Are there cognitive dissonance segments?

Abstract:

Cognitive Dissonance has interested consumer researchers for many years. Until recently, however, there has not been an accepted consumer dissonance scale. Using a recently developed scale, the present study examined the presence of dissonance segments in two customer samples, in which a significant minority of respondents experienced dissonance. Similar segments were found in both clusters, suggesting they may be common across a number of purchase contexts. Further investigation found some background differences, although a number of suggested relationships (e.g. price would affect dissonance) were not supported. Given the potential negative effects of dissonance, managers need to give this construct their attention.

Keywords:

COGNITIVE DISSONANCE; SEGMENTATION; CLUSTER ANALYSIS.

**********

1. Introduction

Cognitive Dissonance has been a controversial phenomenon, debated with frequency and considerable ferocity since Festinger s (1957) early research suggested dissonance could impact on people's decision-making processes, potentially affecting consumers' attitudes towards and satisfaction with their purchase decisions. One author even termed it one of social psychology's greatest theories (Aronson 1969). However, a review by Cummings and Venkatesan (1976) marked a reduction of interest in cognitive dissonance in marketing, despite their conclusion that 'the evidence in favour of dissonance theory in the consumer behaviour literature looks good' (p. 307).

Oliver (1997) argued that this reduction in interest was inexplicable and unfortunate, suggesting more research was needed, especially the development of an operational dissonance measure for use in consumer research. Indeed, dissonance has received renewed attention in recent literature and remains a topic of great interest in social psychology, as can be seen in Harmon-Jones and Mills' (1999) synthesis. Recent research by Sweeney, Hausknecht and Soutar (2000) developed a multidimensional measure of consumer dissonance, enabling researchers to address important managerial questions, such as whether all customers experience dissonance or whether dissonance is experienced differently by different groups of customers. The purpose of the present research was to use their suggested scale to address this issue.

Before discussing the present study in detail, a short review of the relevant literature is provided. Following this review, the study is discussed in detail, the results obtained are outlined and their implications are discussed.

2. Prior Research on Dissonance

2.1 Conceptualising Dissonance

Festinger (1957) described a person as being in a dissonant state if two elements in her/his cognition (that is, in her/his knowledge of her/himself, her/his behaviour, her/his feelings, desires, or in her/his knowledge of the world) are inconsistent. Cognitive dissonance may result when an opinion is formed or a decision taken when cognition and opinions direct us in different directions. However, Festinger (1957, p. 266) also seems to have intended an emotional conceptualisation, suggesting that, 'for some people, dissonance is an extremely painful and intolerable thing.'

Sweeney, Hausknecht and Soutar (2000, p. 383) noted, 'dissonance includes both cognitive aspects, as the title 'cognitive dissonance' implies, as well as an emotional dimension, as many definitions, including Festinger's original definition, imply'. Their 22-item consumer related scale included an emotional dimension and two cognitive dimensions, which were termed 'wisdom of purchase' and 'concern over the deal'. These three dimensions are well supported in the dissonance literature. The emotional dimension, defined as 'a person's psychological discomfort subsequent to the purchase decision' (Sweeney, Hausknecht & Soutar 2000, p. 380), reflects the anxiety related to the purchase, as noted by Festinger (1957) and Mowen (1995). Indeed, Menasco and Hawkins (1978) viewed dissonance as a form of anxiety arising from temporary situations such as purchase. Elliott and Devine (1994) also noted that previous research had focussed on the arousal component of dissonance and found that cognitive dissonance is experienced through psychological discomfort.

Sweeney, Hausknecht and Soutar's (2000, p. 380) 'wisdom of purchase' dimension was defined as 'a person's recognition after the purchase has been made that they may not have needed the product or may not have selected the appropriate one'. This dimension is consistent with several authors discussing the difficulty in the purchase decision. For example, Kassarjian and Cohen (1965) noted that, even after making a decision, an individual is often faced with uncertainty as to its wisdom (e.g. wondering if they should have bought product A when product B was also attractive). That is, a person's decision difficulty reflects the positive attributes of rejected alternatives compared to the negative attributes of the chosen alternative, giving rise to a logical inconsistency between cognitive elements (Festinger 1957). Menasco and Hawkins (1978), recognising this view of dissonance, measured cognitive dissonance arousal in terms of purchase decision difficulty.

The third dissonance dimension identified by Sweeney, Hausknecht and Soutar (2000, p. 380), which was termed 'concern over the deal', reflects 'a person's recognition after the purchase has been made that they may have been influenced against their own beliefs by sales staff'. This final dimension recognises the potential cognitive inconsistency flowing from a consumer's attitude change through the influence of a salesperson. This aspect of dissonance has as its basis the concept of forced compliance that suggests people can be 'forced' to behave in a manner inconsistent with their prior beliefs (Cummings & Venkatesan 1976). While the forced compliance approach to investigating dissonance has been common in social psychology, it has been argued as having little relevance in marketing, where the customer has the final power of decision (Oshikawa 1970). Nonetheless, there are situations in marketing where the forced compliance paradigm is relevant (Cummings & Venkatesan 1976). A forceful or pushy salesperson might bring this about. Indeed, Bell (1967) identified the persuasibility of the customer, or the extent to which they are influenced by a salesperson, as a major influence on the extent of dissonance experienced.

2.2 Variation in Types of Dissonance Experienced

Given the richness of the dissonance construct, it is possible that customers vary in the type and extent of dissonance experienced. For example, some customers may feel concerned over their choice but may not be concerned about a salesperson's influence, while some may experience high dissonance on cognitive and emotional aspects. Indeed some customers may not experience dissonance at all, as not all purchase decisions lead to dissonance. We can examine these aspects one at a time.

There are three well-known conditions for dissonance to arise. The decision needs to be important, irrevocable and voluntary (Cummings & Venkatesan 1976; Oliver 1997). These conditions are a matter of degree and are most apparent in extended problem-solving decisions, such as when purchasing major durables or appliances, which vary in terms of features and price across different brands (Menasco & Hawkins 1978; Oliver 1997).

Further, customers do not always experience dissonance (Bell 1967). People have differential thresholds for dissonance, in the same way as they experience different thresholds for anxiety (Oliver 1997) or stimulation due to environmental stimuli (Bitner 1992). Initial discussions with the senior staff of the retail stores where the main data collection took place illustrate this point. The following quotes reflect the views of the managers and two of the senior staff in these stores:

   'Before the sale this anxiety is normal. They will always mentally
   calculate ... will it fit? They have a mental picture in their mind.
   It is not so much the finances. Once they have bought it, they feel
   quite good with themselves. There is a quick changeover from when
   they pay to when they get to the car. The majority are very happy
   and there is no anxiety'.

