Sie sind auf Seite 1von 9

Journal tifCenmiology: PSYCHOLOGICAL SCIENCES In the Public Do

19X9. Vol. 44. No. 6. PI6I-I69

Age Differences and Changes in the


Use of Coping Mechanisms
Robert R. McCrae

Gerontology Research Center, National Institute on Aging.

Two cross-sectional studies of coping conducted on 405 men and women aged 21 to 91 in 1980 (McCrae, 1982)
formed the basis for longitudinal analyses in 1987. At Time 2,191 of the original subjects were retested, and data were

Downloaded from http://geronj.oxfordjournals.org/ at Penn State University (Paterno Lib) on February 20, 2016
also gathered from a new sample of 207 men and women. Subjects reported on a single event classified as a loss,
threat, or challenge; responses were scored for 28 specific coping mechanisms and 2 broad coping factors. Cross-
sectional analyses showed evidence of age or cohort differences in the use of several coping mechanisms, but none of
these effects was consistently paralleled by changes in repeated measures and cross-sequential analyses. These
findings suggest that aging has little effect on coping behavior in a community-dwelling sample. Retest correlations
demonstrated modest stability of individual differences in most coping mechanisms, suggesting that coping responses
are in part a function of enduring characteristics of the individual.

A LTHOUGH stress has been a familiar topic in social


science research for decades, the systematic study of
sectional methods that confound generational differences
with maturational changes. This article reports a 7-year
coping processes is relatively new (Costa & McCrae, 1989). longitudinal study of coping in adults initially aged 21 to 91
Studies documenting the harmful effects of stress have (McCrae, 1982), and thus offers the opportunity to examine
generally reported only modest associations (Schroeder & age changes over a reasonably long interval as well as age
Costa, 1984), and even major stressors such as bereavement differences.
tend to have relatively short-term effects (McCrae & Costa, Two parallel studies were conducted in 1980. In the first,
1988). Clearly, most people find ways to deal with stress, 255 men and women completed a 118-item Coping Ques-
and researchers in the past decade have begun to study these tionnaire (CQ) concerning the ways in which they had dealt
ways of coping. with a single event they had previously reported on a check-
Gerontologists have a natural interest in the ways in which list of life events. The event was selected and classified by
aging individuals cope with the characteristic stressors of the investigator as a loss, threat, or challenge. In the second
each phase of the life span. Do older individuals bring study, 150 men and women completed a shortened, 50-item
wisdom and experience to their coping efforts, or are the version of the CQ for each of three events. These events
stressors of old age compounded by a declining ability to were selected by the subject as instances of loss, threat, and
cope? Maturational theories of coping and defense (e.g., challenge. Age differences were examined for 28 specific
Pfeiffer, 1977; Vaillant, 1977) received little support from coping mechanisms in the first study. After controlling for
early cross-sectional studies of coping, which showed few the type of stressor, eight mechanisms showed significant
age differences in the use of coping mechanisms (Lazarus & effects; only two of these, however, were replicated in the
DeLongis, 1983; McCrae, 1982). Such findings suggested second study. In both studies, middle-aged and older sub-
that, like personality (Costa & McCrae, 1988), coping styles jects were less likely than younger adults to use hostile
and abilities were stable in adulthood. reactions and escapist fantasy in dealing with stress.
More recently, however, cross-sectional age differences In 1987 the full CQ was administered to 398 subjects aged
have been reported. Folkman, Lazarus, Pimley, & Novacek 20 to 93. Cross-sectional analyses of these data provide one
(1987) found that older respondents were less likely to seek way to estimate age effects, but aging is confounded with
social support or use confrontive coping; they were more generational differences and sampling biases. About half of
likely to use distancing and positive reappraisal. Irion and the subjects tested in 1987 were from the initial 1980 sample;
Blanchard-Fields (1987) replicated the association of age on these subjects, repeated measures analyses could be
with the decreased use of confrontive coping, but only when performed to assess changes over time. These longitudinal
dealing with stressors that could be classified as threats. analyses control for cohort differences, but may be suscepti-
Cross-sectional differences in styles of defense have also ble to practice and time of measurement effects. Because the
been noted: Costa, Zonderman, and McCrae (in press) CQ had not been previously administered to the other half of
suggested that older men and women were "more forgiving the 1987 sample, it was also possible to conduct cross- and
and willing to meet adversity cheerfully, and less prone to time-sequential analyses (Schaie, 1977). Cross-sequential
take offense or to vent frustrations on others." analyses compare groups of people born at the same time but
With the exception of Vaillant's (1977) research on the tested at different times for example, one subsample of
maturity of coping mechanisms between late adolescence people born in 1900 might be tested in 1960, at age 60; a
and middle adulthood, all these studies have relied on cross- second subsample might be tested in 1980, at age 80.

