyendecker, B., Lamb, M., Scholmerich, A., & Fricke, D. M. (1997).

252

HOLDEN AND MILLER McLoyd, V. C. (1990). The impact of economic hardship on Black families and children: Psychological distress, parenting, and socioemotional development. Child Development, 61, 311-346. *McNally, S., Eisenberg, S., & Harris, J. D. (1991). Consistency and change in maternal child-rearing practices and values: A longitudinal study. Child Development, 62, 190-198. *Mekos, D., Hetherington, E. M., & Reiss, D. (1996). Sibling differences in problem behavior and parental treatment in nondivorced and remarried families. Child Development, 67, 2148-2165. Meldrum, B. (1982). Psychological factors in breast feeding versus bottle feeding in the Third World. Bulletin of the British Psychological Society, 35, 229-231. *Merrill, B. (1946). A measurement of mother-child interaction. Journal of Abnormal and Social Psychology, 9, 37-49. Metcalf, K., & Gaier, E. L. (1987). Patterns of middle-class parenting and adolescent underachievement. Adolescence, 22, 919-928. Miller, P. C., Shim, J., & Holden, G. W. (1999). Immediate contextual influences on maternal behaviour: Environmental affordances and demands. Journal of Environmental Psychology, 18, 1-12. Minuchin, P. (1985). Families and individual development: Provocations from the field of family therapy. Child Development, 56, 289-302. Mischel, W. (1977). On the future of personality measurement. American Psychologist, 32, 246-254. Mischel, W. (1979). On the interface of cognition and personality: Beyond the person-situation debate. American Psychologist, 34, 740-754. Mischel, W. (1984). Convergences and challenges in the search for consistency. American Psychologist, 39, 351-364. Mischel, W., & Shoda, Y. (1995). A cognitive-affective system theory of personality: Reconceptualizing situations, dispositions, dynamics, and invariance in personality structure. Psychological Review, 102, 246268. *Moore, G., Cohn, J., & Campbell, S. (1997). Mothers' affective behavior with infant siblings: Stability and change. Developmental Psychology, 33, 856-860. Morelli, G. A., Rogoff, B., Oppenheim, D., & Goldsmith, D. (1992). Cultural variation in infants' sleeping arrangements: Questions of independence. Developmental Psychology, 28, 604-613. *Moss, H. A. (1967). Sex, age, and state as determinants of mother-infant interaction. Merrill-Palmer Quarterly, 13, 19-36. Moulton, R. W., Burnstein, E., Liberty, P. G., & Altucher, N. (1966). Patterning of parental affection and disciplinary dominance as a determinant of guilt and sex typing. Journal of Personality and Social Psychology, 4, 356-363. *Murphey, D., & Alexander, S. (1991, March). Parents' beliefs and parental behavior: A multi-method study. Paper presented at the biennial meeting of the Society for Research in Child Development, Seattle, WA. Nunnally, J. C. (1978). Psychometric theory (2nd ed.). New York: McGraw-Hill. *O'Brien, M. & Nagle, K. J. (1987). Parents' speech to toddlers: The effect of play context. Journal of Child Language, 14, 269-279. *O'Dell, P. C. (1995, March). Maternal stability across contexts: An observational exploration of settings and tasks. Poster presented at the biennial meeting of the Society for Research in Child Development, Indianapolis, IN. *Olson, S., Bates, J., & Bayles, K. (1984). Mother-infant interaction and the development of individual differences in children's cognitive competence. Developmental Psychology, 20, 166-179. Paikoff, R. L., & Brooks-Gunn, J. (1991). Do parent-child relationships change during puberty? Psychological Bulletin, 110, 47-66. *Palmer, S. A. (1993). Mother-infant and father-infant interactions during caregiving and play and their effects on infant cognitive and emotional development. Unpublished doctoral dissertation, University of California, Riverside. *Park, S., Belsky, J., Putnam, S., & Crnic, K. (1997). Infant emotionality,

