The Role of Cognition and Affect in the Formation of Customer Satisfaction: A Dynamic Perspective

Abstract

Despite the strong recognition that customer satisfaction should be viewed from a dynamic perspective, little is known about how the satisfaction judgment develops over time. Therefore, this study provides a dynamic analysis of the simultaneous influence of cognition and affect in the satisfaction formation process. The results of an experimental study based on a real consumption experience indicate that the impact of cognition on the satisfaction evaluation increases and the influence of affect decreases over time. Moreover, these effects are attenuated with inconsistent performance experiences. Finally, the study shows that the variance in customer satisfaction jointly explained by cognition and affect increases as experience accumulates.

The topic of customer satisfaction/dissatisfaction has been of great interest to marketing and consumer researchers for many years. This interest is driven in part by the notion that customer satisfaction can have long-term benefits, including customer loyalty and increased profitability (e.g., Anderson, Fornell, and Lehmann 1994; Anderson, Fornell, and Rust 1997; Anderson and Sullivan 1993; Bearden and Teel 1983; Bolton and Drew 1991; Fornell 1992; LaBarbera and Mazursky 1983; Oliver 1980; Oliver and Swan 1989a, b; Rust, Moorman, and Dickson 2002; Rust and Zahorik 1993). Against this background, it is important to understand the mechanism that leads to customer satisfaction from both an academic and an applied perspective.

Previous research has recognized that both cognition and affect significantly predict satisfaction judgments. Cognition has been studied mainly in terms of the disconfirmation paradigm, which predicts satisfaction to be a function of a comparison between expectations and performance (e.g., Bearden and Teel 1983; LaBarbera and Mazursky 1983; Oliver 1980; Oliver and DeSarbo 1988). Other studies have recognized that the affect experienced during the acquisition and consumption of the product or service (e.g., joy, happiness, disgust) can also have a significant influence on satisfaction judgments (e.g., Mano and Oliver 1993; Westbrook 1987; Westbrook and Oliver 1991). Oliver (1997, p. 319) states that affect “coexists alongside various cognitive judgments in producing satisfaction” and that it is central to understanding customers’ consumption experiences. In addition, a key conclusion of Szymanski and Henard's (2001) meta-analysis is that disconfirmation and, to a lesser extent, affect are strongly related to satisfaction.

However, only a few studies have investigated cognitive and affective antecedents of customer satisfaction simultaneously. For example, in studying affects that are provoked by service failures, Smith and Bolton (2002) find that feelings predict satisfaction levels, after accounting for cognitive factors. Kempf's (1999) study finds that feelings (e.g., arousal, pleasure) are particularly important antecedents of product evaluations of hedonic products, whereas brand cognitions are not. The opposite is true for functional products. Oliver (1993) examines the role of disconfirmation and affective experiences and finds that both are critical to the modeling of satisfaction judgments.

Although previous studies identify the key roles of cognition and affect in influencing satisfaction, it is important to acknowledge that previous work has largely been static (i.e., cross-sectional) in nature. This represents a significant research gap because it is well established that customer satisfaction is a dynamic phenomenon. Even the classic confirmation/disconfirmation paradigm, which has guided much of the research in the area, recognizes that satisfaction judgments can have an important influence on customer expectations for subsequent consumption occasions (Oliver 1980). The dynamic nature of customer satisfaction is also emphasized in the work by Bolton (1998), who proposes a dynamic model of cumulative satisfaction. In a similar vein, Bolton and Lemon (1999) stress the importance of a dynamic perspective with their model of customer service usage. Despite this strong recognition that customer satisfaction should be viewed from a dynamic perspective, the role of cognitive and affective influences has not been systematically studied in this manner.

The few studies that investigate the antecedents of customer satisfaction from a dynamic perspective focus on the cognitive component of customer satisfaction. For example, Mittal, Kumar, and Tsiros (1999) and Mittal, Katrichis, and Kumar (2001) find that the weights given to an attribute in determining satisfaction shift over time on the basis of their salience. Slotegraaf and Inman (2004) show how satisfaction with product attributes can decline over time (particularly those that can be remedied and especially as customers approach the end of their warranty period). Other dynamic studies in this area focus on carryover effects of previous satisfaction judgments to subsequent satisfaction evaluations (Bolton and Drew 1991; Boulding et al. 1993; Mittal, Kumar, and Tsiros 1999). Bolton (1998) proposes a dynamic model in which the duration of customer relationship depends on an anchoring and adjustment process in which cumulative satisfaction serves as the anchor for subsequent experiences. Nevertheless, to the best of our knowledge, there is no study that investigates the effects of cognition and affect simultaneously over time.

Against this background, our study investigates cognition and affect simultaneously as drivers of customer satisfaction in a dynamic context. The fundamental proposition of our study is that the role of cognition and affect may change over time. More specifically, we argue that affect plays its strongest role at early stages of satisfaction development because of a lack of knowledge about the product or service. As experience accumulates, however, the impact of cognition should increase. To study this proposition, we examine customer satisfaction formation, beginning from a point at which customers have no prior experience with the product at all. In this respect, our study differs from previous work in this area, which essentially considers situations in which customers have some prior experience with the product or service. In addition, we examine this process in the context of moderately complex, utilitarian products.

In addition, from a dynamic perspective, the evaluation process that occurs during consumption serves as an input into subsequent decision-making situations. In other words, the nature of the feedback received has a strong influence on the satisfaction judgment, and this affects future decisions. Therefore, we build a case that the phenomenon under study also depends on the level of consistency of the consumption experience (i.e., Is it consistently positive, consistently negative, or inconsistent?). The importance of the concept of feedback consistency has been established in related areas, such as job performance (Alvero, Bucklin, and Austin 2001; Stone and Stone 1985). To the best of our knowledge, however, this concept has not been studied in the customer satisfaction literature.

Furthermore, previous research in social psychology and marketing has indicated that attitudes and judgments become more stable over time after multiple experiences with the attitude object (Fazio 1986; Petty, Unnava, and Strathman 1991; Smith and Swinyard 1983). On the basis of this literature, we argue that over time and with repeated consumption experiences, satisfaction judgments should also become more stable and that the ability of both cognitive and affective factors to predict customer satisfaction should increase.

