Sie sind auf Seite 1von 9

Journal of Consumer Research, Inc.

The Role of Attributions in Customer Satisfaction: A Reexamination Author(s): MichaelTsiros, VikasMittal, and WilliamT.Ross, Jr. Source: Journal of Consumer Research, Vol. 31, No. 2 (September 2004), pp. 476-483 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/10.1086/422124 . Accessed: 10/10/2013 12:33
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

The University of Chicago Press and Journal of Consumer Research, Inc. are collaborating with JSTOR to digitize, preserve and extend access to Journal of Consumer Research.

http://www.jstor.org

Electronic copy available at: http://ssrn.com/abstract=2343894

This content downloaded from 168.7.215.220 on Thu, 10 Oct 2013 12:33:56 PM All use subject to JSTOR Terms and Conditions

The Role of Attributions in Customer Satisfaction: A Reexamination


MICHAEL TSIROS VIKAS MITTAL WILLIAM T. ROSS, JR.*
We investigate the role of disconrmation, responsibility, and stability attributions in the formation of satisfaction judgments. Building on the valence-expectancy framework, we nd that disconrmation and attributions impact satisfaction in a complex manner. Besides its main effect, responsibility moderates disconrmations effect on satisfaction, manifested as a two-way interaction between the two. Disconrmation and responsibility jointly determine the valence component, and stability determines the expectancy component of the satisfaction evaluation. This is consistent with the three-way interaction among stability, responsibility, and disconrmation that we also nd. These results clarify past studies and provide new insights about the relationship among the constructs.

n the customer satisfaction literature, attribution theory has mainly been used to understand consumer satisfaction judgments under differential attributions. Taking a dimensional view, researchers have examined how attribution dimensions such as responsibility, controllability, and stability inuence satisfaction judgments. A survey of the empirical attribution research in the area of satisfaction (see table 1) shows that key questions in the area remain unexplored. First, a conceptual model of how the various types of attributions combine with disconrmation to impact satisfaction is lacking. We posit that the valence-expectancy framework (Oliver 1974; Teas 1981; Vroom 1964) provides the beginnings of such a model. Second, though researchers have examined more than one attribution dimension in a single study, it is not clear how various types of attributions interact with each other or with other antecedents of satisfaction. This article seeks to address these issues with respect to two specic attribution dimensions: responsibility, whether the consumer holds the service provider responsible for the outcome, and stability, whether the consumer holds the outcome to emanate from a stable versus unstable cause. Our results show that disconrmation and attributions in-

uence satisfaction in a complex manner not captured by either a simple main-effects approach or a set of simple twoway interactions. The rival models emanating from these different approaches are summarized in gure 1 and discussed later. Results show that the pattern of observed interactions is consistent with the valence-expectancy framework (Oliver 1974; Teas 1981; Vroom 1964).

CAUSAL ATTRIBUTIONS AND SATISFACTION: BACKGROUND


Most satisfaction studies have taken a multidimensional view of attributions to examine if and how various attribution types inuence satisfaction judgments and behavioral intentions differently. From table 1, two conclusions can be drawn. First, within the multidimensional view of attributions, the specic dimensions used have varied from study to study. Thus, an issue that arises is: what attribution dimensions are salient to post purchase evaluation? Weiners early conceptualization (1985) employed three attributional dimensionslocus of causality, controllability, and stabilitythat have been used in most marketing studies (see table 1). Recently, however, Weiner (2000) concluded that attributions about responsibility and stability are most salient for understanding postconsumption reactions. Following this, we examine stability as one attributional dimension. Responsibility, however, seems to incorporate both controllability and locus of causality. From a consumers perspective the issue is to assign responsibility for the obtained outcome. Clearly, locus of causality, that is, who caused the failure, is an important part of responsibility. But so too is controllability, the degree of control the causal party had on
476
2004 by JOURNAL OF CONSUMER RESEARCH, Inc. Vol. 31 September 2004 All rights reserved. 0093-5301/2004/3102-0021$10.00

*Michael Tsiros is associate professor of marketing, School of Business Administration, University of Miami, Coral Gables, FL 33124 (mtsiros @exchange.sba.miami.edu) and Tassos Papastratos Research Associate Professor, ALBA, Athens, Greece, 16671. Vikas Mittal is associate professor of marketing, Katz Graduate School of Business, and associate professor of psychiatry, School of Medicine, University of Pittsburgh, Pittsburgh, PA 15260 (vmittal@katz.pitt.edu). William T. Ross, Jr., is professor of marketing, Smeal College of Business Administration, Pennsylvania State University, University Park, PA 16802 (wtr2@psu.edu). The authors acknowledge the helpful input of the editor, associate editor, and reviewers.