P161
P162 MCCRAE

Differences between these two groups could not be due to example, Felton and Revenson (1984) reported 7-month
birth cohort effects (because subjects were born at the same retest reliability coefficients of .58 and .63 for measures of
time) or to practice effects, and may indicate maturational information seeking and wish-fulfilling fantasy, respec-
change. Time-sequential analyses, which compare individ- tively, as mechanisms for coping with chronic illnesses.
uals tested at the same ages but at different times, allow an Solomon, Mikulincer, and Avitzur (1988) reported one-year
estimation of time of measurement effects. Taken together, retest coefficients for four coping mechanisms ranging from
these analyses allow reasonable inferences about which .38 to .58 in a sample of Israeli soldiers who had experienced
effects may be attributed to maturation. a combat stress reaction. The present study will examine the
It should be noted that the topic of coping presents both 7-year stability of individual differences in the use of a
conceptual and methodological challenges to longitudinal variety of coping mechanisms.
research. Repeated measures and cross-sequential designs One of the unresolved issues in contemporary coping
normally examine changes in the same variable over time. research is the number, nature, and breadth of coping mech-
Coping behavior, however, is by definition situation- anisms. Although it is universally recognized that there are

Downloaded from http://geronj.oxfordjournals.org/ at Penn State University (Paterno Lib) on February 20, 2016
specific. An individual might have been asked at Time 1 how many different ways to cope, researchers disagree on how
he or she coped with a recent bereavement. The same these specific behaviors can be meaningfully combined into
question seven years later is not really the same question: useful scales. Should we examine a few broad categories like
Most individuals will long since have adapted to the stress of problem-focused and emotion-focused coping (Folkman &
widowhood (McCrae & Costa, 1988). It is unlikely that Lazarus, 1980), several mid-level mechanisms such as con-
many individuals would be found who recently faced the frontive coping and distancing (Folkman, Lazarus, Dunkel-
same stressor that they had reported on seven years earlier. Schetter, DeLongis, & Gruen, 1986), or many fine-grained
Operationally, this entails a change in the design for nomi- mechanisms (McCrae, 1982)? In the present study, mecha-
nating stressors; conceptually, it means that the measures are nisms at two levels are examined. In addition to the 28
"repeated" only approximately. specific mechanisms examined in the original study, two
This does not, however, invalidate the logic of longitudi- broad factors Neurotic Coping and Mature Coping
nal research. Both cross-sectional and longitudinal studies of derived from analysis of these 28 mechanisms (McCrae &
age and coping examine one or a few instances of coping Costa, 1986) are also considered. (Analyses were also per-
behavior as a sample of general coping styles. If there were formed on the Folkman et al., 1986, scales; the results were
no consistent individual differences in coping if all coping similar to those found for the other coping scales and so have
behavior were determined entirely by the specifics of the been omitted.)
situation then it would make no sense to speak of age
differences or changes in coping. Any such differences METHOD
would be the result of age differences in the stressors Because two related studies were carried out in 1980, a
encountered. If, however, maturation has broad effects on number of different longitudinal and sequential comparisons
the ways individuals deal with stress, then these should are possible. An overview of the design is presented in Table
appear even when different stressors are sampled at different 1, which summarizes some characteristics of the subsamples
times. used in the present study. The subsamples are identified by
Although the same stressor cannot be examined at both letters. Note that A' is the subset of A that returned complete
times, it is possible to introduce some control by requiring data in 1987; similarly, B' is the subset of B who completed
that the same type of stressor be examined. For this purpose, the full CQ in 1987. The short CQ can be scored from the full
stressful events have been classified as losses, threats, or version to permit longitudinal analyses of B' data. Subsam-
challenges. This classification, suggested by Lazarus and
Launier (1978), was used in the original cross-sectional
study and has proven useful in understanding situational Table 1. Characteristics of Subsamples and Data
determinants of coping (McCrae, 1984). Other dimensions for Two Administrations
of the stressor, including severity, chronicity, and controlla-
bility, may also affect coping responses. Although they were Time 1 (1980)
not controlled in the present design, these features, as rated Subsample B
by a judge (McCrae, in press), were weakly related to age 255 150
(I r | s = .08 to . 19), and so are unlikely to affect the results % Male 60.4 53.3
of age analyses. Age range 24-91 21-90
Longitudinal studies offer a direct approach to the ques- Instrument Full CQ Short CQ
tion of individual consistency versus situational specificity Time 2 (1987)
in coping. Correlations of coping measures across time can Subsample A' B' C
suggest the degree to which coping behavior is a stable N 113 78 207
characteristic of the individual. A few studies have exam- % Male 56.6 57.7 50.2
ined stability in broad styles of defense (Haan, 1977; Vail- Age range 34-89 33-89 20-93
lant, Bond, & Vaillant, 1986) and reported modest but Instrument Full CQ Full CQ Full CQ
significant associations over 10-year intervals. Additional
Note: CQ = Coping Questionnaire. The Full CQ contains the 68 items of
studies have examined the stability of specific coping mech- the Folkman and Lazarus (1980) Ways of Coping plus 50 additional items;
anisms, but over considerably shorter periods of time. For the Short CQ contains only the 50 additional items.
LONGITUDINAL ANALYSES OF COPING P163

pie C is composed of subjects first tested in 1987. Individual classified events and the follow-up employed self-selected
analyses will be described in terms of this Table. events, the validity of the classification at Time 2 must be
assessed. A judge, blind to the type of questionnaire as-
signed, read each of the narrative paragraphs describing the
Subjects. All participants were members of the Aug- stressful event and coded it as a loss, threat, or challenge.
mented Baltimore Longitudinal Study of Aging (ABLSA). Overall agreement between assigned and judged type of
The BLSA sample itself is composed of generally healthy, stressor was 47%, showing less-than-perfect agreement, but
well-educated, community-dwelling volunteers (Shock et well over the 33% to be expected by chance, ^ 2 (4, N =
al., 1984); the ABLSA consists of a subsample of BLSA 378) = 64.10, p < .001. When examined by categories,
participants, augmented by the addition of their spouses, most discrepancies concerned confusions between threats
who have agreed to complete a variety of psychological and losses. This is understandable, given the association
questionnaires at home. Recruitment into the BLSA is con- between these two categories: Threats are often potential
tinuous, so that new participants joined the study after the losses, and some losses (e.g., loss of a job) pose threats to