Thus, our research is in line with the frequent call for longitudinal studies in marketing (e.g., Bolton 1998; Fournier and Mick 1999; Szymanski and Henard 2001). In addition to the theoretical interest of this topic, it has practical relevance because it aids managers in understanding the development of customer satisfaction and what types of factors are most critical at different stages in the process.

Development of Hypotheses

According to Oliver (1997), a satisfaction evaluation is a specific type of attitude (i.e., postconsumption evaluation). Therefore, to develop our hypotheses on the customer satisfaction formation process, we draw on research on attitudes. Research in this area has found that newly formed attitudes are unstable and held with less certainty and therefore are more difficult to predict (Eagly and Chaiken 1993). Certainty has been shown to increase as the quantity of information about the product increases (Berscheid et al. 1976; Dover and Olson 1977; Farley, Katz, and Lehmann 1978). Over repeated consumption experiences, customers gain more information about the product, and therefore their judgment certainty should increase (Fazio 1986; Smith and Swinyard 1983). Thus, with repeated experience, the cognitive and affective judgments underlying these attitudes become associated with higher certainty (i.e., become stronger and more stable) and should be better able to predict the satisfaction evaluation (Chandrashekaran et al. 2000). We hypothesize the following:

H₁: The variance in customer satisfaction jointly explained by cognitive and affective factors increases as experience accumulates.

As we mentioned previously, a key contribution of this article is its examination of whether the impact of cognition and affect on the satisfaction judgment changes over time. In particular, we are interested in the early stages of the satisfaction formation process (i.e., the first consumption experiences with the product or service). Whenever customers encounter a new product or service, there is typically little information to draw on from memory (i.e., no affect- or cognition-based evaluations). Therefore, customers are more likely to rely on their feelings because feelings are often elicited immediately on exposure to a new stimulus (Pham et al. 2001). In other words, affective responses to stimuli are evoked much quicker than cognitive responses. Consistent with this view, several studies have examined a feelings-based inference called the “how-do-I-feel-about-it” heuristic, whereby customers can make satisfaction or dissatisfaction judgments based on the valence of their feelings (Gorn, Goldberg, and Basu 1993; Pham 1998; Schwartz and Clore 1988).

This notion is also related to strategies outlined in the affect infusion model (AIM; Forgas 1994, 1995), in which affect infusion is highly likely to occur. Affect infusion refers to the process in which affectively loaded information influences and becomes part of the judgmental process, entering the judge's constructive thought process and eventually coloring the judgmental outcome (Forgas 1995, p. 39). The AIM is a cognitively oriented approach for describing the role of affect in judgment processes and thus fits well with the decision context of the current study. The AIM suggests that it is the constructive, generative nature of most judgments that is critical for affect infusion to occur (Fiedler 1991). Constructive processing involves the development of a judgment or decision rule on the spot rather than the reliance on a well-formed rule or judgment (Bettman, Luce, and Payne 1998).

Applied to the current context, for the first consumption experience, customers should have little information stored in their memory on which to base a judgment, and therefore they should rely on affective inputs. As the number of consumption experiences increases, however, customers acquire increasing amounts of information about the product or service (i.e., through trial and other sources). When making a satisfaction judgment on subsequent occasions, customers can recall this prior information to help them make a satisfaction judgment. In support of this notion, studies have found that previous satisfaction judgments affect subsequent satisfaction evaluations (Bolton 1998; Bolton and Drew 1991; Boulding et al. 1993; Mittal, Kumar, and Tsiros 1999). This notion is consistent with the direct access strategy of the AIM model, which involves the strongly cued retrieval of a crystallized judgment from memory, and is a robust process more resistant to affect infusion.

Thus, we expect that affect has a greater impact on the satisfaction judgment in the earlier stages of the satisfaction development process when judgments require a higher degree of constructive processing. In support of this notion, Smith and Bolton (2002) find that affect is more likely to predict transaction-specific customer satisfaction than cumulative satisfaction (which would be based on several consumption experiences). Thus, we offer the following hypothesis:

H₂: As experience accumulates, (a) the impact of affective factors on customer satisfaction decreases, and (b) the impact of cognitive factors on customer satisfaction increases.

Further refining our theoretical reasoning, we propose that the previously mentioned pattern of processing is contingent on the consistency of the consumption experience over time. Thus, we investigate the impact of consistent versus inconsistent performance experiences on the proposed relationships in H₂. A consistent experience would result from performance feedback that is in the same direction for each consumption occasion (i.e., all consumption experiences are positive, or all are negative). This leads to the building of a more reliable and diagnostic knowledge base that can be used as a basis for evaluation on subsequent occasions. In support of this notion, studies have found that as consistency of information increases, attitude certainty also increases (Heslin, Blake, and Rotton 1972; Kahneman and Tversky 1973). Therefore, we expect that customers switch in early stages from a more constructive processing strategy to a direct access strategy.

Conversely, inconsistent experiences would generate conflicting information, which would result in less reliable and less diagnostic information stored in memory. Therefore, customers would need to continue engaging in constructive information processing, which, as we mentioned previously, involves the significant role of affective factors. Thus, we assume that the relative impact of affective factors does not decrease as heavily as in the case of consistent experiences.

H₃: The suggested pattern (i.e., the impact of cognition increases, whereas the impact of affect decreases) is more pronounced in the case of consistent experiences than in the case of inconsistent experiences.

A Longitudinal Experimental Study

Overview

To enhance the realism of the study, we placed the “customers” in an actual consumption situation. Specifically, we asked them to evaluate a newly created CD-ROM tutorial, which could be used and purchased to provide academic assistance in a difficult pricing class. We gave them three sample chapters (trials) of the CD-ROM tutorial over time in a computer-based format. After each trial, we asked them to solve a pricing problem related to the sample chapter, and then we gave them feedback on their performance. We manipulated performance by varying the quality of the tutorial and the feedback. After we provided the feedback, we elicited responses to key measures.

Sample

Our sample consisted of 157 marketing students enrolled in a graduate-level pricing class at a large German university. Because students represent the target market for the CD-ROM study guide, this sample is particularly relevant and appropriate. Participants were aware that previous students had experienced difficulties in this pricing class.