   'Because they are handing over money ... a lady from England, she
   didn't want to part with the money. She wasn't going to rush into
   it. She had a calculation to do, with the money going out of her
   bank account'

   'I don't buy into the anxiety from a sales point of view. Otherwise
   you can lose the sale'

   'Some aren't sure. The husbands leave their wives and go out and
   have a smoke. They've got to go for a walk'

   'One guy recently broke into a sweat [at the point of deciding], he
   was a nervous type'

   'They are often relieved after making the decision. There's not too
   many who are like that [dissonant]. Most come in and are blase,
   matter of fact. They come in make a decision and it is all over and
   done with. Most are positive and excited.'

2.3 Integrity of Realistic Setting

In much previous research, the existence of dissonance has been manipulated in artificial and often trivial situations (e.g. Hunt 1970; Loscuito & Perloff 1967; Cohen & Goldberg 1970; Korgaonkar & Moschis 1982; Elliott & Devine 1994). Such situations reflect decision conflict rather than decision dissonance, since the conditions were not irrevocable, important or even volitational, due to the artificiality of the experiment. Further, the participants were not exposed to contradictory evidence after the decision was made (Korgaonkar & Moschis 1982). A further drawback of much dissonance research is that, consistent with the experimental approach that is most commonly used, the study of cognitive dissonance has largely focussed on students, who do not necessarily represent a population experiencing dissonance (e.g. Loscuito & Perloff 1967; Cohen & Goldberg 1970; Elliott & Devine 1994). The present study, in contrast, tested the existence of cognitive dissonance among real consumers in a realistic consumer setting immediately following a major purchase.

Thus, the purpose of the present study was to examine the variation in dissonance in a realistic setting with real-life consumers. Specifically, various groups or segments of consumers who experience dissonance differently were identified and profiled in terms of their post-purchase behaviours, perceptions and backgrounds. It was hoped that such information would enhance our understanding of the ways customers experience dissonance and assist managers to develop strategies to reduce dissonance, thus improving customers' buying process experience.

2.4 The Decision Phase

A better understanding of the dissonance development process can be gained by examining the psychological responses at various stages of the decision-making process. Consumers are concerned to a greater or lesser extent about decision making at all these stages.

In the pre-decision ('alpha') stage, the choice may exist between similarly attractive alternatives, resulting in predecision conflict (Oliver 1997). At this stage, no commitment has been made to the purchase and the consumer is free to select this or any other product. Nonetheless, the consumer, in assessing the product, can evaluate the degree of risk in the product, which is derived from pre-purchase uncertainty about the product or, more specifically, the possible negative consequences of a choice alternative and the uncertainty of their occurrence (Bauer 1960; Cox 1967; Dowling 1999). Both perceived risk and cognitive dissonance share many of the same characteristics (e.g. the importance of the decision, a lack of product information, a lack of familiarity with the product or brand and the price of the product). However, it is likely that perceived risk is an antecedent to cognitive dissonance--perceived

risk is generally seen as prior to purchase and cognitive dissonance is generally seen as a post purchase construct (Oliver 1997; Sweeney, Hausknecht & Soutar 2000). Finally, while consumers may recognize a level of risk and try to minimise it, for example by purchasing a well known brand, those who experience dissonance post purchase are driven to regain cognitive consistency that favours the purchased item over others that were considered. Cognitive dissonance is therefore also known as 'buyer's regret'.

In the 'beta' stage, immediately following the purchase decision, consumers deal with anticipated regret as they consider forgone alternatives that have been 'left behind' (Oliver 1997). The 'gamma' stage follows purchase but precedes product use and represents the reality of the purchase, since forgone alternatives are gone and there may be a concern that the product will not perform as expected (Oliver 1997). At this stage dissonance is maximized and precedes satisfaction formation (Sweeney, Hausknecht & Soutar 2000). The concept of dissonance addressed in the present research best fits this phase. In the 'delta' phase, that follows the use of a product, the concept of dissonance refers to concern over current and future performance. This is particularly relevant to products that involve continuing consumption, such as durables (Oliver 1997). The consumer may never be completely free of dissonance. Dissonance is not a condition that exists in extremes but is present to a greater or lesser extent at the various stages of decision-making.

The present study examined the dissonance concept in two separate samples of purchasers of furniture and purchasers of car stereo equipment. The approach taken and the results obtained are outlined in subsequent sections.

3. The Present Study

3.1 Measures

Dissonance was measured using the 22-item scale developed by Sweeney, et al. (2000). Latent constructs can be created from reflective or formative indicators (Diamantopoulos & Winklhoffer 2001). Indicators of reflective constructs are affected by the same underlying construct and, so, are parallel measures that covary to the extent that they measure the underlying construct. Formative indicators determine the construct they represent and, consequently, may not covary with the construct they determine. In the present case, the dissonance construct was developed assuming reflective indicators for the various dimensions of the construct. The scale includes three dissonance dimensions (emotional, wisdom of purchase and concern over the deal) that are measured after purchase but before use. The scale has been shown to have high reliability (with alpha coefficients ranging from 0.80 to 0.98 for the various dimensions) and good discriminant validity, and criterion-related validity with respect to perceived value, satisfaction and difficulty in judging the quality of the product (Sweeney, Hausknecht & Soutar 2000).

A number of related constructs, developed in prior research, were also measured within the study. These included perceived value (Sweeney & Soutar 2001), satisfaction with the decision to buy the product (Oliver 1997), the ease of evaluating the quality of the product (Arnthorsson, Berry & Urbany 1991), active loyalty behaviours and likely switching and complaining behaviours (Zeithaml, Berry & Parasuraman 1996). These constructs were measured in order to profile the segments and were part of a data set used in a wider study. Details of the various scales are shown in Appendix A. In addition, the price of the product (measured in dollars) and the length of time (days) since the purchase decision to the completion of the questionnaire were also collected, as were some typical personal background details (e.g. age and gender).

3.2 Data Collection and Samples

Data were collected from customers of two different types of stores selling durable goods. It is generally accepted that dissonance is more likely to occur for such goods as they are more likely to be important to the consumer, given their costs, and the decision is relatively irreversible (Menasco & Hawkins 1978; Oliver 1997). Customers purchasing from a furniture store (with two outlets) and a car stereo centre (with three outlets) were questioned immediately after they committed themselves to a purchase. Customers making a major purchase (worth more than $400) were approached and asked to fill in a self-completion questionnaire and return it in a reply-paid envelope within ten days. As an incentive, participants were included in a draw for a voucher redeemable at the store. The survey was conducted over a six-month period and all customers who made a major were asked to participate. The final sample sizes were 323 for the furniture store and 313 for the car stereo centre, providing response rates of 44% and 31% respectively. The same respondents were surveyed again at a later stage to measure their postpurchase evaluations and behaviours, such as perceived value and satisfaction. (1) Sample sizes for the second stage were 206 and 162 respectively.