Electronic copy available at: http://ssrn.com/abstract=2343894

This content downloaded from 168.7.215.220 on Thu, 10 Oct 2013 12:33:56 PM All use subject to JSTOR Terms and Conditions

TABLE 1 A REVIEW OF EMPIRICAL FINDINGS ON ATTRIBUTION AND SATISFACTION Study/method Oliver and DeSarbo (1988) experimental Folkes (1984; study 1) Experiment (study 2, within subjects) Attribution/valence Locus (internal/external) Valence: positive and negative Valence: negative Locus, controllability, stability (not clear manipulation though: conditions were not perfectly comparable) Valence: negative Independent: locus (L), controllability (C), stability (S) Dependent: expectancy reactions (expectancy of future performance), market equity reactions (refund or apology), anger reaction Variables Independent: locus, expectations, performance, disconrmation, equity Dependent: satisfaction Main ndings Signicant main effects. Signicant two-way interactions between expectations and performance, disconrmation, and equity. Stable and controllable causes predominately lead to product failure (these two dimensions are highly correlated, r p .94). Expectancy reactions are inuenced by stability. Market equity reactions are inuenced by locus. Interaction effects: anger reactions are inuenced by a twoway interaction (L # C). Market equity reactions are inuenced by two- and three-way interactions (L # S, L # C, L # C # S). Apologies and refunds are less deserved when consumer related than when manufacturer related. Manufacturer-uncontrolled-stable causes deserve fewer apologies and refunds than (1) manufacturer-controlled-stable cases and (2) manufacturer-controlled-unstable causes. Consumer-uncontrolled-stable causes deserve more apologies and refunds than (1) consumer-uncontrolled-unstable causes and (2) consumer-controlled-stable causes. Controllability and stability inuence anger, which inuences complaint and repurchase intentions. Controllability and stability also have a direct effect on complaint and repurchase intentions.

Folkes, Koletsky, and Graham (1987) survey, path analysis

Stability, controllability Valence: negative

Curren and Folkes (1987) experiment (study 2)

Stability, controllability, locus Valence: negative Stability, controllability Valence: positive Locus (internal: consumer; external: marketing institution) Valence: negative Controllability, self-reported attribution measures (PREVENT: hotel could have prevented the problem; CONTROL: degree of control hotel had over the problem); these were covariates in study 2 and cant nd their effect in the ndings Valence: negative

Experiment (study 3)

Richins (1983) Survey

Smith, Bolton, and Wagner (1999) experiment (two identical studies: restaurant and hotel). Study 2 examined attributions

Independent: stability, controllability Dependent: expectancy reactions (expectancy of future performance), market equity reactions (refund or apology), anger reaction Independent: stability, controllability, locus Dependent: complaint intentions, negative word of mouth Independent: stability, controllability Dependent: compliment, positive word of mouth Independent: dissatisfaction, problem severity Dependent: complaint behavior, word of mouth Independent: Between subjects: type of failure, (process/outcome), magnitude of failure (low/ high) Within subjects: compensation (high/medium/ none), response speed (immediate/delayed), apology (present/absent), recovery initiation (employee/customer). Dependent: distributive justice, procedural justice, interactional justice, service encounter, satisfaction

Locus inuences complaint and word of mouth, stability inuences word of mouth (same as Folkes 1984). Found signicant main effects.