Research Design

The research design consisted of an 8 (levels of performance) × 3 (trial) full factorial design. Performance was a between-subjects factor, and trial was a within-subjects factor. To manipulate performance, we gave participants a sample chapter and asked them to solve a related pricing problem. In the high-performance condition, the content of the CD-ROM sample chapter made it easy to understand and to solve the pricing problem. This was accomplished by providing the different steps necessary to complete the pricing task successfully in a logical order. Furthermore, participants in this condition received positive feedback that the CD-ROM helped them solve the problem. Note that this feedback was designed to ensure that participants focused on the performance of the CD-ROM rather than on their own performance.

In the low-performance condition, the content of the CD-ROM chapter was not in logical order and therefore was difficult to read. It also contained little information related to the pricing problem. We also provided participants with negative feedback that the CD-ROM did not help them solve the pricing problem. Again, the goal was to focus participants on the performance with the CD-ROM rather than on their own performance.

To examine the proposed dynamic relationships, we manipulated performance across three different trials. In each trial, we presented participants with a different chapter from the CD-ROM tutorial and then asked them to solve a related pricing problem. Note that the three sample chapters covered completely different pricing issues and did not build on one another. Thus, learning effects, in the sense that knowledge gained in previous chapters could be helpful for understanding subsequent chapters and pricing tasks, were not present. In the high-performance condition, participants received the high-performance manipulation on all three trials. In the low-performance condition, participants received the low-performance manipulation on all three trials. These two conditions represent consistent performance experiences. We then manipulated intermediate levels of performance through high- and low-performance combinations across the three trials. To represent inconsistent experiences, we constructed two conditions with a performance experience on each subsequent trial that was different from the most recent one (+ − + and − + −). We outline the different conditions in Table 1.

Table 1

Overview of Experimental Conditions

Experimental Condition	Trial
Experimental Condition	1	2	3
1	+	+	+
2	+	+	–
3	+	–	+
4	+	–	–
5	–	+	+
6	–	+	–
7	–	–	+
8	–	–	–

Notes: + = high performance, and − = low performance.

Experimental Procedure

Before the first trial, we gave participants the introductory section, which described the purpose of the CD-ROM and content. We told them that a CD-ROM study guide had been developed to assist participants in solving difficult pricing problems in the course. Furthermore, we informed them that they would have the chance to test the CD-ROM tutorial before deciding if they wanted to purchase it. We then set up expectations about the CD-ROM tutorial and held them constant across the experimental conditions. We told participants that the CD-ROM tutorial contained 73 chapters, which would be similar to the ones they received in the testing phase but would cover different pricing topics. The introductory section of the CD-ROM tutorial informed participants that the purpose of the study guide was to help course participants understand difficult material in the class, and it provided an overview of the content.

Participants then received one sample chapter of the CD-ROM study guide (first trial), after which they solved a problem related to the material. We manipulated performance in the manner described previously. The measurement of cognition and affect occurred immediately after participants received the feedback from the performance manipulation. We then asked intervening questions to distract participants from the evaluation situation, after which we measured customer satisfaction. After the first trial, participants completed an intervening task. This involved reading a newspaper article about a recent pricing problem in practice and answering some open-ended questions about the content of the article.

We conducted the second and third trials in a similar manner. We told participants that they had the chance to get more experience with the CD-ROM study guide. They then received another sample chapter of the CD-ROM, solved a pricing task, and received feedback. After we measured the key variables, we distracted participants between the second and the third trial with an intervening task similar to the one described previously. After the third trial, participants completed a final set of general questions and provided demographic and classification information.

Measurement of Variables

The key variables of interest were cognition and affect. We measured cognition in terms of disconfirmation, which is the key cognitive component in the confirmation/disconfirmation paradigm (Oliver 1980). We measured disconfirmation with the following item, which parallels previous approaches in measuring disconfirmation (Fornell et al. 1996): “The performance of the CD-ROM meets my expectations.” We evaluated the item on a seven-point Likert-type scale anchored by “I totally disagree” (1) and “I totally agree” (7).

We measured affect with three items. These items included elation, delight, and joy, which have also been measured in other studies examining affect (e.g., Holbrook and Batra 1987; Oliver 1993; Westbrook 1987; Westbrook and Oliver 1991). We asked participants how intensively they experienced these types of affect on a seven-point Likert-type scale.

In addition, we assessed affect and cognition with an open-ended, verbal protocol measures. We asked participants to describe any feelings and/or thoughts they had while using the tutorial. Responses were then coded into affective and cognitive categories. Responses were coded as affective if the statements related to any type of affect or feelings; they were coded as cognitive if they related to the content and functional performance of the tutorial. Research assistants carried out the codings. In terms of coder reliability, the coders agreed in 96% of the cases, and they resolved disagreements through discussions.

The key dependent variable was customer satisfaction, which we measured with the following three items: “All in all, I am satisfied with the CD-ROM tutorial”; “The CD-ROM tutorial compares to an ideal CD-ROM tutorial”; and “Overall, how satisfied are you with the CD-ROM tutorial?” We measured the items on an 11-point Likert-type scale. For the first two items, the scale ranged from “strongly agree” to “strongly disagree,” and for the third item, it ranged from “very satisfied” to “very dissatisfied.” The internal consistency of the satisfaction scale was excellent across the three trials (first trial: Cronbach's α = .95; second trial: Cronbach's α = .97; and third trial: Cronbach's α = .96). Thus, for further analyses, we calculated the satisfaction scores as the means of the satisfaction scale items.

Finally, we collected several variables as possible covariates. These included participants’ age, gender, income, degree to which the budget for studying material is exhausted, perceived pressure to buy the CD-ROM tutorial, price consciousness, value consciousness, and self-confidence. Analyses indicate that the fixed effects of these covariates are not significant.¹ Thus, we dropped them from further analysis.

The individual significance levels were as follows: age: p = .485; gender: p = .140; income: p = .825; degree to which the budget for studying material is exhausted: p = .354; perceived pressure to buy the CD-ROM tutorial: p = .563; price consciousness: p = .223; value consciousness: p = .295; and self-confidence: p = .103.