Downloaded from http://geronj.oxfordjournals.org/ at Penn State University (Paterno Lib) on February 20, 2016
initial data collection in 1980. future well-being. Further, most of the situational effects on
In 1980, 255 ABLSA participants (Subsample A) com- the choice of coping mechanisms contrast challenges with
pleted the CQ in the first of two parallel studies. Their threats and losses (McCrae, 1984). When the threat and loss
responses provide Time 1 data for repeated measures and categories were combined, agreement between rater and
cross-sequential analyses on the full CQ. An additional 150 assigned condition increased to 74%. There was, then,
participants (Subsample B) completed an abbreviated ver- substantial agreement on the most important distinction
sion of the CQ with regard to three self-selected stressors; among types of stressor.
their data are used in supplementary repeated measures and
cross-sequential analyses. Subsamples A' and B' also pro-
vide data for correlational analyses of the stability of individ- Measures. The Coping Questionnaire (CQ) consists of
ual differences in coping. In 1987, responses to the full 118 items. The first 68 items were taken from the Ways of
Coping Questionnaire were obtained from 207 ABLSA Coping (Folkman & Lazarus, 1980); 50 additional items
participants (Subsample C) who had not previously com- were written to cover a wide range of coping mechanisms
pleted either version of the CQ. The age ranges and sex identified in a review of the literature (McCrae, 1982). The
ratios of all these groups are given in Table 1. abbreviated version of the CQ consisted only of the 50 new
items and differed also in some details: Subjects were of-
Procedure. In 1979, ABLSA participants completed a fered a "does not apply" category and were also asked to
life events checklist. The 255 subjects in Subsample A were rate the perceived effectiveness of the coping strategy de-
selected because they indicated that they had experienced a scribed by the item.
life event that could be categorized as a loss, threat, or Responses to the full CQ were scored for use of each of 28
challenge within the past year. These subjects were sent the specific coping mechanisms; these are listed in Table 2. The
118-item CQ in 1980 and asked to complete it with regard to number of items in each mechanism ranged from 1 to 7; for
the selected event. The 150 subjects in Subsample B had not scales with two or more items, internal consistency in the
experienced any of the designated life events within the past initial study ranged from .35 to .83, with a median of .67
year. They were therefore asked to nominate three events (McCrae, 1984), suggesting moderate reliability, t h e scales
a loss, a threat, and a challenge which had occurred in the have shown validity and utility in studies of environmental
past six months, and to complete a 50-item version of the CQ and dispositional determinants of coping (Costa & McCrae,
with regard to each. Further details are given in McCrae 1989; McCrae, 1984; McCrae & Costa, 1986). Twenty-
(1982)." seven of the specific coping mechanisms were also scored
In 1987, participants in Subsamples A', B', and C were from the short version of the CQ, although the scales were
mailed one of three questionnaires which asked them to necessarily shorter. Table 2 gives the number of items in
recall a stressful event they had experienced in the past six both versions of each of the scales. Scores were computed
months. The first asked about harm or loss, and instructed separately for each of the three types of stressor for each
subjects to recall an event "in which you were hurt, disap- subject in Subsample B.
pointed, or lost something of value to you"; the second Parallel factor analyses of mechanisms from the 1980
asked about threat ("some kind of threat or clanger, in which administrations of the long and short versions of the CQ
you were worried about how things would turn out"); the showed two well-replicated factors (McCrae & Costa, 1986).
third asked about challenge ("some challenge or big oppor- Neurotic coping was defined by loadings on hostile reaction,
tunity"'). All subjects were asked to describe the incident in a escapist fantasy, self-blame, sedation, withdrawal, passiv-
paragraph, rate its stressfulness, and then decide whether ity, and indecisiveness; each of these specific mechanisms
they had ever used each of the 118 coping responses in was found to be associated with the personality dimension of
dealing with the stressor. As in the original study, subjects Neuroticism. Mature coping was defined by rational action,
were told to leave blank inapplicable items. Subjects in perseverance, positive thinking, restraint, and self-adapta-
Subsample A were assigned the same type of stressor at tion. These mechanisms were summed to estimate the factors
Time 2 as they had reported on at Time 1; other individuals in the present study; coefficient alphas for the composites
were randomly assigned to one of the three types. were .74 for both factors. (The other coping mechanisms
Because the original study involved investigator- loaded on different factors in the two different analyses.)
PI 64 McCRAE

Table 2. Correlations and Stability Coefficients for Coping Mechanisms


Time 2 Correlations Stability Coefficients
Coping mechanism Gender Education Age Subsample A' Subsample B'
Individual mechanisms
Hostile reaction (7,6) 1 Q** *
-.09 _ |7*** 22* 37***
Rational action (7,4) -.01 .03 -.12* .28** .20
Seeking help (5,2) -.01 .07 -.07 .20* -.10
Perseverence (3.1) .08 .04 .02 .09 .18
Isolation of affect (2.1) -.06 -.05 .10* .16 -.07
Fatalism (6,2) .03 -.06 .12* 32*** .13
Expression of feelings (5,1) .15** -.06 -.25*** 39*** 43***
Positive thinking (7,3) -.01 -.04 _ 22*** .21* .25*
Distraction (2,1) .23*** -.06 -.04 .18 .20