Positive correlation between problem severity and word of mouth. External attributions (blame the marketing institution for poor performance) lead to more complaints and negative word of mouth Main effects: service encounter satisfaction after a service failure is inuenced by type and magnitude of failure. More dissatisfaction was experienced by customers experiencing process failures than outcome failures. High-magnitude failures lead to more dissatisfaction than lo-magnitude failures. All three types of justice inuence satisfaction, with distributive justice having the largest effect. Compensation increases distributive justice. Speedy recovery increases procedural justice. Apology and employee-initiated recovery increase interactional justice. Interaction effects: A speedy recovery has a greater effect on procedural justice after an outcome failure. Compensation and recovery speed have a greater effect on perceived justice when failure magnitude is low.

This content downloaded from 168.7.215.220 on Thu, 10 Oct 2013 12:33:56 PM All use subject to JSTOR Terms and Conditions

478
FIGURE 1 PROPOSED MODELS OF CAUSAL ATTRIBUTIONS AND SATISFACTION

JOURNAL OF CONSUMER RESEARCH

tributions exist and should be examined, though such an examination is beyond the scope of this article. Second, at least some attributions directly affect satisfaction judgments. In response to the same outcome, satisfaction can vary based on the type of attributions made by the consumer. It seems that attributions can have an interactive effect on satisfaction judgments as well as direct effects. However, within the satisfaction literature, we are aware of only two studies that have actually included interactive effects (Folkes 1984, study 2; Oliver and DeSarbo 1988). Folkes (1984, study 2) examined negative disconrmation, with anger, expectancy and market-equity reactionsnot satisfaction judgments or behavioral intentionsas dependent variables. She found that fewer apologies and refunds were warranted if the negative disconrmation was due to a cause that was stable but uncontrollable than if the cause was stable but controllable. This study provides a conceptual basis to argue for a complex approach to understanding the role of attributions as antecedents of satisfaction and intentions. Oliver and DeSarbo (1988) experimentally manipulated locus of causality and disconrmation but did not nd a signicant interaction between them in determining satisfaction with stocks. They operationalized locus of causality as being internal (respondents picked the stock themselves) or external (respondents picked the stock based on a brokers recommendation). However, locus of causality may have been confounded with stability of attributions. Those attributing the stock pick to their decision (internal control) may also have made stable attributions under positive disconrmation (their own ability) but unstable attributions under negative disconrmation (bad luck). This is all the more likely, as their sample of MBA students had prior stock market experience.

DISCONFIRMATION, ATTRIBUTIONS, AND SATISFACTION: A VALENCEEXPECTANCY APPROACH


Within the valence-expectancy framework people independently gauge two factorsthe objects valence and expectancyto evaluate the object. Valence is the extent to which someone feels positively or negatively about an event occurring, and expectancy estimates enable the person to gauge the likelihood of the event reoccurring. We argue that, just as the valence-expectancy model has been used to determine factors that lead to motivation (Vroom 1964), it can be used to consider how attributions and disconrmation combine to affect satisfaction. Disconrmation is evaluative in nature and should be a contributor to the valence component. Similarly, responsibility should be a contributor to the consumers evaluation of the event and, thus, related to the valence component. Finally, stability deals with how likelyin the futurethe outcome is perceived by the consumer to continue. Thus, it should be a primary contributor to the expectancy component. As a result, we suggest that attributions and disconrmation interact to determine sat-

the circumstances. So, we follow Weiner (2000) and suggest that the combination of these two dimensions enables the consumer to determine responsibility.1 Thus, the second attribution dimension we use is responsibility. We do this with the explicit acknowledgement that other dimensions of at1 Folkes (1984) measured these two attribution dimensions separately and found a correlation (r) of .94. Thus, it seems that the locus of causality and controllability dimensions should be highly correlated and jointly determine responsibility.