Results

H₁ predicted that the variance in customer satisfaction jointly explained by cognitive and affective factors increases as experience accumulates. As participants gain more experience with the CD-ROM tutorial across the three trials, the satisfaction judgment becomes more cumulative. To test H₁, we estimated the following regression model for each of the three trials:

CS = b_{0} + b_{1} \times CO + b_{2} \times AF + r,

(1)

where CS is customer satisfaction, CO is the cognitive predictor, AF is the affective predictor, and r is the random error.

We estimated the three regression models with the maximum likelihood method using the procedure MIXED in SAS 8.02.² The results appear in Table 2. First, in each of the three trials, there is a positive and statistically significant relationship between the cognitive predictor and satisfaction, and there is a positive and statistically significant relation between the affective predictor and satisfaction. Second, the results indicate that the R-square values substantially increase from the first (.645) to the third (.731) trial. These results support H₁.

We used full maximum likelihood estimation because the focus of the analysis is on deviance tests and not on the random part parameters, for which the restricted maximum likelihood method is preferable (Singer and Willett 2003; Snijders and Bosker 1999).

Table 2

Regression Results Across Trials

		Customer Satisfaction Trial 1			Customer Satisfaction Trial 2			Customer Satisfaction Trial 3
−2 log-likelihood		940.9			525.8			499.7
R²		.645			.706			.731
		Solutions for Fixed Effects
Parameter	Effect	Estimate	t-Value	p	Estimate	t-Value	p	Estimate	t-Value	p
b₀	Intercept	6.318	58.940	.000	6.074	58.950	.000	5.823	61.400	.000
b₁	Cognition	.682	10.510	.000	.787	12.890	.000	.849	15.870	.000
b₂	Affect	.353	6.060	.000	.303	5.160	.000	.171	3.340	.001
		Solutions for Residual Effects
Parameter		Estimate	Z Value	p	Estimate	Z Value	p	Estimate	Z Value	p
σ² (variance of r_ij)		1.804	8.800	.000	1.667	8.860	.000	1.412	8.860	.000

Our second analysis pertained to the changing role of cognitive and affective factors over time. H_2a predicted that the impact of affective factors decreases as experience accumulates, and H_2b predicted that the role of cognitive factors increases as experience accumulates. To conduct a test of these hypotheses, we estimated the following regression model with interaction effects:

\begin{array}{l} {CS}_{i} = b_{0} + b_{1} \times {CO}_{i} + b_{2} \times {AF}_{i} + b_{3} \times {time}_{i} \\ + b_{4} \times {CO}_{i} \times {time}_{i} + b_{5} \times {AF}_{i} \times {time}_{i} + r_{i}, \end{array}

(2)

where CS_i is customer satisfaction, CO_i is cognition, and AF_i is affect at time_i. The different points in time are represented by the different trials (e.g., trial 1 = time 1). The random error associated with time_i is represented by r_i.

In the testing of H_2, the coefficients of the two interaction terms, b₄ and b₅, are of key interest. These interaction terms are similar to slope-shift parameters (Mittal, Katrichis, and Kumar 2001), which enable us to estimate the differential slopes over time as experience accumulates. A significant, positive coefficient for the interaction term implies that the slope is higher for those who have more experience with the CD-ROM. In other words, the corresponding factor has a larger weight for customers who have more experience with the CD-ROM than for those who have less experience with it. On the basis of our theoretical reasoning, we expect the coefficient b₄ to be positive.

A negative coefficient for the interaction term implies that the slope is lower for those who have more experience with the CD-ROM. The corresponding factor has a smaller weight for customers who have more experience with the CD-ROM than for those who have less experience with it. On the basis of our theoretical reasoning, we expect the coefficient b₅ to be negative. Furthermore, statistically nonsignificant coefficients imply that the weight of the corresponding factor is invariant to the experience a customer has with the CD-ROM. We estimated the model with the maximum likelihood method using the procedure MIXED in SAS 8.02. The estimation results appear in Table 3.

Table 3

Regression Results with Interaction Effects

−2 log-likelihood	1566.9
		Solution for Fixed Effects
Parameter	Effect	Estimate	t-Value	p
b₀	Intercept	6.072	103.200	.00
b₁	Cognition	.773	22.380	.00
b₂	Affect	.275	8.490	.00
b₃	Time	−.248	–3.440	.00
b₄	Cognition × time	.083	1.960	.05
b₅	Affect × time	−.091	–2.330	.02
		Solution for Residual Effect
Parameter		Estimate	Z Value	p
σ²		1.630	15.350	.00

The results indicate that there is an overall statistically significant, positive relationship between cognition and satisfaction (b₁ = 22.380, p < .001) and between affect and satisfaction (b₂ = 8.490, p < .001). In addition, there is a significant, positive interaction between cognition and time (b₄ = 1.960, p < .05). This means that the impact of cognitive factors on the satisfaction judgment increases over time. Thus, there is support for H_2a.

Furthermore, there is a significant, negative interaction between affect and time (b₅ = −2.330, p < .05), supporting the notion that the impact of affective factors on customer satisfaction decreases over time. Thus, there is also support for H_2b. Taking these two findings together, the impact of cognitive factors on the satisfaction judgment increases over time, whereas the impact of affective factors decreases.

Further support for these hypotheses is provided by examining the results for the verbal protocol measures. As we mentioned previously, open-ended responses were coded into affective and cognitive categories. Overall, there were more cognitive statements than affective ones. We expected this because the product category is cognitive in nature. More important, however, the pattern of responses over the trials was consistent with the hypotheses. Specifically, the largest number of emotional statements was made in the first trial (23) and then decreased (14 in both the second and the third trials). For cognitive statements related to the content and functional performance, participants exhibited an increasing trend. On the first trial, 74 such statements were made, followed by 89 in the second trial and 90 in the third.

H₃ proposed that an inconsistent performance experience would attenuate the increasing impact of cognitive factors and the decreasing impact of affective factors over time (compared with a consistent performance experience). To test this hypothesis, we created two subgroups from the data set. The first group had consistent satisfaction/performance experiences (+ + + or − − −), whereas the second group had completely inconsistent performance experiences (+ − + or − + −).