To check for possible non-response bias, background information, such as area of residence and expenditure levels, were compared with company records for all customers over the same time period. Chi-square tests found no significant differences, suggesting the samples were representative of major purchasers during the research period, at least in terms of the background data collected in the present study.

3.3 Analysis

As the major purpose of the study was to examine the level and pattern of cognitive dissonance, cluster analysis was used to see whether there were sub-groups of consumers with similar dissonance profiles. 'Clustering problems arise in many different applications ... and there are many methods for finding clusters (, although) perhaps the most widely used is k-means clustering' (Kanungo, Mount, Netanyahu, Piatko, Silverman & Wu 2002). Indeed, k-means clustering is the most common way through which market segments are identified (Punj & Stewart 1983; Helsen & Green 1991; Arabie & Hubert 1994; Brusco, Cradit & Stahl 2002), often because sample sizes are relatively large (Green, Carmone & Kim 1990). Consequently, it was decided to use a k-means approach in the present study.

Since there are no formal tests of statistical significance in cluster analysis, the stability of these clusters was examined in a variety of ways, as suggested by Calantone and Sawyer (1978) and Soutar and McNeil (1991). In this case, the two samples were each split randomly into two halves and a series of separate analyses conducted on each half to determine their stability, as outlined in the results section. The clusters obtained at each store were then compared to see whether there were similar segments across the samples. Finally, differences between group members' backgrounds and consumption constructs were examined. The results of the various analyses are outlined in the following section.

4. The Results Obtained

4.1 A Basic Analysis of Dissonance Dimensions

Some descriptive statistics were initially computed to assess the characteristics of the three dissonance dimensions, as shown in table 1. As can be seen from the low mean values, there was relatively little dissonance across the sample as a whole, although the standard deviations showed that there was some variation across respondents. In particular, it seemed that the wisdom of the purchase dimension was the dominant dissonance dimension. Correlations between the dissonance dimensions varied between 0.31 and 0.68, with the highest in both samples being between emotional dissonance and the concern over the deal. The alpha reliabilities were high, demonstrating internally consistent dissonance scales that can be used with confidence. A series of t-tests confirmed that there were no significant differences for any of the three dissonance dimensions across store types (the probabilities of equal means in the two store types were all greater than 0.30).

However, examining sample means can mask the presence of subgroups. Since the purpose of the present study was to see whether there were groups of consumers who viewed dissonance differently, the potential presence of sub-groups or dissonance segments was examined.

4.2 Clustering of Dissonance Types

Howard and Harris's (1966) k-means clustering procedure (Smith 1988) was used to examine this issue as it has been found to provide useful clusters in a number of prior studies (e.g. Soutar, Ramaseshan & Molster 1994; Soutar & McNeil 1995; Soutar, McNeil & Maisey 1995). The data from each store were analysed separately, with the number of clusters being varied from two to eight to determine the most appropriate number of clusters in each case. Following Milligan and Mahajan (1980), the point biserial correlation was used to determine the most appropriate number of clusters.

The point biserial correlation is used to estimate the degree of relationship between a dichotomous scale and an interval (or ratio) scale (Brown 1988). In this case, the interval scale is the distance between every pair of observations while the dichotomous scale is a corresponding vector of zeroes (when the two observations are in the same cluster) and ones (when the two observations are not in the same cluster). Since pairs of observations with large distances should not be in the same cluster, while pairs of observations with small distances should be in the same cluster, a high correlation implies a logical grouping. Milligan and Mahajan (1980) argued that the best solution was the one that maximized this correlation or was the solution at which the correlations flattened out (a reverse scree). The point biserial coefficient varied between 0.53 and 0.71 for the furniture store and between 0.56 and 0.71 for the car stereo centre, with the maximum value associated with the three-group solution in each case. However, as noted previously, the stability of these results needs to be assessed.

Following Calantone and Sawyer (1978) and Soutar and McNeil (1991), four analyses were undertaken to assess stability. First, respondents from each store were randomly assigned to one of two subsamples and separate cluster analyses were undertaken on each subsample. The point biserial correlations for the furniture store varied between 0.54 and 0.75 for the first sub-sample and between 0.52 and 0.69 for the second sub-sample. The point biserial correlations for the car stereo centre varied between 0.59 and 0.73 for the first subsample and 0.57 and 0.68 for the second subsample. In all cases, however, the maximum value was associated with the three-group solution, suggesting there was consistency as to the appropriate number of clusters.

Second, similar segments were identified for each of the two samples (furniture and car stereo) based on their mean scores over the three dissonance dimensions. Following this, the correlation of the mean scores over the twenty-two items in the scale was calculated for each sub-sample pair (one pair from the furniture sample and one pair from the car stereo sample). The lowest correlation was 0.80 and the average correlation was 0.88, which suggests responses to the various items were consistent across both sub-samples for both stores.

Third, apparently similar segments from each store were identified based on their mean scores, and t-tests were calculated between the matched groups from each store to see whether there were significant differences in their mean dissonance scores. According to the procedures recommended by Calantone and Sawyer (1978), a total of nine t-tests were computed (3 dissonance dimensions by 3 groups), with only three showing a significant difference between the two samples, providing further evidence of stability between the subsamples.

Finally discriminant analysis was used to develop classification rules for group membership in each subsample, and their effectiveness in classifying the matched group from the other subsample was examined. In this case, 80% of the cases for the furniture store were correctly classified, while 90% of the cases for the car stereo centre were correctly classified, again suggesting stability in the cluster analysis results. It would seem that the obtained clusters can be used with confidence. Consequently, the three-group solution for each store was accepted and the mean scores and percentages of the samples in each group are shown in table 2. As can be seen from table 2, the largest cluster in each case had very low scores across the three dissonance dimensions and was termed the 'low dissonance' group. Most respondents experienced minimal dissonance. The second cluster in each store also had a similar pattern, with low scores on the emotional and concern over the deal dimensions but a high score on the wisdom of purchase dimension. Hence this group was termed the 'concerned about needing the purchase' group. The third, and smallest, group in each store had relatively high scores on the three dissonance dimensions, and was termed 'the high dissonance' group. (2)

These results suggest that the three groups for each store type are similar. To investigate this further, differences between the six groups (3 groups x 2 stores) were analysed over the three dissonance dimensions using discriminant analysis. While the three estimated discriminant functions were significant, the [I.sup.2] statistic (Peterson & Mahajan 1976) suggested that the first two functions explained 89% of the variance between the groups. Consequently, the two-function results are reported. Both functions were important in differentiating the groups (explaining 49% and 40% of the variation between the groups respectively) and the positions of the groups on the two functions can be shown diagrammatically as in figure 1. The structural correlations, which are drawn as vectors in the diagram, show the relationship between the dissonance dimensions and the estimated discriminant functions. The head of the vector shows the direction of the relationship, while the length shows the strength of the relationship. As can be seen from the diagram, the first function is related to concern about the wisdom of the purchase, while the second is related to the emotional and concern about the deal dimensions. The three clusters from each store were very closely aligned, suggesting stability in the types of dissonance segments across the two stores.