Downloaded from http://geronj.oxfordjournals.org/ at Penn State University (Paterno Lib) on February 20, 2016
Escapist fantasy (5,2) .09 -.05 _ 23*** 43*** .24*
Intellectual denial (4,3) -.02 -.13* .05 .22* 29**
Self-blame (4,1) -.02 -.02 -.08 .22* .15
Taking one step at a time (2, ) .00 -.06 .11* .18
Social comparison (1,1) .08 .04 .01 .18 .31**
Sedation (3,1) 29*** -.12* -.14** .26** .60***
Substitution (1,1) .12* .01 -.08 27** .25*
Restraint (4,2) -.05 .00 - .12* .24* -.01
Drawing strength from adversity (5,1) -.03 -.01 -.02 47*** .07
Avoidance (1,1) .07 -.10 .05 .12 -.07
Withdrawal (2,1) .07 -.07 -.10* 33*** 55***
Self-adaptation (5,3) -.04 -.13* 39*** .29*
Wishful thinking (4,1) 2| *** -.13* -.05 .26** 41 ***
Active forgetting (3,1) .00 -.17** .09 .05 .26*
Humor (3.1) .03 -.05 -.16** 29** .03
Passivity (3,2) .12* -.08 .03 .45*** 34**
Indecisiveness (2,1) .16** -.09 .00 27** .28*
Assessing blame (1,1) -.01 -.12* .00 .16 .12
Faith (1,1) -.02 -.06 14** 4Q*** .27*
Coping Factors
Neurotic coping (26,14) Ig*** -.11* 15** 51 *** .55***
Mature coping (26,13) .06 .00 -.16** .38*** 34**

Note: The number of items in each scale for the full and short version of the Coping Questionnaire are given in parentheses. /Vs = 398 for gender and age,
315 for education, 113 for Subsample A', and 78 for Subsample B'. Stability coefficients for Subsample. B' are based on the short version of the Coping
Questionnaire; the coping mechanism Taking one step at a time was not measured in the short version. Gender is coded I for males, two for females.
*p < .05; **p < .01; ***/; < .001.

RESULTS Costa & McCrae, 1988, for details on personality measure-


Maturational effects are best inferred from converging ment). The respondents appear to be a fairly representative
evidence from several different designs (Costa & McCrae, subgroup of ABLSA participants.
1982). Cross-sectional, longitudinal, and sequential analy- Because coping is determined in part by type of stressor,
ses are considered separately for the long and short versions and because types of stressor vary with age, some control for
of the CQ, and results are then compared across the designs. type of stressor is needed in cross-sectional analyses. Two
None of the coping mechanisms shows a consistent pattern dummy variables representing the three types of stressors
of maturational effects, suggesting that ways of coping do were therefore partialed from each of the coping mechanism
not change predictably with age. scores to provide event-adjusted measures. Table 2 gives
correlations of these adjusted scores with gender, education,
Cross-sectional analyses. Cross-sectional age relations and age. (A comparison of unadjusted scores shows a very
are examined by correlating age with scores on the coping similar pattern of results.)
mechanisms at Time 2 for Subsamples A', B', and C. These Women appear more likely to use hostile reaction, expres-
398 subjects represented 57% of the 704 participants active sion of feelings, distraction, sedation, self-adaptation, wish-
in the ABLSA in 1987. The subjects who responded with ful thinking, passivity, and indecisiveness in dealing with
complete data were slightly better educated and higher in the stress; overall, they are somewhat higher on the use of
personality disposition of Conscientiousness than were non- neurotic coping mechanisms. These findings are consistent
respondents, but the two groups did not differ in age, with previous evidence that women are higher in the person-
gender, or the personality dispositions of Neuroticism, Ex- ality disposition of Neuroticism, which is a determinant of
traversion, Openness to Experience, or Agreeableness (see neurotic coping (McCrae & Costa, 1986). Education is not
LONGITUDINAL ANALYSES OF COPING P165

strongly related to the use of coping mechanisms, although it both Time 1 and Time 2, and repeated measures analyses
appears that better educated individuals are somewhat less were conducted for these subjects on all coping mechanisms.
likely to use intellectual denial, sedation, wishful thinking, Because of attrition from the larger study and nonresponse
active forgetting, and assessment of blame. at Time 2, only 44% of Subsample A was retested. The
Half of the 28 specific coping mechanisms are related to subjects in A' were four years younger and had had one year
age at Time 2. Only four of these findings, however, repli- more of education than the dropouts and nonrespondents.
cate the cross-sectional results from Time I analyses of They did not differ from them, however, in gender, in the
Subsample A (McCrae, 1982). In both studies, older men personality dispositions of Neuroticism, Extraversion,
and women were less likely to use hostile reaction, escapist Openness, Agreeableness, or Conscientiousness, or in the
fantasy, and sedation, and more likely to use faith. In the use of mature or neurotic coping mechanisms at Time 1. It
present study there are relatively strong negative correlations appears unlikely that results of the repeated measures analy-
between age and the use of expression of feelings and ses are seriously affected by attrition.
positive thinking, but neither of these effects was seen at Because each individual had been assigned the same type

Downloaded from http://geronj.oxfordjournals.org/ at Penn State University (Paterno Lib) on February 20, 2016
Time I. Overall, older individuals are slightly less likely to of stressor at Time 2 as he or she had reported on at Time 1,
use neurotic coping mechanisms, but also less likely to use this variable was controlled. By measuring the same individ-
mature coping mechanisms. The direction and significance ual twice, repeated measures designs also control for birth
of cross-sectional age effects are summarized in the first cohort, sex, education, and other variables that are constant
column of Table 3. for the individual. Subjects were cross-classified by gender
and type of stressor. Effects of gender generally paralleled
Repeated measures analyses. For the 113 subjects in those shown in Table 2, in part because the two samples
Subsample A', complete data on the CQ were available at overlap at Time 2. Main effects for type of stressor were