We then ran the regression model with interaction effects described previously for each group separately. As we show in Table 4, for the consistent performance experiences, the impact of cognition increases slightly, and the impact of affect decreases over time, as is evidenced by the results for the interaction effects (b₄ = .119, p < .06; b₅ = -.156, p < .05). In the case of inconsistent performance experiences, however, neither of these interaction terms is significant (see Table 4). In other words, in the case of inconsistent performance experiences, there is neither a significant increase of cognition over time nor a significant decrease of affect over time in predicting customer satisfaction. Thus, inconsistent performance experiences attenuate these effects. These results support H₃.

Table 4

Regression Results for Consistent and Inconsistent Performance Experience

		Consistent Performance Experience			Inconsistent Performance Experience
−2 log-likelihood		449.5			422.8
		Solution for Fixed Effects			Solution for Fixed Effects
Parameter	Effect	Estimate	t-Value	p	Estimate	t-Value	p
b₀	Intercept	6.095	59.660	.00	6.268	48.320	.00
b₁	Cognition	.742	14.610	.00	.833	10.450	.00
b₂	Affect	.335	6.600	.00	.249	8.490	.00
b₃	Time	−.197	–1.590	.12	−.119	3.220	.45
b₄	Cognition × time	.119	1.870	.06	.048	−.750	.62
b₅	Affect × time	−.156	–2.500	.01	.111	1.200	.23
		Solution for Residual Effect			Solution for Residual Effect
Parameter		Estimate	Z Value	p	Estimate	Z Value	p
σ²		1.419	15.350	.00	1.9844	15.350	.00

Discussion

The goal of this study was to examine the joint effects of cognitive and affective factors on satisfaction judgments in a dynamic setting. The results of our study show that together, cognitive and affective factors explain the variance in satisfaction judgments well and that the strength of association increases over time. More important, we found that as the number of experiences increases over time, the influence of cognitive factors increases, whereas the influence of affective factors decreases. However, this effect attenuates with inconsistent consumption experiences. These findings have several theoretical and managerial implications.

Implications

Several implications follow from our study. Most important, our study contributes to the existing literature on customer satisfaction. First, we illustrate the importance of examining customer satisfaction from a dynamic perspective. We show that the antecedents leading to customer satisfaction vary over time. As we mentioned previously, it is well recognized that both cognition and affect have important influences on the satisfaction judgment. However, the key finding of this study points to the differential roles of these factors as experience accumulates. To the best of our knowledge, this is the first study to examine the role of cognition and affect simultaneously in a dynamic setting. Thus, we provide a deeper integrative understanding of how cognition and affect combine to form the customer satisfaction judgment. Our findings show that these judgments are based on different factors, depending on the amount of experience that has been accumulated over time. Thus, capturing satisfaction judgments at only one time may not provide a full picture of the underlying process.

Second, most studies in this area examine in a cross-sectional manner the satisfaction judgment process after customers already have some experience or knowledge with the product or service. In the current study, we investigated the customer satisfaction formation process beginning with the very first consumption experience. This provides a deeper insight into how the satisfaction judgment crystallizes over time. In this context, we found that affective factors have their strongest role in the early stages. The cognitive component maintains its importance and increases its role over time.

Third, we examine the role of cognition and affect in influencing satisfaction judgment processes. Our findings suggest that affect is particularly important in the early stages of the judgment formation process in which customers have little knowledge or experience related to the product. In other words, when a stimulus is new and unfamiliar, feelings appear to play a critical role in the construction of satisfaction judgments, consistent with the how-do-I-feel-about-it heuristic (Gorn, Goldberg, and Basu 1993; Pham 1998; Schwartz and Clore 1988). As experience accumulates, however, the role of cognition becomes more prominent. That is, a direct access strategy (Forgas 1994, 1995) can be employed in which evaluations from previous trials (based on a more detailed evaluation of product attributes) can be recalled from memory.

Fourth, we provide insights into how the inconsistency versus consistency of performance feedback can also affect the customer satisfaction formation process. Specifically, when experiences are inconsistent, the related judgments are less stable and predictable. In this case, the changing patterns of affect and cognition do not emerge. An implication of our findings is that affective factors are particularly important and relevant when the judgments are based on little information (i.e., they are newly formed) or when information is inconsistent. In these cases, the cognitive information has unstable predictive validity, so customers are less likely to rely on this information. This finding emphasizes the importance of identifying how the pattern or nature of experiences can influence satisfaction judgments as well as the importance of managing customer experiences. If customers have inconsistent experiences, their evaluations of the product/service will be less crystallized and less stable. Thus, in an effort to improve customer satisfaction and loyalty, it is critical to ensure that customer experiences are positive and consistent.

Fifth, the findings provide further support for research on attitude strength in an alternative context (Jaccard et al. 1995; Jacoby et al. 2002; Petty and Krosnick 1995). That is, cognitive and affective factors were better able to predict satisfaction evaluations as the number of trials increased, particularly in the case of consistent experiences. This is in line with the notion that newly formed attitudes are less stable and, therefore, less predictive of information processing and behavior (Eagly and Chaiken 1993). As knowledge and experience accumulates, the relationship between the cognitive and affective factors and satisfaction is strengthened.

Sixth, our study also makes a methodological contribution by developing a procedure for studying these dynamic effects in a tightly controlled, yet realistic setting. This methodology could be employed to examine judgment formation processes in various other product/service contexts as well.

Finally, on a general level, our study suggests that it is insightful to investigate marketing phenomena from a dynamic perspective. As our study shows, judgment patterns change over time.