[FIGURE 1 OMITTED]

4.3 Profiling the Segments

Having established the existence of three segments and, importantly, found that over a quarter of the furniture purchasers and over a third of the car stereo purchasers experience some form of dissonance, it is important to understand the nature of the two dissonant, as well as the non-dissonant, groups in order to develop appropriate marketing strategies. We can see whether dissonance varies according to product and purchase features or customers' backgrounds. Taking product and purchase features, we tested whether dissonance was greater for high priced than for low priced purchases. This question is central to the suggested conditions for dissonance that were discussed earlier, which argued that dissonance is more likely for high importance purchases (Festinger 1957; Oliver 1997). We also tested whether dissonance diminishes over time after purchase. Mowen (1995) argued that the extent of dissonance is likely to vary in the time period after purchase, being highest immediately after purchase and reducing over time. Indeed, it has been suggested that an individual's natural tendency is to attempt to reduce dissonance. Dissonance is lower if previous experience has increased a customer's confidence and familiarity with the store, since familiarity has been shown to reduce dissonance (Holloway 1967).

Turning to the customers themselves, the relationship between characteristics (such as age and gender) and dissonance was assessed. Research in psychology has found that men are more prone to dichotomous thinking and cognitive distortions than women and that they also experience more anger (Willert Jr. 1995; Pinto 2001), although women are more able to elaborate their reasons for dissonance (Dittmar & Drury 2000). Further, considering the rise in consumer empowerment and increasing consumer expectations (Vernarec 1997; Lovelock, Patterson & Walker 2001), younger consumers are believed to be more sophisticated and have higher expectations, and they also have a higher desire for involvement and higher associated dissonance (e.g. Thompson, Pitts & Schwankovsky 1993). Older people, in contrast, tend to be more trusting and less questioning of sales staff. Finally, we examined whether highly dissonant consumers have lower store and product evaluations overall. Do they perceive lower value and satisfaction with the product? Are their intentions towards the retailer less favourable than their low dissonant counterparts (for example, in terms of word of mouth and propensity to complain)? These questions were examined through a series of chi-square, ANOVA and MANOVA tests, as outlined in the following sections.

4.3.1 Dissonance Variations in Terms of Purchase Features It was clear that there were no differences in the prices of the goods purchased across the high dissonance (group 3), low dissonance (group 1) and 'concerned about needing the purchase' group (group 2), in either sample. On average, the purchases were valued at $830 for the car stereo store and $2000 for the furniture store (as can be seen in table 3).

There was also no evidence that dissonance decreased over time. Respondents completed the questionnaire at their convenience after purchase and were asked to report the time in days since purchase. Dissonance grouping was not related to the time after purchase at which dissonance was reported (through the questionnaire) (as can be seen in table 4). As the questionnaires were completed up to 12 days after the purchase, the length of time was relatively short. However, dissonance could be expected to vary in this short time period, especially in the case of the car stereo centre, as installation and experience with the product took place within 2 to 3 days. Further, the levels of each specific dissonance dimension did not vary across the time period. That is, customers' dissonance did not decrease over time (furniture F = 2.29, 0.26, 0.66 p > 0.10; car stereo F = 0.11, 0.00, 0.54 p > 0.10; for levels of emotional, wisdom of purchase and concern over the deal dimensions over three discrete post purchase time periods). Finally, dissonance groups did not vary with the number of previous purchases in the store (furniture [chi square] = 0.81 p > 0.10; car stereo [chi square] = 0.54 p > 0.10).

4.3.2 Dissonance Variations in Terms of Background Characteristics Significant differences were found in the profiles of the three groups. Older consumers were more common in the low dissonance group, while there were relatively more young consumers in the high dissonance group. Differences were significant in the car stereo sample but only marginally significant in the furniture sample (as can be seen in table 5 and table 6).

There were no significant gender differences, although the low dissonance group for the car stereo sample had proportionally more females and the high dissonance group proportionally more males (furniture [chi square] 0.83 p > 0.10; car stereo [chi square] = 4.15, p > 0.10). Similarly there was no occupational difference (furniture [chi square] = 19.78 df = 18p > 0.10; car stereo [chi square] = 15.47.15, df = 18, p > 0.10).

The three groups were also assessed in terms of the nine attitudinal and behavioural criteria shown in table7. Significant differences were found. Differences across the three groups in terms of all nine evaluative variables were assessed using MANOVA, which accounts for the interrelationships between the evaluative variables, providing a conservative test of differences. (3)

This analysis revealed significant differences (furniture: Wilk's A = 0.77, F = 2.51 df = 18. p < 0.01; car stereo: Wilk's [LAMBDA] = 0.57, F = 4.68 df = 18. p < 0.01), indicating that further analysis was warranted to find the nature of these differences. This was conducted through a series of ANOVA's, shown in table 7. The sample size for this phase of the analysis was restricted to the sample collected in the second stage, as the evaluative data were collected in this second phase. The results of the ANOVA tests showed, as expected, that respondents in the low dissonance group were most favourable in their evaluations, while respondents in the high dissonance group were least favourable. Those in the group concerned about whether they needed the purchase fell between these extremes. Differences were most apparent for the car stereo sample, where the high dissonance group held significantly more negative views than did either of the other groups. In particular, highly dissonant group members in the car stereo sample were more likely to have experienced difficulty in evaluating the product, were less likely to be satisfied with the product, were less likely to be actively loyal to the store, were more likely to switch to a competitive store and were less likely to perceive value in terms of price, emotional or quality benefits (as can be seen in table 8). However, none of the groups were particularly likely to complain on experiencing a problem. Thus, it might be difficult to identify dissonant consumers if the dissonance was caused by a problem experienced in the purchase process. While these results were based on small group, the results are remarkably similar, although they were less extreme in the furniture store sample than in the car stereo sample (as can be seen by comparing tables 7 and 8).