Table 3. Summary of Effects for Age-Related Analyses

Repeated Measures Cross-Sequential


Coping mechanism Cross-sectional Subsample A' Subsample B' Full CQ Short CQ
Specific mechanisms
Hostile reaction _ *** 0 0 0 0
Rational action _* 0 0 0 0
Seeking help 0 0 _* 0 0
Perseverance 0 0 0 0 0
Isolation of affect +* _* 0 _* _ #**
Fatalism +* 0 +* _* -j. + * *

Expression of feelings _ *** 0 +* 0 0


Positive thinking 0 0 _ ** 0
Distraction 0 0 0 0 +*
Escapist fantasy 0 0 0 0"
Intellectual denial 0 _** 0 0 0
Self-blame 0 0 0 0 0
Taking one step at a time +* 0 0
_j_ * *
Social comparison 0 0 0 0
Sedation _ ** 0 0 0 0
Substitution 0 0 0 0 0
Restraint _* 0 0 0 +*
Drawing strength from adversity 0 _ ** 0 0
Avoidance 0 0 0 0 0
Withdrawal _* 0 0 0 0
Self-adaptation _* _* _* 0 0
Wishful thinking 0 0 +*
Active forgetting 0 _ ** 0 0 0
Humor _ ** 0 0 0 0
Passivity 0 0 0 0 +*
Indecisiveness 0 0 0 0 0
Assessing blame 0 0 0 0 0
Faith + ** _ *** 0 _ ** 0
Coping Factors
Neurotic coping _ ** 0 _* 0 0
Mature coping _ ** 0 0 0 0

Note: Minus signs indicate a decrease in the variable with age or time and plus signs indicate an increase; zeros indicate no significant effect. The coping
strategy Taking one step at a time was not measured in the Subsample B' analyses.
*/J < .05; **/> < .01: ***/; < .001.
P166 McCRAE

consistent with previous research on the situational determi- age in time-sequential designs, these differences could not
nants of coping responses (McCrae, 1984). be due to age. Administration, rather than aging, seems to be
The second column of Table 3 indicates the direction and responsible for these effects.
significance of main effects for time. Between 1980 and In the present design, the second administration differed
1987, the use of isolation of affect, intellectual denial, from the first not only in time, but also in the procedure for
drawing strength from adversity, self-adaptation, active for- determining the stressor. Stressful events were selected by
getting, and faith declined; the use of wishful thinking the investigator at Time 1; they were nominated by the
increased. The largest effects were for wishful thinking and subject at Time 2. Although an attempt was made to control
faith, but even here time accounted for less than 2% of the for type of stressor, subjectively selected events may differ
variance in coping scores. in many ways from investigator-assigned stressors.
In general, results of the longitudinal analysis do not
square with those from the cross-sectional analysis. A com- Supplementary longitudinal analyses. In the analyses
parison of the first and second columns in Table 3 shows of the full CQ, historical time period changes associated

Downloaded from http://geronj.oxfordjournals.org/ at Penn State University (Paterno Lib) on February 20, 2016
only a single replication of age differences by age changes with social events between 1980 and 1987 is confounded
for self-adaptation. In the cases of isolation of affect and with procedural changes in the way the stressor was selected.
faith there are direct contradictions: Cross-sectional results It is possible to separate these two effects by employing
point to an increase in the use of these coping mechanisms, coping mechanism scores derived from the short version of
whereas longitudinal results point to a decrease. In other the CQ. Data from Subsamples B and B' can be used in
cases, significant effects in one design are not significant in repeated measures analyses; data from Subsample B and
the other. These discrepancies suggest that cohort differ- Subsamples A' and C can be used in cross-sequential analy-
ences may be responsible for the cross-sectional correla- ses. In these analyses, the procedure for selecting stressors is
tions, or that practice or time of measurement effects or the same, so any effects ought to be the result of time of
procedural differences between the two administrations measurement or maturation.
may be responsible for the repeated measures effects. For the repeated measures analyses, the responses of the
78 subjects in Subsample B' were examined at Times 1 and
Sequential analyses. An alternative way to estimate 2. At Time 1, subjects had responded to three separate
age changes is provided by the cross-sequential design, in events; for this analysis, only data from the same type of
which two different samples, matched on date of birth, are event as that assigned at Time 2 were considered. Thus, 31
compared at two different times. In the present study, Time individuals were compared on responses to self-selected
1 data from Subsample A were compared with Time 2 data losses at two times; 28 individuals were compared on self-
from Subsamples B' and C. Subjects were cross-classified selected threats; and 19 were compared on self-selected
by type of event to control for its effects, and were also challenges. Subjects were cross-classified by gender and
classified by gender and seven-year birth cohorts composed type of event.
of individuals born between 1907 and 1956. Data from 451 Main effects for time of administration were seen for
subjects were included in this range of birth years. Multivari- seeking help, self-adaptation, and neurotic coping, which
ate analyses of variance (ANOVAs) were conducted for the decreased, and for fatalism and expression of feelings,
28 specific coping mechanisms. Significant multivariate which both increased between 1980 and 1987. These effects
effects were found for gender, event type, birth cohort, and are shown in column 3 of Table 3. All the effects were small
administration main effects. in magnitude, and only one the decline in use of self-
The fourth column of Table 3 summarizes the effects of adaptation replicated the findings of repeated measures
administration of the CQ in 1980 versus 1987. Over this analyses conducted on the full CQ scales. The increased use
interval there were declines in the use of isolation of affect, of fatalism and decreased use of self-adaptation and neurotic
fatalism, positive thinking, drawing strength from adversity, coping are consistent in direction with the cross-sectional
and faith; there was an increase in the use of wishful findings, but the increased use of expression of feelings is
thinking. In general, this pattern of results resembles that inconsistent with the significant negative cross-sectional
found in the repeated measures analyses, but not that found correlation.
in cross-sectional analyses. It is also possible to conduct cross-sequential analyses on
The effects seen in repeated measures and cross-sequen- scales from the short version of the CQ. Subjects in Subsam-
tial analyses might represent true maturational effects that ple B provided data on all three types of events in 1980;
failed to appear in cross-sectional analyses because they subjects in Subsamples A' and C provided data on one event
were somehow masked by cohort differences. Alternatively, in 1987. When cross-classified by birth cohort (with dates of
they might be time of measurement effects that reflect birth between 1907 and 1956), comparisons of these two
differences between the 1980 and 1987 administrations. By groups can be used to estimate age changes.
comparing individuals matched for age at time of testing but Note that this analysis differs from traditional cross-
tested on two different occasions, time-sequential analyses sequential analyses in that the same individuals provide three
can be used to test this hypothesis. When 495 subjects, aged sets of measures at Time 1, whereas different individuals
26 to 81 at time of testing, were compared on the use of provide measures for each of the three types of stressor at
coping mechanisms in a time-sequential analysis, 7 of the 8 Time 2. Administration X Event Type interactions would be
significant administration effects seen in the cross-sequential difficult to interpret in this design, but the main effects for
analyses were replicated. Because subjects are matched on administration, which collapse data across type of event, are
LONGITUDINAL ANALYSES OF COPING P167