This content downloaded from 168.7.215.220 on Thu, 10 Oct 2013 12:33:56 PM All use subject to JSTOR Terms and Conditions

ATTRIBUTIONS AND CUSTOMER SATISFACTION

479

isfaction. We propose specic patterns of interactive effects based on the valence-expectancy model. Prior research (see table 1) has shown a consistent main effect for responsibility. Hence, although we do not focus on the direct effects of responsibility, we include it as a direct predictor in our models. Evidence for stability is decidedly mixed in table 1, and our conceptualization, in fact, implies that stability might not have a main effect; testing our conceptualization against previous work requires inclusion of stability as a main effect in the models. Thus, disconrmation (positive and negative) and responsibility attributions jointly affect the valence component of the valence-expectancy framework. Specically, responsibility attributions should moderate the effect of disconrmation on satisfaction such that, when consumers hold the rm responsible for an experienced outcome; disconrmation will have a stronger effect on satisfaction than when they do not. Our reasoning here is consistent with Weiner (2000, p. 385) who argues that responsibility attributions link to inferences regarding personal responsibility, moral judgment, and moral emotions including anger, sympathy, and gratitude, so that they are at the very heart of social behavior. Additionally, responsibility attributions may also intensify the impact of disconrmation on satisfaction via the level of experienced regret (Tsiros and Mittal 2000). If the company is responsible for negative disconrmation, consumers may regret choosing the brand more than if the negative disconrmation occurs due to non-company-related factors (e.g., a natural disaster). Conversely, under positive disconrmation, consumers may rejoice in their choice more if they nd that the company itself is responsible than an external factor. Counter to past studies, this model does not posit responsibility attributions as affecting satisfaction only directly. In addition to a direct impact, responsibility attributions affect the valence component of the valence-expectancy framework by moderating the role of disconrmation. Next, we argue that stability attributions affect the expectancy component. As such, stability attributions should have a multiplicative effect with the valence component (disconrmation and responsibility attributions). In other words, once consumers have formed a valence (as a result of disconrmation and responsibility), stability attributions affect the expectancy associated with such valence. Thus, when the cause of the experienced outcome is seen as stable, the valence component (disconrmation # responsibility) should have a stronger effect on satisfaction than when the cause is seen as unstable. This occurs because, when the outcome is attributed to an unstable cause, it is likely to be deemed transitory and have a weaker impact on satisfaction than when stable causal attributions are made. Conceptually, this implies that stability attributions will moderate the moderating impact of responsibility attributions on disconrmation (the valence component) but is silent about whether it will have its own direct effect. This is stated as follows: H1: Attributions of stability moderate the effect of responsibility on the disconrmation-satisfaction re-

lationship. Specically, the moderating effect of responsibility attributions is stronger when the experienced outcome is attributed to a stable cause, but weaker when it is attributed to an unstable cause. Hypothesis 1 is consistent with prior attribution literature suggesting that attributions regarding stability are likely to affect consumer expectancies and thus inuence probabilistic judgments about the likelihood of an outcome (Curren and Folkes 1987; Folkes 1984; Weiner 1985, 2000). It is also consistent with the satisfaction literature, where the notion of stability affecting expectancies is consistent with the belief-updating framework (cf., Anderson and Sullivan 1993). Accordingly, customers update their priors and incorporate them in their satisfaction judgment only when the cause is also deemed to be stable.

Rival Models. Empirically, the valence-expectancy framework posited above is contrasted against two rival models shown in gure 1. In model A, the base model used in most of the empirical research in table 1, disconrmation and attribution dimensions directly inuence satisfaction. Support for this model would manifest itself as statistically signicant main effects of disconrmation, responsibility, and stability. In model B, a simple interaction model, stability and responsibility moderate the effect of disconrmation on satisfaction. This model would be supported by statistically signicant two-way interactions between responsibility and disconrmation, and stability and disconrmation, but no three-way interaction. Model C is consistent with the valence-expectancy framework where responsibility attributions and disconrmation jointly affect the valence component. The valence part of this model should manifest itself as signicant main effects of disconrmation and responsibility and as a two-way interaction between responsibility and disconrmation. The expectancy part of the model should manifest as a signicant three-way interaction among disconrmation, responsibility, and stability. A main effect of stability may or may not be present.

STUDY Design and Procedure


The design was a two (disconrmation: positive vs. negative) # two (stability of the cause: stable vs. unstable) # two (responsibility: company-related vs. company-unrelated) between-subjects design. Respondents were 202 executive MBAs at a midwestern university. Averages were: age, 29.3 yr.; work experience, over 6 yr. full-time; employees supervised, six or more full-time. Almost half were involved in purchasing decisions, and one-third had evaluated a partner rms performance.