Contexts as moderators of observed interactions: A study of Costa Rican mothers and infants from differing socioeconomic backgrounds. International Journal of Behavioral Development, 21, 15-34. Light, R. J., & Pillemer, D. B. (1984). Summing up: The science of reviewing research. Cambridge, MA: Harvard University Press. Luster, R., & Okagaki, L. (Eds.) (1993). Parenting: An ecological perspective. Hillsdale, NJ: Erlbaum. Lytton, H., & Romney, D. M. (1991). Parents' differential socialization of boys and girls: A meta-analysis. Psychological Bulletin, 109, 267-296. *Lytton, H., Watts, D., & Dunn, B. E. (1986). Stability and predictability of cognitive and social characteristics from age 2 to age 9. Genetic, Social and General Psychology Monographs, 112, 363-398. Lytton, H., Watts, D., & Dunn, B. E. (1988). Continuity and change in child characteristics and maternal practices between ages 2 and 9: An analysis of interview responses. Child Study Journal, 18, 1-15. *Lytton, H., & Zwirner, W. (1975). Compliance and its controlling stimuli observed in a natural setting. Developmental Psychology, 11, 769-779. Maccoby, E. E. (1984). Socialization and developmental change. Child Development, 55, 317-328. Maccoby, E. E. (1992). The role of parents in the socialization of children: An historical overview. Developmental Psychology, 28, 1006-1017. Maccoby, E. E., & Martin, J. A. (1983). Socialization in the context of the family: Parent-child interaction. In P. H. Mussen (Series Ed.) & E. M. Hetherington (Vol. Ed.), Handbook of child psychology: Vol. 4. Socialization, personality, and social development (4th ed., pp. 1-101). New York: Wiley. Magnusson, D., & Stattin, H. (1998). Person-context interaction theories. In W. Damon (Series Ed.) & R. M. Lerner (Vol. Ed.), Handbook of child psychology: Vol. 1. Theoretical models of human development (5th ed., pp. 685-759). New York: Wiley. Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood, and adulthood: A move to the level of representation. In I. Bretherton & E. Waters (Eds.), Growing points of attachment theory and research. Monographs of the Society for Research in Child Development, 50 (209, Serial No. 1-2). Martin, B. (1975). Parent-child relations. In F. D. Horowitz (Ed.), Review of child development research (Vol. 4, pp. 463-540). Chicago: University of Chicago. Martin, J. A. (1981). A longitudinal study of the consequences of early mother-infant interaction: A microanalytic approach. Monographs of the Society for Research in Child Development, 190 (46, Serial No. 3). Mash, E. J., & Johnston, C. (1990). Determinants of parenting stress: Illustrations from families of hyperactive children and families of physically abused children. Journal of Consulting and Clinical Psychology, 19, 313-328. *Mash, E., Johnston, C., & Kovitz, K. (1983). A comparison of the mother-child interactions of physically abused and non-abused children during play and task situations. Journal of Clinical Child Psychology, 12, 337-346. McCall, R., Appelbaum, M. I., & Hogarty, P. S. (1973). Developmental changes in mental performance. Monographs of the Society for Research in Child Development, 38 (3, Serial No. 150). McCord, J. (1988). Parental aggressiveness and physical punishment in long-term perspective. In G. T. Hotaling, D. Finkelhor, J. T. Kirpatrick, & M. A. Straus (Eds.), Family abuse and its consequences: New directions in research (pp. 93-98). Beverly Hills, CA: Sage. McGillicuddy-DeLisi, A. V., & Sigel, I. E. (1995). Parental beliefs. In M. H. Bornstein (Ed.), Handbook of parenting: Vol. 3. Status and social conditions of parenting (pp. 333-358). Mahwah, NJ: Erlbaum. McHale, J. P. (1995). Coparenting and triadic interactions during infancy: The roles of marital distress and child gender. Developmental Psychology, 31, 985-996. *McHale, S. M., & Pawletko, T. M. (1992). Differential treatment of siblings in two family contexts. Child Development, 63, 68-81.