probably meaningful and preferable to separate analyses sistency of coping across both time and situations and give a
within type of event; these would multiply opportunities for conservative estimate of stability.
capitalizing on chance. Results must, however, be inter- The fourth column of Table 2 gives retest correlations for
preted with caution. the 113 individuals in Subsample A' who completed the full
Significant main effects for administration were found for CQ at both times. Of the 30 correlations, 22 are significant;
isolation of affect, which decreased, and for fatalism, dis- the median correlation is .26. It might be argued that the
traction, social comparison, restraint, wishful thinking, and magnitude of the correlations is inflated by the fact that the
passivity, which all increased between 1980 and 1987. same type of stressor is assessed at each time, and consist-
These findings are summarized in the fifth column of Table ency of stressor may account for the observed stability.
3. Only the increase in the use of fatalism replicated the When retest correlations are examined within type of
results of the repeated measures analyses of the short version stressor, Ns are greatly reduced, so many correlations fail to
of the CQ. As a whole, this pattern of results also fails to attain significance. The general pattern of results, however,
replicate either cross-sectional or cross-sequential analyses appears to be similar, and the median correlation across the

Downloaded from http://geronj.oxfordjournals.org/ at Penn State University (Paterno Lib) on February 20, 2016
of scales from the full CQ. three types of stressors is .26. These analyses suggest mod-
It appears from the supplementary analyses of the short est stability for a wide variety of coping mechanisms.
CQ scales that many of the effects seen in the longitudinal The fifth column of Table 2 gives correlations between
analyses of the full CQ may be due to procedural differences scores on 27 of the specific coping mechanisms and the two
in the 1980 and 1987 administrations. Neither age nor time coping factors for the 78 individuals in Subsample B'. As in
of measurement itself appears to be systematically related to the repeated measures analyses, only Time 1 responses for
the changes seen there. the same type of stressor as that assessed at Time 2 are
scored. Of the 29 correlations, 17 are significant; the median
Summary of mean level differences and changes. correlation is .25.
Maturational changes in the use of coping mechanisms It should be recalled that the correlations for Subsample
might be seen in cross-sectional differences, in increases or B' are based on short versions of the scales in a smaller
decreases over time in repeated measures analyses, or in sample. The fact that 13 of the significant stability coef-
differences between administrations in cross-sequential ficients from the analyses of the full CQ are replicated under
analyses. The most meaningful pattern of effects would these conditions is remarkable. In addition, analyses of
show changes or differences in the same direction for all Subsample B' suggest significant stability for two mecha-
these kinds of analyses (Costa & McCrae, 1982). Analyses nisms social comparison and active forgetting that are
on the full CQ and supplementary analyses on the short not significant in analyses of Subsample A'. Together, 22 of
version summarized in Table 3 gave five opportunities for 28 coping mechanisms show some consistency in coping
maturational effects to occur; none of the coping mecha- with two different events separated by seven years.
nisms showed a consistent pattern of replicated effects There appear to be patterns of differential stability among
across all five sets of analyses. The closest approximations coping mechanisms. Six of the seven coping mechanisms
are found for self-adaptation and wishful thinking. Evidence that define neurotic coping show significant retest stability in
for a maturational decline in self-adaptation is found in the both studies; by contrast, only two of the five coping mecha-
cross-sectional analyses and the two repeated measures nisms from the mature coping factor show this pattern. It
analyses. Cross-sequential analyses did not show significant appears that coping mechanisms related to the personality
effects for self-adaptation. Increases in wishful thinking are disposition of Neuroticism are moderately stable over a
suggested by the two cross-sequential analyses and the period of seven years. By contrast, coping mechanisms that
repeated measures analysis of Subsample A', but not by the are determined more directly by situational requirements
repeated measures analysis of Subsample B', nor by the show lower levels of stability.
cross-sectional results.
Reporting inconsistent results across methods of analysis DISCUSSION
is not unusual when cross-sectional, longitudinal, and cross- Do ways of coping change as people age? Consistent with
sequential designs are all employed in the same data set most of the literature, the present study suggests that such
(e.g., Costa & McCrae, 1988). When large samples are changes, if they occur at all, are quite subtle in their in-
used, as in the present study, very subtle effects of sampling fluence. Cross-sectional analyses show significant associa-
bias, practice, or time of measurement effects can appear as tions of age with many different coping mechanisms, but all
significant findings. In isolation, they might be misinter- the associations are modest in magnitude, and only a few
preted as evidence of maturational change. Together, their replicate previous findings. Further, changes over a seven-
mutual inconsistency suggests that there are no robust age year period assessed by both repeated measures and cross-
effects in this sample of community-dwelling individuals. sequential designs provide no support for the view that age
differences are due to maturation. Instead, they might better
Retest stability. Correlations between 1980 data and be interpreted as cohort effects.
1987 data can be used to assess the stability of individual Perhaps the most consistent cross-sectional finding is that
differences in the use of coping mechanisms. As in studies of older individuals use less interpersonal aggression in dealing
mean levels, these analyses are complicated by the fact that with stress. In both Time 1 and Time 2 analyses, age was
individuals coped with different specific events at the two negatively associated with the use of hostile reaction. (In
times. In consequence, the retest correlations estimate con- cross-sectional analyses of the Folkman et al. scales, older
P168 MCCRAE