Decision Scenario
Participants assumed the role of the vice president of an electronic equipment manufacturer currently evaluating a

This content downloaded from 168.7.215.220 on Thu, 10 Oct 2013 12:33:56 PM All use subject to JSTOR Terms and Conditions

480

JOURNAL OF CONSUMER RESEARCH


TABLE 2 ANALYSIS OF VARIANCE Source Disconrmation (D) Responsibility (R) Stability (S) D#R D#S R#S D#R#S Error df 1 1 1 1 1 1 1 193 Mean square 371.79 2.07 .41 20.85 .80 2.26 6.12 1.13 F-value 329.02 1.83 .359 18.45 .71 2.00 5.42 P-value .001 .178 .550 .001 .405 .159 .020

distributor and read the scenario in the appendix. In the scenario they were told that they had constant sales (a growth of 0%) based on their partnership with their current distributor (Alpha). Expectations for the next years sales were also set at 0%, that is, constant. The scenario then shifted to 1 yr. later, and participants were told about company Alphas performance, which was either a 10% increase (positive disconrmation) or a 10% decrease (negative disconrmation). In addition, they were informed of the cause that may have contributed to that level of observed outcome. The cause varied on two dimensions, stability and responsibility. It was manipulated using a widely accepted classication (cf. Teas and McElroy 1986; Valle and Frieze 1976) as follows:
Company related (company is responsible) Ability Effort Company unrelated (company is not responsible) Task difculty Luck

Stable Unstable

Next, respondents provided measures on satisfaction and manipulation checks.

Measures
Dependent Measure. Satisfaction was measured using a seven-point scale (1 p strongly disagree, 7 p strongly agree). The three items are shown in the appendix and loaded on a single factor (alpha p .95). They were averaged to create a single measure where high score indicates a high level of satisfaction. Manipulation Checks. One-tailed t-tests were used to assess the manipulations. Participants rated disconrmation using a seven-point scale (7 p much better than expected, 1 p much worse than expected). The positive disconrmation group had higher ratings than the negative disconrmation group (5.83 vs. 2.92, p ! .01). For responsibility, participants rated their agreement (7 p strongly agree, 1 p strongly disagree) with the statement Company Alpha is responsible for the recent level of performance. The mean was higher for the high than the low-responsibility condition (5.49 vs. 2.58, p ! .01). For stability, participants rated their agreement (7 p strongly agree, 1 p strongly disagree) with the statement The cause of Alphas performance is stable. The mean was higher for the stable-attribution group than the unstable-attribution group (5.69 vs. 2.93, p ! .01). Thus, all three manipulations were successful. Orthogonality Check. The correlations between the manipulation checks were statistically nonsignicant for positive (r p .12, p p .15) and negative (r p .04, p p .69) disconrmation. Thus the manipulations were orthogonal.

rmation, stability, and responsibility. As shown in table 2, there is a signicant three-way interaction among disconrmation, responsibility, and stability (F(1, 193) p 5.42, p ! .02, h2 p .04). The moderating effect of responsibility in the disconrmation-satisfaction relationship is much stronger for the stable than for the unstable condition. In panel 1 of gure 2, where attributions are unstable, there is no difference in the effect of responsibility on disconrmation and satisfaction. Whether the company is responsible or not, the effect of disconrmation is identical on satisfaction. We constructed a contrast (2.43 2.26 vs. 5.80 5.71) and found it to be statistically nonsignicant (p 1 .10). Conversely, in panel 2, where attributions are stable, a moderating effect of responsibility on disconrmation and satisfaction is found. The effect of disconrmation on satisfaction is stronger when the company is responsible than when it is not. A contrast (1.31 3.50 vs. 6.12 5.54) was signicant ( p ! .01) and conrms this.