SIMILARITY IN PARENTING parenting, and 3-year inhibition: Exploring stability and lawful discontinuity in a male sample. Developmental Psychology, 33, 218-227. Parpal, M., & Maccoby, E. E. (1985). Maternal responsiveness and subsequent child compliance. Child Development, 56, 1326-1334. Patterson, G. R. (1982). Coercive family process. Eugene, OR: Castalia. *Pease, D., & Melby, J. N. (1988, March). Stability of maternal and paternal perceptions of parent behaviors. Paper presented at the biennial meeting of the Southwestern Society for Research in Human Development, New Orleans, LA. *Pettit, G. S., & Bates, J. E. (1984). Continuity of individual differences in the mother-infant relationship from six to thirteen months. Child Development, 55, 729-739. *Pettit, G., & Bates, J. (1989). Family interaction patterns and children's behavior problems from infancy to 4 years. Developmental Psychology, 25, 413-420. *Pianta, R. C., Sroufe, L. A., & Egeland, B. (1989). Continuity and discontinuity in maternal sensitivity at 6, 24, and 42 months in a high-risk sample. Child Development, 60, 481-487. Plomin, R. (1990). Nature and nurture: An introduction to human behavioral genetics. Pacific Grove, CA: Brooks/Cole. Plomin, R. (1994). Nature, nurture, and social development. Social Development, 3, 37-53. *Poresky, R. H., & Hendrix, C. (1989, March). Parenting priorities: Stability, change, and impact on young children. Paper presented at the biennial meeting of the Society for Research in Child Development, Kansas City, MO. *Pratt, M. N., Kerig, P., Cowan, P. A., & Cowan, C. P. (1988). Mothers and fathers teaching 3-year-olds: Authoritative parenting and adult scaffolding of young children's learning. Developmental Psychology, 24, 832-839. *Pridham, K., Van Riper, M., Schroeder, M., & Thoyre, S. (1997, April). Mothers' working models of feeding: How stable are they through the first year? Poster presented at the biennial meeting of the Society for Research in Child Development, Washington, DC. Pulkkinen, L. (1982). Self control and continuity from childhood to adolescence. P. B. Baltes & O. B. Brim (Eds.), Life span development and behavior (Vol. 4, pp. 63-105). New York: Academic Press. Radke-Yarrow, M. (1989). Development and contextual analysis of continuity. Human Development, 32, 204-209. Radke-Yarrow, M., Zahn-Waxler, C. Z., & Chapman, M. (1983). Children's prosocial dispositions and behavior. In P. H. Mussen (Series Ed.), & E. M. Hetherington (Ed.), Handbook of child psychology: Vol. 4. Socialization, personality, and social development (pp. 469-545). New York: Wiley & Sons. Ragozin, A. S., Basham, R. B., Crnic, K. A., Greenberg, M. T., & Robinson, N. M. (1982). Effects of maternal age on parenting role. Developmental Psychology, 18, 627-634. Raphael-Leff, L. (1986). Facilitators and regulators: Conscious and unconscious processes in pregnancy and early motherhood. British Journal of Medical Psychology, 59, 43-55. *Roberts, G. C., Block, J. H., & Block, J. (1984). Continuity and change in parents' child-rearing practices. Child Development, 55, 586-597. Roggman, L. A., Langlois, J. H., Hubbs-Tait, L., & Rieser-Danner, L. A. (1994). Infant day-care, attachment, and the "file drawer problem." Child Development, 65, 1429-1443. Rohner, R. P. (1986). The warmth dimension: Foundations of parental acceptance-rejection theory. Beverly Hills, CA: Sage. Rollins, B. C., & Thomas, D. L. (1979). Parental support, power, and control techniques in the socialization of children. In W. R. Burr, R. Hill, F. I. Nye, & I. L. Reiss (Eds.), Contemporary theories about the family: Vol. 1. Research-based theories (pp. 317-364). New York: Free Press. Rosenthal, R. (1984). Meta-analytic procedures for social research. Newbury Park, CA: Sage.