5. Conclusions and Managerial Implications

The present study was the first to investigate the presence of segments with different levels and types of dissonance, probably because, until recently, there has not been an accepted measure of consumer dissonance. The study used a recently developed instrument and found such segments, at least in the case of consumer durable purchases. Data from two samples suggested the presence of low- and high-dissonance segments, as well as a 'concerned about needing the purchase' segment. It is clear that dissonance should be of concern, as a significant minority of consumers (27% in the furniture store sample and 40% in the car stereo sample) experienced some form of dissonance. Given the current age of consumer empowerment, in which customers, particularly younger customers, have more involvement and higher expectations of service, including that offered by retail stores (Vernarec 1997; Lovelock, Patterson & Walker 2001), consumers are more likely to experience dissonance, leading to a clear call for management attention to dissonance.

There were few significant background predictors of dissonance. Table 6 suggested that younger consumers, particularly those purchasing car stereos, were more likely to experience dissonance, suggesting that such consumers need to be reassured in different ways. What was also apparent was that people who did not feel confident to evaluate product quality were more likely to experience some form of dissonance. This highlights the association between performance risk, which measures pre-purchase uncertainty associated with product performance (Murray & Schlacter 1990), and cognitive dissonance, as mentioned in section 2.4.

Managers would be wise to spend time and resources to reduce the perceived risk associated with a purchase, since this is likely to significantly reduce customers' dissonance experienced after the purchase. This has implications for the recruitment and training of sales staff and for the service strategies used, as sales staff play a crucial role in 'reassuring customers that a product will perform in the long term' (Sweeney, Soutar & Johnson 1999, p. 100). Sales staff, rather than managers, are best able to identify and act to mitigate the level of dissonance experienced, which will vary across customers, as highlighted in the qualitative comments of sales staff. Specifically a salesperson can, as appropriate, offer assurance by communicating an organisation's credibility and can reassure the customer that they have made the right choice (Zeithaml & Bitner 2000). The salesperson can also reduce dissonance by being empathetic, attentive, listening and by being adaptable in responding to customer needs (Zeithaml & Bitner 2000). A salesperson should act to reduce potential dissonance at the time of purchase, rather than leaving a customer with negative attitudes and behavioural intentions, such as lower satisfaction and perceived value, an increased likelihood of negative word of mouth about their experiences and a greater likelihood of switching (as seen in tables 7 and 8).

The salesperson can, for example, change a customer's cognitions about alternatives, by emphasizing the desirable characteristics of the chosen alternative and comparing the chosen alternative with the rejected alternative in the light of customer needs (' in your case I recommend product A because ... product B would not be so good because ...'). As dissonance, when it occurs, does not seem to fall in the few days after purchase (as seen in the insignificant relationship between the time after purchase and the dissonance groupings shown in table 4), there is an opportunity for the retailer to contact customers in this period to reassure them of the wisdom of their purchase and to allay any risk perceptions they may have. Consistent with Mowen's (1995) suggestion, the aim of such a follow up should be to give customers more information to support their decision and to assist them in evaluating their chosen product more highly than rejected alternatives. This may be achieved through advertising that is deliberately aimed at recent purchasers. Early researchers such as Engel (1963) recognised the interest in this post-purchase role of advertising, although studies have had mixed results. Nonetheless, such information strategies can be used to reduce dissonance through reassurance, rather than by complaining to others.

Further, it is important to recognise that dissonance can be increased by sales staff, particularly given the nature of the 'concern over the deal' dimension, which is based on a sense of being persuaded against one's own will. Thus, while sales staff can act to reduce dissonance through providing information and reassurance, they should not be too pushy at the point of sale or be artificially enthusiastic when following up after the sale. Managers need to be aware of the pivotal role that sales staff play and ensure that they are supported in dissonance reducing tactics, as a failure to assess and address dissonance can have negative outcomes.

6. Limitations and Future Research

Although the present study contributed to our understanding of consumer dissonance, it had several limitations. First, the response rates to the two surveys (44% from the furniture store and 31% from the car stereo outlet) were lower than the response rates reported by Dillman (2000). However, they were higher than those obtained in many other customer-evaluation studies that used self-completion survey methods (e.g. Parasuraman, Berry & Zeithaml 1991; Parasuraman, Zeithaml & Berry 1994). A second limitation is that dissonance was recalled a few days after purchase, rather than reported at the time of purchase. Several authors have noted that actual information processing differs from recall or memory of processing (e.g. Ericsson & Simon 1980). However, memory data is noted as being appropriate and useful in that it is linked to processing (Gardial, Clemons, Woodruff, Schmann & Burns 1994). While, ideally, respondents would have completed their questionnaire in the store, the length of the questionnaire meant this was not possible. Questionnaires were completed soon after purchase, however, (over 70% in both samples within 4 days) suggesting recall was less of an issue than if recall of a more distant event was attempted.

A further limitation is that the research was conducted in only two retail contexts (a furniture store and a car stereo centre). Evaluating the extent of dissonance and existence of clusters of dissonant consumers in other environments is a potential avenue for future research. In particular, the Internet provides a rich ground for future research on dissonance. With issues such as perceived control by the user (Dayal, Landsberg & Zeisser 1999; Han & McLaurin 2002), security issues (Furnell & Karweni 1999; Yoon 2002) and familiarity with the Internet (Miyazaki & Fernandez 2001; Yoon 2002) playing such a key role in Internet commerce, it is likely that negative aspects of such factors induce dissonance in an online environment. Indeed, greater dissonance may be evoked in an online retail setting than an offline retail setting. Dissonance is more likely in the relatively early stages of the evolution of the Internet and may also be particularly likely among newer users to the Internet.

An important finding was that dissonance did not fall in the few days after purchase. However, this finding was based on measuring the self-reported time since purchase and evaluating the corresponding dissonance level. Obtaining a second measure of dissonance at a later stage of the purchase process, such as after use, could have better assessed the reduction of dissonance. Budgetary limitations prevented such data being obtained in the present study.

Further, many preconceptions of dissonance theory were not supported in this study. For example, the price of the product purchased and the number of days after the purchase that the dissonance level was reported did not affect the nature of the dissonance experienced. (4) Hence, the contention that dissonance is higher for higher-priced products (Oliver 1997; Sweeney, Hausknecht & Soutar 2000) and that dissonance decreases after purchase (Mowen 1995) was not supported. However, there is a need to look into these propositions further. For example, the price range did not extend to very low-priced grocery goods or extremely high-priced lifestyle goods (cars, houses, etc.). Also, as the time in which questionnaires were completed after the purchase was mostly very short (almost 75% completed the survey 3 days or less after purchase), dissonance reduction may not have fully taken place.

Further investigation is needed to see whether the three dissonance clusters are present across other contexts and to develop richer descriptions of the clusters to assist managers in developing better marketing strategies. An important consideration is the value of the customer in each dissonance cluster. For example, based on customer relationship management (CRM) principles, if the customers in the small, high-dissonance group are valuable customers, then it is important to identify them early in the decision process and act to reduce their dissonance (Brady 2000).