Results
Table 2 shows the ANOVA, and gure 2 shows the means. Hypothesis 1 posits a three-way interaction among discon-

Model Comparison. It is useful to consider the results of the rival models from gure 1. Model A, the base model, tests previous research investigating only main effects and has an adjusted R2 of 54%. Disconrmation and responsibility have statistically signicant main effects on satisfaction, and stability has a moderately signicant main effect. Model B, the simple-interaction model, includes only the main effects and two-way interactions and has an adjusted R2 of 60%. The main effect of stability approaches statistical signicance, and responsibility and disconrmation have signicant main effects. There is only one statistically signicant interaction, between responsibility and disconrmation. Model C, the valence-expectancy framework, has an adjusted R2 of 63%. Consistent with the valence-expectancy framework, disconrmation has a statistically significant main effect and the main effect of responsibility only approaches signicance perhaps because the interaction between disconrmation and responsibility attributions is statistically signicant. Jointly, these components determine the valence aspect of the model. Consistent with the expectancy part of the model, there is a statistically signicant threeway interaction. In summary, model C is the preferred model.

This content downloaded from 168.7.215.220 on Thu, 10 Oct 2013 12:33:56 PM All use subject to JSTOR Terms and Conditions

ATTRIBUTIONS AND CUSTOMER SATISFACTION


FIGURE 2 EFFECT OF DISCONFIRMATION, RESPONSIBILITY, AND STABILITY ON SATISFACTION

481

DISCUSSION
These results were replicated in two additional studies. The rst consisted of a survey of 109 business managers who evaluated an actual business partner rather than a ctitious business partner used in the current study. In the second study, 250 undergraduates evaluated a recent purchase they had made. In both these studies we found an identical pattern to the one reported here. Thus, we are comfortable about the generalizability of the claims made here. Replication in these other studies also attests to the context invariance of the results, effectively demonstrating that attributional responses are important in both business and consumer environments. Finally, we note that in the study of managers, we used a performance measure, rather than disconrmation, with identical results. This is reassuring, since other research (e.g., Churchill and Suprenant 1982; Mittal, Ross, and Baldasare 1998; Tse and Wilton 1988) has shown performance to have an independent effect on satisfaction beyond disconrmation. Thus, both performance and disconrmation can be subsumed within the valence component of the model. The key theoretical contribution of this research is in providing the valence-expectancy framework as a basis for understanding the way in which disconrmation and attributions affect satisfaction (model C in g. 1). Responsibility attributions and disconrmation jointly form the valence component. More importantly, as per the expectancy component, stability attributions moderate the extent to which the valence component (disconrmation and responsibility attributions) affects satisfaction. Such a conceptualization best accounts for the observed pattern of results. A simple main-effects model (model A in g. 1), or a model examining only the two-way interactions (model B in g. 1), is unable to account for the complex pattern of results observed. Yet, it is these two types of models that underlie most of the research examining attributions and satisfaction. These results have several theoretical implications. First, the dimensional view of attributions taken by previous studies is appropriate, though the impact of these dimensions on satisfaction is better interpreted within the context of the valence-expectancy framework. This being the case, future studies should conceptually determine if an attribution dimension affects the valence and/or the expectancy part. Only those dimensions affecting the valence part should theoretically interact with disconrmation, while those affecting expectancies should be more relevant to the higher-order interactions. Thus the two-way interaction model (panel B in g. 1) may only be appropriate if the researcher believes that the attribution dimensions are part of the valence component. For attribution dimensions that may constitute the expectancy component, a three-way interaction is appropriate. When researchers have no a priori notions in this regard, it is not advisable simply to add two-way interactions among disconrmation and attribution dimensions but rather to include higher order interactions as well. Such a theory-driven approach can guide researchers in deter-

This content downloaded from 168.7.215.220 on Thu, 10 Oct 2013 12:33:56 PM All use subject to JSTOR Terms and Conditions