253

Rosenthal, R. (1991). Meta-analytic procedures for social research (Rev. ed.). Newbury Park, CA: Sage. *Rothbart, M. (1971). Birth order and mother-child interaction in an achievement situation. Journal of Personality and Social Psychology, 17, 113-120. *Rothbaum, F. (1988). Maternal acceptance and child functioning. MerrillPalmer Quarterly, 34, 163-184. Rothbaum, F., & Weisz, J. R. (1994). Parental caregiving and child externalizing behavior in nonclinical samples: A meta-analysis. Psychological Bulletin, 116, 55-74. Rutter, M. (1984). Continuities and discontinuities in socioeconomic development. In R. N. Emde & R. J. Harmon (Eds.), Continuities and discontinuities in development (pp. 41-68). New York: Plenum. Sameroff, A. J., & Feil, L. A. (1985). Parental concepts of development. In I. Sigel (Ed.), Parental belief systems: The psychological consequences for children (pp. 83-105). Hillsdale, NJ: Erlbaum. Scarr, S., & McCartney, K. (1983). How people make their own environments: A theory of genotype environment effects. Child Development, 54, 424-435. *Scarr, S., & McCartney, K. (1988). Far from home: An experimental evaluation of the mother-child home program in Bermuda. Child Development, 59, 531-543. Schaefer, E. S. (1959). A circumplex model for maternal behavior. Journal of Abnormal and Social Psychology, 59, 226-235. *Schaefer, E. S., & Bayley, N. (1960). Consistency of maternal behavior from infancy to preadolescence. Journal of Abnormal and Social Psychology, 61, 1-6. Schaffer, H. R., & Liddell, C. (1984). Adult-child interaction under dyadic and polyadic conditions. British Journal of Developmental Psychology, 2, 33-42. Schmidt, S. E., Liddle, H. A., & Dakof, G. A. (1996). Changes in parenting practices and adolescent drug abuse during multidimensional family therapy. Journal of Family Psychology, 10, 12-27. *Scholmerich, A., Lamb, M., Leyendecker, B., & Fracasso, M. P. (1997). Mother-infant teaching interactions and attachment security in EuroAmerican and Central-American immigrant families. Infant Behavior & Development, 20, 165-174. Sears, R. R., Maccoby, E. E., & Levin, H. (1957). Patterns of childrearing. Evanston, IL: Row, Peterson. *Sigel, I. E. (1982). The relationship between parental distancing strategies and the child's cognitive behavior. In L. M. Laosa & I. E. Sigel (Eds.), Families as learning environments for children (pp. 47-86). New York: Plenum. Sigel, I. E., McGillicuddy-DeLisi, A. V., & Goodnow, J. J. (Eds.). (1992). Parental belief systems: The psychological consequences for children (2nd ed.). Hillsdale, NJ: Erlbaum. *Sigman, M., Neumann, C., Carter, E., & Cattle, D. (1988). Home interactions and the development of Embu toddlers in Kenya. Child Development, 59, 1251-1261. Sigman, M., & Wachs, T. D. (1991). Structure, continuity, and nutritional correlates of caregiver behavior patterns in Kenya and Egypt. In M. H. Bornstein (Ed.), Cultural approaches to parenting (pp. 123-137). Hillsdale, NJ: Erlbaum. Smetana, J. G. (1989). Toddlers' social interactions in the context of moral and conventional transgressions in the home context. Developmental Psychology, 25, 499-508. Smetana, J. G. (1997). Parenting reconceptualized: A social domain analysis. In J. E. Grusec & L. Kuczynski (Eds.), Parenting and children's intemalization of values: A handbook of contemporary theory (pp. 162-192). New York: Wiley. Spitz, R. A. (1965). The first year of life: A psychoanalytic study of normal and deviant patterns of object relations. New York: International Universities. Sroufe, L. A., & Jacobvitz, D. (1989). Diverging pathways, developmental

254

HOLDEN AND MILLER Valsiner, J. (1989). Human development and culture: The social nature of personality and its study. Lexington, MA: Heath. van den Boom, D. C., & Hoeksma, J. B. (1994). The effect of infant irritability on mother-infant interaction: A growth-curve analysis. Developmental Psychology, 30, 581-590. Vaughn, B., Egeland, B., Sroufe, L. A., & Waters, E. (1979). Individual differences in infant-mother attachment at twelve and eighteen months: Stability and change in families under stress. Child Development, 50, 971-975. *Vuchinich, S., Angelelli, J., & Gatherum, A. (1996). Context and development in family problem solving with preadolescent children. Child Development, 67, 1276-1288. *Vuchinich, S., Bank, L., & Patterson, G. (1992). Parenting, peers, and the stability of antisocial behavior in preadolescent boys. Developmental Psychology, 28, 510-521. *Wachs, T. D. (1987). Short-term stability of aggregated and nonaggregated measures of parental behavior. Child Development, 58, 796-797. Wahler, R. G., & Dumas, J. E. (1989). Attentional problems in dysfunctional mother-child interactions: An interbehavioral model. Psychological Bulletin, 105, 116-130. *Ward, M. J., Vaughn, B. E., & Robb, M. D. (1988). Social-emotional adaptation and infant-mother attachment in siblings: Role of the mother in cross-sibling consistency. Child Development, 59, 643-651. Waters, E. (1978). The reliability and stability of individual differences in infant-mother attachment. Child Development, 49, 483-494. Watson, G. (1934). A comparison of the effects of lax versus strict home training. Journal of Social Psychology, 5, 102-105. Webster-Stratton, C. (1990). Stress: A potential disrupter of parent perceptions and family interactions. Journal of Child Clinical Psychology, 19, 302-312. Whiting, J. W. M., & Child, I. L. (1953). Child training and personality: A cross-cultural study. New Haven, CT: Yale University. Wierson, M., & Forehand, R. (1994). Parent behavioral training for child noncompliance: Rationale, concepts, and effectiveness. Current Directions in Psychological Science, 3, 146-150. Zahn-Waxier, C., & Chapman, M. (1982). Immediate antecedents of caretakers' methods of discipline. Child Psychiatry and Human Development, 12, 179-192. *Zegiob, L., Arnold, S., & Forehand, R. (1975). An examination of observer effects in parent-child interactions. Child Development, 46, 509-512. *Ziv, Y., Gini, M., Guttmann, S., & Sagi, A. (1997, April). Dyadic emotional availability and quality of infant-mother attachment: A three point longitudinal study. Paper presented at the biennial meeting of the Society for Research in Child Development, Washington, DC. *Zussman, J. S. U. (1980). Situational determinants of parental behavior: Effects of competing cognitive activity. Child Development, 51, 792800.