Limitations and Avenues for Further Research

Several limitations of this study are worth mentioning. In turn, these provide avenues for further research. First, the study was based on a highly utilitarian product (i.e., a CD-ROM tutorial). In general, the consumption of utilitarian goods is highly cognitively driven and accomplishes a functional or practical task (Strahilevitz and Myers 1998). This may partially explain why the role of cognition (affect) became stronger (weaker) over trials. In other words, as customers gained experience in evaluating the utilitarian benefits of the product, these cognitive factors gained importance in influencing satisfaction. However, it is not clear whether the same pattern would occur for highly hedonic products/services. The consumption of hedonic products/services is characterized by highly affective and sensory experiences of sensual and/or aesthetic pleasure (Dhar and Wertenbroch 2000; Hirschman and Holbrook 1982; Okada 2005). Thus, in such cases, the impact of affect might not decline over time. It is possible that a direct access strategy will still be employed as experience accumulates (i.e., a recall of a previous crystallized evaluation from memory), but this prior evaluation would be based more on affective than on cognitive factors. In support of this view, Kempf (1999) finds that for hedonic products in particular, affective responses to the trial are powerful antecedents of consumers’ evaluations of a trial experience and, subsequently, brand attitude. In addition, it might be the case that the relative importance of cognitive versus affective drivers of customer satisfaction varies across product categories. It is plausible to assume that the relative importance of cognitive factors is higher for utilitarian products than for hedonic ones. These speculations, which are beyond the scope of this article, show that a more detailed investigation of affective and cognitive drivers of customer satisfaction that accounts for different product categories is a fruitful avenue for further research.

Second, in addition to the product category type, customers’ product involvement could also play an important role in this context. Product involvement can be defined as the perceived personal relevance of a product and a customer's inherent interest in a product (Wilkie 1994; Zaichkowsky 1985). This study investigated the proposed relationships in a high-involvement context. In these situations, customers usually employ a more analytical information-processing strategy and exhibit increased cognitive elaborations (Petty, Cacioppo, and Schumann 1983; Wright 1973). Thus, they are likely to base their decisions more on cognitive than on affective factors. Conversely, in low-involvement situations, information processing is usually more holistic, and thus decisions might be influenced more by affective factors. Apart from these effects, it might be interesting to investigate whether and how these potential effects of customers’ product involvement interact with the product category type (hedonic and utilitarian). Thus, studying the relevance of involvement in the context of affective and cognitive drivers of customer satisfaction is another interesting area for further research.

Third, in our empirical study, we measured only the prevalence of positive affect and did not consider negative affect. Because it has been shown that positive and negative affect can have asymmetric effects on customers’ judgments and choices (Mittal, Ross, and Baldasare 1998), a promising avenue for further research might be to investigate positive and negative affect simultaneously. Prior research also demonstrates that different types of (negative) emotions can have opposing moderating effects (Garg, Inman, and Mittal 2005). Thus, additional research on the role of cognition and affect in the formation of customer satisfaction might further differentiate between different types of emotions.

Implications for Marketing Practice

Our findings also have several important implications for marketing managers. Our study helps managers understand customer satisfaction in a more thorough way. It sheds light on the formation process of customer satisfaction, and managers learn that customer satisfaction has a more stochastic character in the early stages. This gives managers the opportunity to influence the satisfaction judgment to a greater extent in the early stages because the satisfaction has not been crystallized.

Furthermore, it is common in practice for managers to consider customer satisfaction in a logical, rational manner (i.e., if the product or service performs well, satisfaction will be higher). The results of our study point out that affective factors can play a critical role as well, particularly in the early stages of the satisfaction formation process. Thus, achieving customer satisfaction involves not merely disconfirmation judgments but a subjective, affective component as well. This point is particularly important when companies are in the early stages of establishing a relationship with a customer. Thus, for new relationships or new products, managers must pay close attention to affective aspects and be careful to manage them effectively.

Moreover, our study provides insights into managing customer satisfaction. For example, it shows how important it is to manage affect when customers have inconsistent experiences over time. Thus, if companies are trying to recover from a situation in which their products/services have suffered from inconsistent performance, it appears that it is important to manage the affective factors surrounding the satisfaction in addition to product/service quality considerations. In addition, providing customers with consistent, positive experiences is critical to the development of stable satisfaction judgments.

References

Alvero

Alicia M.

, Bucklin

Barbara R.

, and Austin

John

(2001), “An Objective Review of the Effectiveness and Essential Characteristics of Performance Feedback in Organizational Settings,” Journal of Organizational Behavior Management, 21 (1), 3–29.

Anderson

Eugene W.

, Fornell

Claes

, and Lehmann

Donald R.

(1994), “Customer Satisfaction, Market Share, and Profitability: Findings from Sweden,” Journal of Marketing, 58 (July), 53–66.

Anderson

Eugene W.

, Fornell

Claes

, and Rust

Roland T.

(1997), “Customer Satisfaction, Productivity, and Profitability: Differences Between Goods and Services,” Marketing Science, 16 (2), 129–45.

Anderson

Eugene W.

, and Sullivan

Mary W.

(1993), “The Antecedents and Consequences of Customer Satisfaction for Firms,” Marketing Science, 12 (2), 125–43.

Bearden

William O.

, and Teel

Jesse E.

(1983), “Selected Determinants of Consumer Satisfaction and Complaint Reports,” Journal of Marketing Research, 20 (February), 21–28.

Berscheid

Ellen

, Graziano

William

, Monson

Thomas

, and Dermer

Marshall

(1976), “Outcome Dependency: Attention, Attribution, and Attraction,” Journal of Personality and Social Psychology, 34 (3), 978–89.

Bettman

James R.

, Luce

Mary Frances

, and Payne

John W.

(1998), “Constructive Consumer Choice Processes,” Journal of Consumer Research, 25 (3), 187–217.

Bolton

Ruth N.

(1998), “A Dynamic Model of the Duration of the Customer's Relationship with a Continuous Service Provider: The Role of Satisfaction,” Marketing Science, 17 (1), 45–65.

Bolton

Ruth N.

, and Drew

James H.

(1991), “A Longitudinal Analysis of the Impact of Service Changes on Customer Attitudes,” Journal of Marketing, 55 (January), 1–9.

10.

Bolton

Ruth N.

, and Lemon

Katherine N.

(1999), “A Dynamic Model of Customers’ Usage of Services: Usage as an Antecedent and Consequence of Satisfaction,” Journal of Marketing Research, 36 (May), 171–86.

11.

Boulding

William

, Kalra

Ajay

, Staelin

Richard

, and Zeithaml

Valarie A.

(1993), “A Dynamic Process Model of Service Quality: From Expectations to Behavioral Intentions,” Journal of Marketing Research, 30 (February), 7–27.

12.