subjects used confrontive coping less, as Folkman et al., individual differences in coping. The retest correlations in
1987, and Irion & Blanchard-Fields, 1987, had reported Table 2 are modest in comparison to those typically found
earlier.) Felton and Revenson (1987) similarly found that for personality measures (e.g., Costa & McCrae, 1988), but
older subjects were less prone to use emotional expression, considering the reliability of the individual scales, the differ-
which resembles hostile reaction. Finally, Costa, Zonder- ences in the stressor on the two occasions, and the lack of
man, and McCrae (in press) found that older men and aggregation across stressors, the retest correlations in Table
women were more forgiving and less likely to vent frustra- 2 may be seen as remarkably high. The 7-year stability
tions on others. Younger cohorts apparently have a greater coefficients for neurotic coping approach the level of
tolerance for aggression; they may believe that the expres- 7-month and 1-year retest reliabilities of coping mechanisms
sion of hostility is an effective and healthy way to respond to (Felton & Revenson, 1984; Solomon et al., 1988). Further,
frustration and conflict. Although this is a common notion in there is strong agreement on the stability of individual
popular psychology, it may be incorrect: Hostile reaction mechanisms between two studies conducted on independent
was rated as one of the least effective ways to cope with samples using different versions of the CQ. Some of the

Downloaded from http://geronj.oxfordjournals.org/ at Penn State University (Paterno Lib) on February 20, 2016
stress by those who had used it (McCrae & Costa, 1986). stability of coping behaviors can be attributed to enduring
The reliability of coping scales is a concern in the present personality variables, such as Neuroticism, which influence
study. The original research (McCrae, 1982) was explor- the choice of coping mechanisms. However, coping behav-
atory; a large number of coping mechanisms were examined iors may also be learned and retained independently of
in order to detect possible age differences. Precision of personality. Longitudinal studies that include measures of
measurement was traded for scope: Many coping mecha- both personality and coping behaviors can address that issue.
nisms were assessed by a single item or a very brief scale. The fact that age does not seem to have a central impact on
These features were carried into the follow-up study. It is, of coping does not mean that gerontologists should abandon
course, possible that age differences and changes might be interest in the topic. The major issue in coping research is the
identified if the coping mechanisms examined here were ultimate effectiveness of different coping strategies in help-
more reliably measured. This is particularly true for some of ing individuals deal with stress (Aldwin & Revenson, 1987;
the single-item scales such as avoidance and assessing McCrae & Costa, 1986). Basic research on coping can offer
blame, which fail to show significant retest correlations. a rational and systematic basis for interventions designed to
However, other single-item scales, such as faith and wishful help older people cope with such stressors as physical
thinking in the short form of the CQ, show substantial impairment, cognitive decline, and bereavement.
stability across seven years; such scales must possess at least
moderate reliability and could be expected to show age ACKNOWLEDGMENTS

changes if they were present. It is also important to recall The author thanks Stephanie Stone for coding characteristics of the
that many of the scales do show acceptable levels of reliabil- stressors.
ity. Hostile reaction and escapist fantasy, which have shown Address correspondence and reprint requests to Dr. Robert R. McCrae,
consistent cross-sectional associations with age, have alpha Personality, Stress, and Coping Section, Gerontology Research Center,
reliabilities of .83 and .68, respectively (McCrae, 1984) 4940 Eastern Avenue, Baltimore, MD 21224.
yet neither shows any evidence of longitudinal change.
Similarly, the two coping factors are broad, reliable, and REFERENCES