Appendix A
Scales Used

Emotional Dissonance   After I bought this     Furniture [alpha]=0.97
dimension (15 items)   product:                Car stereo [alpha]=0.95

                       I was in despair
                       I resented it
                       I felt disappointed
                       with myself
                       I felt scared
                       I felt hollow
                       I felt angry
                       I felt uneasy
                       I felt I'd let myself
                       down
                       I felt annoyed
                       I felt frustrated
                       I was in pain
                       I felt depressed
                       I felt furious with
                       myself
                       I felt sick
                       I was in agony

Wisdom of purchare     I wonder if I really    Furniture [alpha]=0.86
Dissonance dimension   need this product       Car stereo [alpha]=0.79
(4 items)              I wonder whether I
                       should have bought
                       anything at all
                       I wonder if I have
                       made the right choice
                       I wonder if I have
                       done the right thing
                       in buying this
                       product

Concern over deal      After I bought this     Furniture [alpha]=0.83
Dissonance dimension   product I wondered if   Car stereo [alpha]=0.81
(4 items)              I'd been fooled
                       After I bought this
                       product I wondered if
                       they had spun
                       me a line
                       After I bought this
                       product I wondered
                       whether there
                       was something wrong
                       with the deal I got

Satisfaction           Examples:               Furniture [alpha]=0.95
(12 items)             This is one of the      Car stereo [alpha]=0.95
                       best..... I could
                       have bought
                       This..... is exactly
                       what I need
                       This..... hasn't
                       worked out as well as
                       thought it would

Active loyalty         Examples:               Furniture [alpha]=0.94
(5 items)              I say positive things   Car stereo [alpha]=0.95
                       about X to other
                       people
                       I encourage friends
                       and relatives to do
                       business with X

Propensity to switch   Take some of your       Furniture [alpha]=0.50
(2 items)              business to another     Car stereo [alpha]=0.60
                       car stereo or hi-fi
                       store with lower
                       prices
                       Do less business with
                       X in the next few
                       years

Propensity to          Examples:               Furniture [alpha]=0.50
complain behaviour     If you were to          Car stereo [alpha]=0.56
(4 items)              experience a problem
                       with service at X,
                       how likely would you
                       be to complain to
                       employees to
                       help them improve the
                       service
                       If you were to
                       experience a problem
                       with service at X,
                       how likely would you
                       be to complain to
                       other consumers

Value--price           Examples:               Furniture [alpha]=0.87
(4 items)              This product is a       Car stereo [alpha]=0.85
                       good one for the
                       price
                       This product offers
                       value for money

Value--quality         Examples:               Furniture [alpha]=0.93
(6 items)              This product has an     Car stereo [alpha]=0.93
                       acceptable standard
                       of quality
                       This product has a
                       high standard of
                       workmanship

Value--emotional       Examples:               Furniture [alpha]=0.92
(5 items)              This product is one     Car stereo [alpha]=0.92
                       that I enjoy
                       This product makes me
                       feel good

Value--social          Examples:               Furniture [alpha]=0.91
(4 items)              This product helps me   Car stereo [alpha]=0.91
                       to feet acceptable
                       This product makes a
                       good impression on
                       other people

Difficulty of          I had considerable      Single item
evaluating quality     difficulty in
(1 item)               evaluating the
                       quality of this
                       product

Table 1
Some Descriptive Statistics--Between the Stores

Dissonance Dimension     Store      Mean (a)   Standard       Alpha
                                               Deviation   Reliability
Emotional              Furniture      1.32       0.74         0.97
                       Car Stereo     1.36       0.71         0.95
Concern over the
  deal                 Furniture      1.46       0.88         0.86
                       Car Stereo     1.53       0.92         0.79
Wisdom of purchase     Furniture      2.46       1.48         0.83
                       Car Stereo     2.53       1.46         0.81

Note: (a.) 1 = strongly disagree (low dissonance) to 7 = strongly agree
(high dissonance).

Table 2
Group Percentages and Mean Scores

Cluster in Each   Emotional   Concern Over   Wisdom of   % in cluster
  Store                         the Deal     purchase

Furniture
1                  1.16          1.29          1.70           73
2                  1.19          1.36          4.49           20
3                  3.51          3.71          4.67            7

Car Stereo
1                  1.10          1.12          1.51           60
2                  1.32          1.47          3.94           27
3                  2.59          3.47          4.21           13

Table 3
Amount Spent on Purchase by Group

                            Amount Spent

                                       Car
                          Furniture   Stereo
Low dissonance              $1950      820
'Concerned about
  needing the purchase'     $2100      890
High dissonance             $2150      760
Average across groups       $2000      830

Note: Furniture: F = 0.424, p > 0.10; and
Car stereo: F = 0.415, p > 0.10.

Table 4
Time After Purchase by Group

                              Average Time After Purchase to Completion
                                            of Survey Form

                                 Furniture                Car Stereo

Low dissonance                   3.6 days                  3.1 days
Concerned about needing the
  purchase                       4.2 days                  2.9 days
High dissonance                  2.0 days                  3.0 days
Average across groups            3.7 days                  3.0 days

Note: Furniture: F = 0.0.236, p > 0.10; and
Car stereo: F = 0.153, p > 0.10.

Table 5
Age of Groups--Furniture Store

                   Low       Concerned About     High        Total
                Dissonance     Needing the     Dissonance   (N = 322)
                  Group         Purchase         Group
                (N = 237)      (N = 64)         (N = 21)

25-29 Years *      16%             6%             10%          13%
30-39 Years        33%            39%             52%          35%
40-49 Years        32%            25%             29%          30%
50+ Years          20%            30%             10%          21%
TOTAL             100%           100%            100%         100%

Note: [chi square] = 10.69, p < 0.10); and

* No furniture buyers were under 25 years of age.

Table 6
Age of Groups--Car Stereo

                    Low       Concerned About       High      Total
                 Dissonance     Needing the      Dissonance
                   Group         Purchase          Group

Under 18 years       4%             9%               7%          6%
18-24 years         47%            54%              55%         50%
25-29 years         13%            22%              21%         17%
30-39 years         18%             6%               7%         13%
40-49 years         10%             6%               7%          9%
50+ years            8%             2%               2%          5%
TOTAL              100%           100%             100%         100%
                 (N = 186)      (N = 85)         (N = 42)     (N = 313)

Note: [chi square] = 20.13, p < 0.05.