Chandrashekaran

Murali

, McNeilly

Kevin

, Russ

Frederick A.

, and Marinova

Detelina

(2000), “From Uncertain Intentions to Actual Behavior: A Threshold Model of Whether and When Salespeople Quit,” Journal of Marketing Research, 37 (November), 463–79.

13.

Dhar

Ravi

, and Wertenbroch

Klaus

(2000), “Consumer Choice Between Hedonic and Utilitarian Goods,” Journal of Marketing Research, 37 (February), 60–71.

14.

Dover

Philip A.

, and Olson

Jerry C.

(1977), “Dynamic Changes in an Expectancy-Value Attitude Model as a Function of Multiple Exposures to Product Information,” in Contemporary Marketing Thought: Proceedings of the 1977 AMA Conference, Greenberg

Barnett A.

, and Bellenger

Danny N.

, eds. Chicago: American Marketing Association, 455–60.

15.

Eagly

Alice H.

, and Chaiken

Shelly

(1993), The Psychology of Attitudes. Fort Worth, TX: Harcourt Brace Jovanovich.

16.

Farley

John U.

, Katz

Jerrold P.

, and Lehmann

Donald R.

(1978), “Impact of Different Comparison Sets on Evaluation of a New Subcompact Car Brand,” Journal of Consumer Research, 5 (September), 138–42.

17.

Fazio

Russel H.

(1986), “How Do Attitudes Guide Behavior?” in Handbook of Motivation and Cognition: Foundations of Social Behavior, Sorrentino

Richard M.

, and Higgins

Edward T.

, eds. Chichester: John Wiley & Sons, 204–243.

18.

Fiedler

Klaus

(1991), “On the Task, the Measures and the Mood: Research on Affect and Social Cognition,” in Emotion and Social Judgments, Forgas

Joseph P.

, ed. Elmsford, NY: Pergamon Press, 83–104.

19.

Forgas

Joseph P.

(1994), “The Role of Emotion in Social Judgments: An Introductory Review and an Affect Infusion Model (AIM),” European Journal of Social Psychology, 24 (1), 1–24.

20.

Forgas

Joseph P.

(1995), “Mood and Judgment: The Affect Infusion Model (AIM),” Psychological Bulletin, 117 (1), 39–66.

21.

Fornell

Claes

(1992), “A National Customer Satisfaction Barometer: The Swedish Experience,” Journal of Marketing, 56 (January), 6–21.

22.

Fornell

Claes

, Johnson

Michael D.

, Anderson

Eugene W.

, Cha

Jaesung

, and Bryant

Barbara Everitt

(1996), “The American Customer Satisfaction Index: Nature, Purpose, and Findings,” Journal of Marketing, 60 (October), 7–18.

23.

Fournier

Susan

, and Mick

David G.

(1999), “Rediscovering Satisfaction,” Journal of Marketing, 63 (October), 5–23.

24.

Garg

Nitika

, Inman

J. Jeffrey

, and Mittal

Vikas

(2005), “Incidental and Task-Related Affect: A Re-Inquiry and Extension of the Influence of Affect on Choice,” Journal of Consumer Research, 32 (June), 154–59.

25.

Gorn

Gerald J.

, Goldberg

Marvin E.

, and Basu

Kunal

(1993), “Mood, Awareness, and Product Evaluation,” Journal of Consumer Psychology, 2 (3), 237–56.

26.

Heslin

Richard

, Blake

Brian

, and Rotton

James

(1972), “Information Search as a Function of Stimulus Uncertainty and the Importance of the Response,” Journal of Personality and Social Psychology, 23 (3), 333–39.

27.

Hirschman

Elizabeth C.

, and Holbrook

Morris B.

(1982), “Hedonic Consumption: Emerging Concepts, Methods, and Propositions,” Journal of Marketing, 46 (Summer), 92–101.

28.

Holbrook

Morris B.

, and Batra

Rajeev

(1987), “Assessing the Role of Emotions as Mediators of Consumer Responses to Advertising,” Journal of Consumer Research, 14 (December), 404–420.

29.

Jaccard

James

, Radecki

Carmen M.

, Wilson

Tracey E.

, and Dittus

Patricia J.

(1995), “Methods for Identifying Consequential Beliefs: Implications for Understanding Attitude Strength,” in Attitude Strength: Antecedents and Consequences, Petty

Richard E.

, and Krosnick

Jon A.

, eds. Mahwah, NJ: Lawrence Erlbaum Associates, 337–59.

30.

Jacoby

Jacob

, Morrin

Maureen

, Jaccard

James

, Gurhan

Zeynep

, Kuss

Alfred

, and Maheswaran

Durairaj

(2002), “Mapping Attitude Formation as a Function of Information Input: Online Processing Models of Attitude Formation,” Journal of Consumer Psychology, 12 (1), 21–34.

31.

Kahneman

Daniel

, and Tversky

(1973), “On the Psychology of Prediction,” Psychological Review, 80 (4), 237–51.

32.

Kempf

DeAnna S.

(1999), “Attitude Formation from Product Trial: Distinct Roles of Cognition and Affect for Hedonic and Functional Products,” Psychology & Marketing, 16 (1), 35–50.

33.

LaBarbera

Pricilla A.

, and Mazursky

David

(1983), “A Longitudinal Assessment of Consumer Satisfaction/Dissatisfaction: The Dynamic Aspect of the Cognitive Process,” Journal of Marketing Research, 20 (November), 393–404.

34.

Mano

Haim

, and Oliver

Richard L.

(1993), “Assessing the Dimensionality and Structure of the Consumption Experience: Evaluation, Feeling, and Satisfaction,” Journal of Consumer Research, 20 (3), 451–67.

35.

Mittal

Vikas

, Katrichis

Jerome M.

, and Kumar

Pankaj

(2001), “Attribute Performance and Customer Satisfaction over Time: Evidence from Two Field Studies,” Journal of Services Marketing, 15 (5), 343–56.

36.

Mittal

Vikas

, Kumar

Pankaj

, and Tsiros

Michael

(1999), “Attribute-Level Performance, Satisfaction, and Behavioral Intentions over Time: A Consumption-System Approach,” Journal of Marketing, 63 (April), 88–101.