stable, but fail to show maturational effects. Aldwin, C. M., & Revenson, T. A. (1987). Does coping help? A reexami-
It is useful to recall that the present study examines coping nation of the relation between coping and mental health. Journal of
in response to a single event, which provides a limited Personality and Social Psychology, 53, 337-348.
Costa, P. T., Jr., & McCrae, R. R. (1982). An approach to the attribution of
sample of coping behavior. In order to see the systematic age, period, and cohort effects. Psychological Bulletin, 92, 238-250.
effects of age on coping, it may be necessary to aggregate Costa, P. T., Jr., & McCrae, R. R. (1988). Personality in adulthood: A six-
coping responses to a large number of stressful events. It year longitudinal study of self-reports and spouse ratings in the NHO
might also be necessary to extend the retest interval. Al- Personality Inventory. Journal of Personality and Social Psychology, 54,
though long in comparison with other published studies, it is 853-863.
Costa, P. T., Jr., & McCrae, R. R. (1989). Personality, stress, and coping:
short in comparison with the full adult life span. Over seven Some lessons from a decade of research. In K. S. Markides & C. L.
years, slowly progressing changes may not be seen. A Cooper (Eds.), Aging, stress and health, (pp. 267-283). New York:
definitive study would need to cover a longer interval, use Wiley.
more reliable measures, aggregate over many events, and Costa, P. T., Jr., Zonderman, A. B., & McCrae, R. R. (in press).
use the same procedure for selecting the stressors at each Personality, stress, and coping in older adulthood. In A. L. Greene,
time point. In the meantime, however, there is little evidence E. M. Cummings, & K. H. Karraker (Eds.), Life-span developmental
psychology. Vol. 11: Perspectives on stress and coping. Hillsdale, NJ:
of pervasive maturational effects on coping, and a conserva- Lawrence Erlbaum Associates.
tive assessment would be that the influence of aging on Felton, B. J., & Revenson, T. A. (1984). Coping with chronic illness: A
coping processes is at best "limited in scope and modest in study of illness controllability and the influence of coping strategies on
strength" (Felton & Revenson, 1987, p. 168). psychological assessment. Journal of Consulting and Clinical Psychol-
ogy. 52, 343-353.
If age is not a major determinant of coping, what is? Felton, B. J., & Revenson, T. A. (1987). Age differences in coping with
Certainly characteristics of the situation play a major role; in chronic illness. Psychology and Aging, 2, 164-170.
the present study, for example, rational action was used Folkman. S., & Lazarus, R. S. (1980). An analysis of coping in a middle-
aged community sample. Journal of Health and Social Behavior, 21,
more in dealing with challenge, faith more in dealing with
2*19-239.
threat and loss. However, there also appear to be consistent Folkman, S., Lazarus, R. S., Dunkel-Schetter, C , DeLongis, A., &
LONGITUDINAL ANALYSES OF COPING P169

Gruen, R. J. (1986). Dynamics of a stressful encounter: Cognitive Pfeiffer, E. (1977). Psychopathology and social pathology. In J. E. Birren
appraisal, coping, and encounter outcomes. Journal of Personality and & K. W. Schaie (Eds.), Handbook of the psychology of aging (1st ed.,
Social Psychology, 50, 992-1003. pp. 650-671). New York: Van Nostrand Reinhold.
Folkman. S., Lazarus, R. S., Pimley, S., & Novacek, J. (1987). Age Schaie, K. W. (1977). Quasi-experimental research designs in the psychol-
differences in stress and coping processes. Psychology and Aging, 2, ogy of aging. In J. E. Birren & K. W. Schaie (Eds.), Handbook of the
171-184. psychology of aging (1st ed., pp. 36-69). New York: Van Nostrand
Haan. N. (1977). Coping and defending. New York: Academic Press. Reinhold.
Irion, J. C & Blanchard-Fields, F. (1987). A cross-sectional comparison Schroeder, D. H., & Costa, P. T., Jr. (1984). Influence of life event stress
of adaptive coping in adulthood. Journal of Gerontology, 42, 502-504. on physical illness: Substantive effects or methodological Raws'? Journal
Lazarus, R. S., & DeLongis, A. (1983). Psychological stress and coping in of Personality and Social Psychology, 46, 853-863.
aging. American Psychologist, 38, 245-254. Shock, N. W., Greulich, R. C , Andres, R., Arenberg, D., Costa, P. T.,
Lazarus, R. S., & Launier, R. (1978). Stress-related transactions between Jr., Lakatta, E. G., & Tobin, J. D. (1984). Normal human aging: The
person and environment. In L. A. Pervin & M. Lewis (Eds.), Perspec- Baltimore Longitudinal Study of Aging (NIH Publication No. 84-2450).
tives in interactional psychology (pp. 287-327). New York: Plenum. Bethesda, MD: National Institutes of Health.
McCrae, R. R. (1982). Age differences in the use of coping mechanisms. Solomon, Z., Mikulincer, M., & Avitzur, E. (1988). Coping, locus of
Journal of Gerontology, 37, 454-460. control, social support, and combat-related posttraumatic stress disorder:

Downloaded from http://geronj.oxfordjournals.org/ at Penn State University (Paterno Lib) on February 20, 2016
McCrae, R. R. (1984). Situational determinants of coping responses: Loss, A prospective study. Journal of Personality and Social Psychology, 55,
threat, and challenge. Journal of Personality and Social Psychology, 46, 279-285.
919-928. Vaillant, G. E. (1977). Adaptation to life. Boston: Little, Brown.
McCrae, R. R. (in press). Situational determinants of coping. In B. N. Vaillant, G. E., Bond, M., & Vaillant, C. O. (1986). An empirically
Carpenter (Ed.), Personal coping: Theory, research, and applications. validated hierarchy of defense mechanisms. Archives of General Psychi-
New York: Praeger. atry, 43, 786-794.
McCrae. R. R.. & Costa, P. T., Jr. (1986). Personality, coping, and coping
effectiveness in an adult sample. Journal of Personality, 54, 385-405.
McCrae, R. R.. & Costa, P. T., Jr. (1988). Psychological resilience among
widowed men and women: A 10-year followup of a national survey. Received September 26, 1988
Journal of Social Issues, 44, 129-142. Accepted February 14, 1989

Das könnte Ihnen auch gefallen