transformations, multiple etiologies and the problem of continuity in development. Human Development, 32, 196-203. Sroufe, L. A., Jacobvitz, D., Mangelsdorf, S., DeAngelo, E., & Ward, M. J. (1985). Generational boundary dissolution between mothers and their preschool children: A relationship systems approach. Child Development, 56, 317-325. *Stattin, H., & Klackenberg, G. (1992). Discordant family relations in intact families: Developmental tendencies over 18 years. Journal of Marriage and the Family, 53, 940-956. Steinberg, L. D. (1981). Transformations in family relations at puberty. Developmental Psychology, 17, 833-840. Steinberg, L., Lamborn, S. D., Dornbusch, S. M., & Darling, N. (1992). Impact of parenting practices on adolescent achievement: Authoritative parenting, school involvement, and encouragement to succeed. Child Development, 63, 1266-1281. Stolz, L. M. (1967). Influences on parent behavior. Stanford: Stanford University Press. Symonds, P. (1938). A study of parental acceptance and rejection. American Journal of Orthopsychiatry, 8, 679-688. *Tamis-LeMonda, C. S., & Bornstein, M. H. (1989). Habituation and maternal encouragement of attention in infancy as predictors of toddler language, play, and representational competence. Child Development, 60, 738-751. *Tamis-LeMonda, C. S., & Bornstein, M. H. (1991). Individual variation, correspondence, stability, and change in mother and toddler play. Infant Behavior and Development, 14, 143-162. *Tamis-LeMonda, C. S., Chen, L. A., & Bornstein, M. H. (1998). Mothers' knowledge about children's play and language development: Short-term stability and interrelations. Developmental Psychology, 34, 115-124. Taylor, M., & Kogan, K. L. (1973). Effects of birth of a sibling on mother-child interactions. Child Psychiatry & Human Development, 4, 53-58. *Teti, D. M., Sakin, J. W., Kucera, E., Corns, K. M., & Eiden, R. D. (1996). And baby makes four: Predictors of attachment security among preschool-age firstborns during the transition to siblinghood. Child Development, 67, 579-596. Thomas, A., & Chess, S. (1977). Temperament and development. New York: Brunner-Mazel. Thompson, E., & Miller, P. C. (1997, April). Parental beliefs and use of parental discipline: The role of religious affiliation. Poster presented at the biennial meeting of the Society for Research in Child Development, Washington, DC. Thompson, R. A. (1998). Early sociopersonality development. In W. Damon (Series Ed.) & N. Eisenberg (Vol. Ed.), Handbook of child psychology: Vol. 3. Social, emotional, and personality development (5th ed., pp. 25-104). New York: Wiley. Thompson, R. A., Lamb, M. E., & Estes, D. (1982). Stability in infantmother attachment and its relationship to changing life circumstances in an unselected middle-class sample. Child Development, 53, 144-148. Trickett, P. K., & Kuczynski, L. (1986). Children's misbehavior and parental discipline in abusive and non-abusive families. Developmental Psychology, 22, 166-176. Valsiner, J. (1984). Construction of the zone or proximal development in adult-child joint action: The socialization of meals. In B, Rogoff & J. Wertsch (Eds.), Children's learning in the zone of proximal development (pp. 65-76). San Francisco: Jossey-Bass.

Received June 30, 1997 Revision received August 4, 1998 Accepted August 19, 1998

Das könnte Ihnen auch gefallen