482

JOURNAL OF CONSUMER RESEARCH

mining which of the three rival models is appropriate for their data. Clearly, a pressing research need is to classify how different attribution dimensions map on to the valence and/or expectancy component. Such attempts at careful theoretical mapping can probably shed light on inconsistent past ndings. Smith, Bolton, and Wagner (1999) did not nd a separation of the two types of attribution in their empirical work. Our work suggests that it may not be enough to use conceptually distinct attribution dimensions. Rather, it is imperative to understand how these conceptually distinct attribution dimensions map on to the valence and/or expectancy component of the valence-expectancy framework. Another research opportunity is to examine the applicability of the valence-expectancy framework in areas other than postpurchase satisfaction, which is our focus. For example, McGill has an interesting series of articles (e.g., Iacobucci and McGill, 1990; McGill 1989) examining how background or context inuences respondents attributions, including one cross-cultural article (McGill 1995) that suggests that cultural background affects whether the respondent actually makes external attributions at all. Incorporating this line of reasoning is an opportunity for further work with the valence-expectancy framework in a satisfaction context. The valence-expectancy framework could be further expanded to include instrumentality (Vroom 1964). This would be especially useful in understanding consumers behavioral responses, in addition to satisfaction. In including instrumentality, we believe that taking an attribute-level approach to formulating satisfaction judgment (see Mittal et al. 1998) is a useful way to proceed. It is likely that the instrumentality of different attributes in determining satisfaction and behavioral responses varies, and that may moderate the extent to which both the valence and expectancies can have an effect.

goals. In addition, this estimate is used as a benchmark against which the performance of your department is measured. Therefore, it is important neither to inate nor to deate this estimate. Based on past experience and on the current economic conditions, you expect the sales gures to continue to remain at the same level as last year (an increase of 0%).

NOW, ONE YEAR LATER


One year has passed since you predicted the sales gures to remain constant. The gures for the past year have just been computed and you nd out that sales have increased (decreased) at the annual rate of 10%. Thus, company Alpha has performed better than it had in the past. This will not only improve your companys nancial standing but will also improve your standing in the company.

LOOKING BACK
Ability. Company Alphas performance has been consistent during the last several years. Their sales representatives are (are not) being continuously trained and educated on the latest changes in technology and on the products that they carry. They also have several years of (very little) work experience in this business. Their performance, therefore, is mainly due to their high (low) level of ability. Effort. Company Alphas performance has been inconsistent during the last several years. They have a low (high) turnover ratio and, therefore, most of their sales representatives have considerable (little) work experience in this business. In addition, they always (do not always) devote the same amount of effort and attention to their customers. Their performance on this task, therefore, is mainly due to their high (low) level of effort. Luck. Company Alphas performance has been inconsistent during the last several years. Their performance is determined mainly by the competitive strategies employed by your major competitors during a particular period. In addition, other unpredictable events and specic situations may play a major role in determining company Alphas performance in selling your products. Their performance, therefore, is mainly due to bad (good) luck. Difculty. Company Alphas performance has been consistent during the last several years. The geographic area that they service is (is not) a difcult one and that mainly determines their level of performance. This is so because the area that they service has been historically one of the most (least) difcult in terms of the presence (absence) of major competition from other manufacturers. Their performance, therefore, is mainly due to the difculty (lack of difculty) of the task.

APPENDIX
Imagine you are the Sales Vice President of BestTools, Inc., an electronics equipment manufacturer that supplies companies with a wide selection of industrial machinery. Your company has been in business for the past 30 years and you have been at your current position since 1986. BestTools hires independent distribution companies to sell its products. These companies, however, carry similar equipment from several of your competitors. One of the companies that you currently employ is Alpha Inc. Recently, however, you have been approached by other distribution companies. Although these other distribution companies provided attractive offers, you decided to go with company Alpha, since hooking up with them would include some additional investment and their superior performance was questionable. Tomorrow, you have to provide an estimate of the predicted sales gures for the next year to your superiors. This estimate is crucial because it determines the amount of resources that your department will be given to achieve its

This content downloaded from 168.7.215.220 on Thu, 10 Oct 2013 12:33:56 PM All use subject to JSTOR Terms and Conditions