Table 7
Evaluations by Group--Furniture Store

                           Low       Concerned      High
                       Dissonance      About      Dissonance
                          Group     Needing the     Group
                        (N = 132)     Purchase     (N = 11)
                                      (N = 33)

Satisfaction            5.96 (a)     5.80 (a,b)    5.01 (b)
with product (x)

Difficulty in
evaluating
product (x)             1.80 (a)     2.12 (a)      4.09 (b)

Active loyalty
(word of mouth
etc.) (x)               5.66 (a)     5.34 (a,b)    4.60 (b)

Propensity to
switch (x)              3.05         3.12          4.00

Likelihood
of complaining
if problem
experienced (x)         4.12         4.21          4.54

Perceived
price value of
product (y)             5.55 (a)     5.30 (a,b)    4.68 (b)

Perceived emotional
value of product (y)    5.81         5.76          5.31

Perceived
quality value
of product (y)          6.04 (a)     5.75          5.30 (b)

Perceived social
value of product (y)    4.37         4.21          4.44

Evaluations by Group--Furniture Store

                       Total for     F(z)
                        Entire
                        Sample
                       (N = 176)

Satisfaction            5.88        5.01 ***
with product (x)

Difficulty in
evaluating
product (x)             2.01       16.09 ***

Active loyalty
(word of mouth
etc.) (x)               5.53        4.53 **

Propensity to
switch (x)              3.12        2.84 *

Likelihood
of complaining
if problem
experienced (x)         4.16        0.70

Perceived
price value of
product (y)             5.44        4.16 **

Perceived emotional
value of product (y)    5.77        1.19

Perceived
quality value
of product (y)          5.94        4.43 **

Perceived social
value of product (y)    4.35        0.18

Note: (x): 1 = not at all likely, 7 = extremely likely;
(y): 1 = strongly disagree, 7 = strongly agree;
(z): p value *** p < 0.01, ** p < 0.05, * p < 0.10; and
group means with different superscripts in table refer to groups where
differences are significantly differentatp < 0.05.

Table 8
Evaluations by Group--Car Stereo

                            Low        Concerned        High
                         dissonance   about needing   dissonance
                           group      the purchase      group
                          (N = 83)      (N = 39)       (N = 19)

Satisfaction with         5.94 (a)       5.74 (a)       4.71 (b)
product (y)
Difficulty in             1.53 (a)       2.12 (a)       3.37 (b)
evaluating product (x)
Active loyalty            5.87 (a)       5.72 (a)       4.49 (b)
(word of mouth
etc.) (x)
Propensity to             2.44 (a)       2.67 (a)       3.92  (b)
switch (x)
Likelihood of             3.75           4.04           3.73
complaining if
problem
experienced (x)
Perceived price           5.74 (a)       5.50 (a)       4.37 (b)
value of product (y)
Perceived emotional       5.95 (a)       5.90 (a)       5.01 (b)
value of product (y)
Perceived quality         6.11 (a)       6.29 (a)       4.86 (b)
value of product (y)
Perceived social          4.36 (a)       4.41 (a)       3.88 (b)
value of product (y)

                           Total        F (z)
                         for entire
                           sample
                          (N = 141)

Satisfaction with           5.72      12.48 ***
product (y)
Difficulty in               1.94      16.41 ***
evaluating product (x)
Active loyalty              5.64      10.35 ***
(word of mouth
etc.) (x)
Propensity to               2.70       9.69 ***
switch (x)
Likelihood of               3.82       0.79
complaining if
problem
experienced (x)
Perceived price             5.49      16.34 ***
value of product (y)
Perceived emotional         5.81       7.88 ***
value of product (y)
Perceived quality           5.99      17.88 ***
value of product (y)
Perceived social            4.32       0.86
value of product (y)

Note: x: 1 =not at all likely, 7=extremely likely;
y: 1= strongly disagree, 7=strongly agree; and
z: p value *** p < 0.01, ** p < 0.05, * p < 0.10.

(1.) Dissonance was not measured in the second stage as respondents had experience with the product by this time and, consequently, dissonance should have converted to satisfaction or dissatisfaction.

(2.) Cluster analysis sometimes produces two clusters--one with respondents who gave high ratings to all measures and one with respondents who gave low ratings to all measures. Such a result may be a function of response style. However, the presence of a third cluster, with high scores on one dimension and low scores on the other two, suggests the result is not merely an artifact of response style.

(3.) MANOVA was used to assess the relationship between the clusters and outcome variables such as value and loyalty. MANOVA was used initially, rather than a series of ANOVA's to maintain control over the error rate that can occur by chance when conducting a series of ANOVA's (Hair, Anderson, Tatham and Black 1998).

(4.) This analysis was conducted across the overall data set. Hence the results indicated that dissonance was no different for higher priced goods than for lower priced goods, and that this did not change in the few days after purchase.

References

Arabic, P. & Hubert, L. 1994, 'Cluster analysis in marketing research', in Advanced Methods of Marketing Research, ed. R. P. Bagozzi, Oxford: Blackwell & Company, pp. 160-89.

Arnthorsson, A., Berry, W.E. & Urbany, J.E. 1991, 'Difficulty of pre-purchase quality inspection: Conceptualization and measurement', Advances in Consumer Research, vol. 18, pp. 217-24.

Aronson, E. 1969, 'The theory of cognitive dissonance: A current perspective', Advances in Experimental Social Psychology, vol. 4, ed. L. Berkowitz, Academic Press, New York, pp. 1-34.

Bauer, R.A. 1960, 'Consumer behaviour as risk taking', in Dynamic Marketing for a Changing World, ed R.S. Hancock, American Marketing Association, Chicago, Illinois, pp. 389-98.

Bell, G.D. 1967, "The automobile buyer after purchase', Journal of Marketing, vol. 31, no. 3, pp. 12-6.

Bitner, M.J. 1992, 'Servicescapes: The impact of physical surroundings on customers and employees', Journal of Marketing, vol. 56, no. 2, pp. 57-71.

Brady, D. 2000, 'Why service stinks', Business Week, October 23, pp. 118-28.

Brown, J.D. 1988, Understanding Research in Second Language Learning: A Teacher's Guide to Statistics and Research Design. Cambridge University Press, Cambridge.

Brusco, M.J., Cradit, J.D. & Stahl, S. 2002, 'A simulated annealing heuristic for a bicriterion partitioning problem in market segmentation', Journal of Marketing Research, vol. 39, no. 1, pp. 99-109.

Calantone, R.J. & Sawyer, A.G. 1978, 'The stability of benefit segments', Journal of Marketing Research, vol. 15, no. 3, pp. 395-404.

Cohen, J.B. & Goldberg, M.E. 1970, 'The dissonance model in post-decision product evaluation', Journal of Marketing Research, vol. 7, no. 3, pp. 315-21.

Cox, D. 1967, 'The sorting rule model of the consumer product evaluation process', Risk Taking and Information Handling in Consumer Behaviour, ed. C. Donald, Graduate School of Business Administration, Harvard University, Boston, Mass, pp. 324-69.