37.

Mittal

Vikas

, Ross

William T.

, and Baldasare

Patrick M.

(1998), “The Asymmetric Impact of Negative and Positive Attribute-Level Performance on Overall Satisfaction and Repurchase Intentions,” Journal of Marketing, 62 (January), 33–47.

38.

Okada

Erica M.

(2005), “Justification Effects on Consumer Choice of Hedonic and Utilitarian Goods,” Journal of Marketing Research, 42 (February), 43–53.

39.

Oliver

Richard L.

(1980), “A Cognitive Model of the Antecedents and Consequences of Satisfaction Decisions,” Journal of Marketing Research, 17 (November), 460–69.

40.

Oliver

Richard L.

(1993), “Cognitive, Affective, and Attribute Bases of Satisfaction Response,” Journal of Consumer Research, 20 (December), 418–30.

41.

Oliver

Richard L.

(1997), Satisfaction: A Behavioral Perspective on the Consumer. Boston: McGraw-Hill.

42.

Oliver

Richard L.

, and DeSarbo

Wayne S.

(1988), “Response Determinants in Satisfaction Judgments,” Journal of Consumer Research, 14 (March), 495–507.

43.

Oliver

Richard L.

, and Swan

John E.

(1989a), “Consumer Perceptions of Interpersonal Equity and Satisfaction in Transactions: A Field Survey Approach,” Journal of Marketing, 53 (April), 21–35.

44.

Oliver

Richard L.

, and Swan

John E.

(1989b), “Equity and Disconfirmation Perceptions as Influences on Merchant and Product Satisfaction,” Journal of Consumer Research, 16 (December), 372–83.

45.

Petty

Richard E.

, Cacioppo

John T.

, and Schumann

David T.

(1983), “Central and Peripheral Routes to Advertising Effectiveness: The Moderating Effect of Involvement,” Journal of Consumer Research, 10 (September), 135–46.

46.

Petty

Richard E.

, and Krosnick

Jon A.

(1995), Attitude Strength: Antecedents and Consequences. Mahwah, NJ: Lawrence Erlbaum Associates.

47.

Petty

Richard E.

, Unnava

H. Rao

, and Strathman

Alan J.

(1991), “Theories of Attitude Change,” in Handbook of Consumer Behavior, Robertson

Thomas S.

, and Kassarjian

Harold H.

, eds. Englewood Cliffs, NJ: Prentice Hall, 241–80.

48.

Pham

Michel Tuan

(1998), “Representativeness, Relevance, and the Use of Feelings in Decision Making,” Journal of Consumer Research, 25 (September), 144–60.

49.

Pham

Michel Tuan

, Cohen

Joel B.

, Pracejus

John W.

, and Hughes

G. David

(2001), “Affect Monitoring and the Primacy of Feelings in Judgment,” Journal of Consumer Research, 28 (September), 167–88.

50.

Rust

Roland T.

, Moorman

Christine

, and Dickson

Peter R.

(2002), “Getting Return on Quality: Revenue Expansion, Cost Reduction, or Both?” Journal of Marketing, 65 (October), 7–24.

51.

Rust

Roland T.

, and Zahorik

Anthony J.

(1993), “Customer Satisfaction, Customer Retention, and Market Share,” Journal of Retailing, 69 (2), 193–215.

52.

Schwartz

Nobert

, and Clore

Gerald L.

(1988), “How Do I Feel About It? Informative Functions of Affective States,” in Affect, Cognition, and Social Behavior, Fielder

Klaus

, and Forgas

Joseph

, eds. Toronto: Hogrefe International, 44–62.

53.

Singer

Judith D.

, and Willett

John B.

(2003), Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence. New York: Oxford University Press.

54.

Slotegraaf

Rebecca J.

, and Inman

J. Jeffrey

(2004), “Longitudinal Shifts in the Drivers of Satisfaction with Product Quality: The Role of Attribute Resolvability,” Journal of Marketing Research, 41 (August), 269–80.

55.

Smith

Amy K.

, and Bolton

Ruth N.

(2002), “The Effect of Customers’ Emotional Responses to Service Failures on Their Recovery Effort Evaluations and Satisfaction Judgments,” Journal of the Academy of Marketing Science, 30 (1), 5–23.

56.

Smith

Robert E.

, and Swinyard

William R.

(1983), “Attitude–Behavior Consistency: The Impact of Product Trial Versus Advertising,” Journal of Marketing Research, 20 (August), 257–67.

57.

Snijders

Tom A.B.

, and Bosker

Roel J.

(1999), Multilevel Analysis: An Introduction to Basic and Advanced Multilevel Modeling. London: Sage Publications.

58.

Stone

Dianna L.

, and Stone

Eugene F.

(1985), “The Effects of Feedback Consistency and Feedback Favorability on Self-Perceived Task Competence and Perceived Feedback Accuracy,” Organizational Behavior and Human Decision Processes, 36 (2), 167–85.

59.

Strahilevitz

Michal

, and Myers

John

(1998), “Donations to Charity as Purchase Incentives: How Well They Work May Depend on What You Are Trying to Sell,” Journal of Consumer Research, 24 (2), 434–46.

60.

Szymanski

David M.

, and Henard

David

(2001), “Customer Satisfaction: A Meta-Analysis of the Empirical Evidence,” Journal of the Academy of Marketing Science, 29 (1), 16–35.

61.

Westbrook

Robert A.

(1987), “Product/Consumption-Based Affective Responses and Postpurchase Processes,” Journal of Marketing Research, 24 (August), 258–70.

62.

Westbrook

Robert A.

, and Oliver

Richard L.

(1991), “The Dimensionality of Consumption Emotion Patterns and Consumer Satisfaction,” Journal of Consumer Research, 18 (June), 84–91.

63.

Wilkie

William L.

(1994), Consumer Behavior, 3d ed. New York: John Wiley & Sons.

64.

Wright

Peter L.

(1973), “The Cognitive Processes Mediating Acceptance of Advertising,” Journal of Marketing Research, 10 (February), 53–62.

65.

Zaichkowsky

Judith L.

(1985), “Measuring the Involvement Construct,” Journal of Consumer Research, 12 (December), 341–52.