ATTRIBUTIONS AND CUSTOMER SATISFACTION

483 tion, Journal of Personality and Social Psychology, 57 (2), 189200. . (1995), American and Thai Managers Explanations for Poor Company Performance: Role of Perspective and Culture in Causal Selection, Organizational Behavior and Human Decision Processes, 61 (January), 1627. Mittal, Vikas, William T. Ross, Jr., and Patrick M. Baldasare (1998), The Asymmetric Impact of Negative and Positive Attribute-Level Performance on Overall Satisfaction and Repurchase Intentions, Journal of Marketing, 62 (January), 3347. Oliver, Richard L. (1974), Expectancy Theory Predictions of Salesmens Performance, Journal of Marketing Research, 11 (August), 24353. Oliver, Richard L. and Wayne S. DeSarbo (1988), Response Determinant in Satisfaction Judgments, Journal of Consumer Research, 14 (March), 495507. Richins, Marsha L. (1983), Negative Word-of-Mouth by Dissatised Consumers: A Pilot Study, Journal of Marketing, 47 (Winter), 6878. Smith, Amy K., Ruth N. Bolton, and Janet Wagner (1999), A Model of Customer Satisfaction with Service Encounters Involving Failure Recovery, Journal of Marketing Research, 36 (August), 35672. Teas, R. Kenneth (1981) An Empirical Test of Models of Salespersons Job Expectancy and Instrumentality Perceptions, Journal of Marketing Research, 18 (May), 20926. Teas, R. Kenneth and James C. McElroy (1986), Causal Attributions and Expectancy Estimates: A Framework for Understanding the Dynamics of Salesforce Motivation, Journal of Marketing, 50 (January), 7586. Tse, David K. and Peter C. Wilton (1988), Models of Consumer Satisfaction Formation: An Extension, Journal of Marketing Research, 25 (May), 20412. Tsiros, Michael and Vikas Mittal (2000), Regret: A Model of Its Antecedents and Consequences in Consumer Decision Making, Journal of Consumer Research, 26 (March), 40117. Valle, Valerie A. and Irene Hanson Frieze (1976), Stability of Causal Attributions as a Mediator in Changing Expectations for Success, Journal of Personality and Social Psychology, 33 (5), 57987. Vroom, Victor H. (1964), Work and Motivation, New York: Wiley. Weiner, Bernard (1985), An Attributional Theory of Achievement Motivation and Emotion, Psychological Review, 92 (October), 54873. (2000), Attributional Thoughts about Consumer Behavior, Journal of Consumer Research, 27 (December), 38287.

MEASURES Manipulation Checks


1. Company Alphas performance was ___ (7, much better that expected; 1, much worse than expected). 2. Company Alpha is responsible for the recent level of performance (1, strongly disagree; 7, strongly agree). 3. The cause of Alphas performance is stable (1, strongly disagree; 7, strongly agree). Satisfaction (a p .95): (7-point scale: 1, strongly disagree; 7, strongly agree) 1. I feel satised with company Alphas performance. 2. I am happy with company Alphas performance. 3. I am pleased with company Alphas performance. [Dawn Iacobucci served as editor and Kent B. Monroe served as associate editor for this article.]

REFERENCES
Anderson, Eugene W. and Mary W. Sullivan (1993), The Antecedents and Consequences of Customer Satisfaction for Firms, Marketing Science, 12 (Spring), 12543. Churchill, Gilbert A. and Carol Suprenant (1982), An Investigation into the Determinants of Customer Satisfaction, Journal of Marketing Research, 19 (November), 491504. Curren, Mary T. and Valerie S. Folkes (1987), Attributional Inuences on Consumers Desires to Communicate about Products, Psychology and Marketing, 4 (Spring), 3145. Folkes, Valerie S. (1984), Consumer Reactions to Product Failure: An Attributional Approach, Journal of Consumer Research, 10 (March), 398409. Folkes, Valerie S., Susan Koletsky, and John Graham (1987), A Field Study of Causal Inferences and Consumer Reaction: The View from the Airport, Journal of Consumer Research, 13 (March), 53439. Iacobucci, Dawn and McGill, Ann L. (1990), Analysis of Attribution Data: Theory Testing and Effects Estimation, Journal of Personality and Social Psychology, 59 (3), 42641. McGill, Ann L. (1989), Context Effects in Judgment of Causa-

This content downloaded from 168.7.215.220 on Thu, 10 Oct 2013 12:33:56 PM All use subject to JSTOR Terms and Conditions

Das könnte Ihnen auch gefallen