Cummings, W.H. & Venkatesan, M. 1976, 'Cognitive dissonance and consumer behaviour: A review of the evidence', Journal of Marketing Research, vol. 13, no. 3, pp. 303-08.

Dayal, S., Landsberg, H. & Zeisser, M. 1999, 'How to build trust online', Marketing Management. vol. 8, no. 3, pp. 64-8.

Diamantopoulos, A. & Winklhoffer, H.M. 2001, 'Index construction with formative indicators: An alternative to scale development', Journal of Marketing Research, vol. 38, no. 2, pp. 269-77.

Dillman, D.A. 2000, Mail and Internet Surveys: The Tailored Design Method, Wiley, New York.

Dittmar, H. & Drury, J. 2000, 'Self-image--Is it in the bag? A qualitative comparison between ordinary and excessive consumers', Journal of Economic Psychology, vol. 21, no. 2, pp. 10-42.

Dowling, G. 1999, 'Perceived risk', in The Elgar Companion to Consumer Research and Economic Psychology, eds. P.E. Earl & S. Kemp, Edward Elgar Publishing Ltd, Cheltenham, pp. 419-24.

Elliot, A.J. & Devine, P.G. 1994, 'On the motivational nature of cognitive dissonance: Dissonance as psychological discomfort', Journal of Personality and Social Psychology, vol. 67, no. 3, pp. 382-94.

Engel, J.F. 1963, 'Are automobile purchasers dissonant consumers?' Journal of Marketing, vol. 27, April, pp. 55-8.

Eriksson, K.A. & Simon, H.A. 1980, 'Verbal reports as data', Psychological Review, vol. 87, no. 2, pp. 215-50.

Festinger, L. 1957, A Theory of Cognitive Dissonance, Stanford University Press, Stanford, CA.

Furnell, S.M. & Karweni, T. 1999, 'Security implications of electronic commerce: A survey of consumers and businesses', Internet Research." Electronic Networking Application and Policy, vol. 9, no. 5, pp. 372-82.

Gardial, S.F., Clemons, D.S., Woodruff, R.B., Schumann, D.W. & Burns, M.J. 1994, 'Comparing consumer's recall of prepurchase and postpurchase product evaluation experiences', Journal of Consumer Research, vol. 20, March, pp. 548-60.

Green, P.E., Carmone Jr, F.J. & Kim, J. 1990, 'A preliminary study of optimal weighting on K-means clustering', Journal of Classification, vol. 7, pp. 271-85.

Hair, J.F., Anderson, R.E., Tatham, R.L. & Black W.C. 1998, Multivariate Data Analysis, fifth edition, Prentice Hall, Upper Saddle River, NJ.

Han, P. & Maclaurin, A. 2002, 'Do consumers really care about online privacy?' Marketing Management, vol. 11, no. 1, pp. 35-8.

Harmon-Jones, E. & Mills, J. 1999, Cognitive Dissonance." Progress on a Pivotal Theory in Social Psychology, American Psychological Association, Washington, DC.

Helsen, K. & Green, P.E. 1991, 'A computational study of replicated clustering with an application to marketing research', Decision Sciences, vol. 22, no. 5, pp. 1124-41.

Holloway, R.J. 1967, 'An experiment on consumer dissonance', Journal of Marketing, vol. 31, no. 1, pp. 39-43.

Howard, H. & Harris, R.D. 1966, 'A hierarchical grouping routine; IBM 360/65 FORTRAN IV program', University of Pennsylvania Computing Centre.

Hunt, S.D. 1970, 'Post-transaction communications and dissonance reduction', Journal of Marketing, vol. 34, no. 3, pp. 46-51.

Kanungo, T., Mount, D.M., Netanyahu, N.S., Piatko, C.D., Silverman, R, & Wu, A.Y. 2002, 'The analysis of a simple K-means clustering algorithm: Analysis and implementation', IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 24, no. 6, pp. 1-12.

Kassarjian, H.H. & Cohen, J.B. 1965, 'Cognitive dissonance and consumer behaviour', California Management Review, vol. 8, no. 1, pp. 55-64.

Korgaonkar, P.K. & Moschis, G.P. 1982, 'An experimental study of cognitive dissonance, product involvement, expectations, performance and consumer judgement of product performance', Journal of Advertising, vol. 11, no. 3, pp. 32-44.

Losciuto, L.A. & Perloff, R. 1967, 'Influence of product preference on dissonance reduction', Journal of Marketing Research, vol. 4, no. 3, pp. 286-90.

Lovelock, C.H., Patterson, P.G. & Walker, R.H. 2001, Services Marketing: An Asia Pacific Perspective, second edition, Prentice Hall Australia Pty Ltd, Sydney.

Menasco, M.B. & Hawkins, D.I. 1978, 'A field test of the relationship between cognitive dissonance and state anxiety', Journal of Marketing Research, vol. 15, no. 4, pp. 650-55.

Milligan, G.W. & Mahajan, V. 1980, 'A note on procedures for testing the quality of clustering a set of objects', Decision Sciences, vol. 11, no. 4, pp. 669-77.

Miyazaki, A.D. & Fernandez, A. 2001, 'Consumer perceptions of privacy and security risks for online shopping', Journal of Consumer Affairs, vol. 35, no. 1, pp. 27-44.

Mowen, J.C. 1995, Consumer Behaviour, fourth edition, Prentice-Hall, Englewood Cliffs, NJ.

Murray, K. & Schlacter, J.L. 1990, 'The impact of services versus goods on consumers' assessment of perceived risk and variability', Journal of the Academy of Marketing Science, vol. 18, no. l, pp. 51-65.

Oliver, R.L. 1997, Satisfaction." A Behavioral Perspective on the Consumer, McGraw-Hill Book Company, New York.

Oshikawa, S. 1970, 'Consumer pre-decisional conflict and post-decision dissonance', Behavioral Science, vol. 15, no. 1, pp. 132-40.

Parasuraman, A., Berry, L.L. & Zeithaml, V.A. 1991, 'Refining and reassessment of the SERVQUAL scale', Journal of Retailing, vol. 67, no. 4, pp. 420-50.

Parasuraman, A., Zeithaml, V.A. & Berry, L. 1994, 'Alternative scales for measuring service quality: A comparative assessment based on psychometric and diagnostic criteria', Journal of Retailing, vol. 70, no. 3, pp. 201-30.

Peterson, R.A. & Mahajan, V. 1976, 'Practical significance and partitioning variance in discriminant analysis', Decision Sciences, vol. 7, no. 4, pp. 649-58.

Pinto, D. 2001, Driving Anger, Articulated Cognitive Distortions, Cognitive Deficiencies and Aggression, Doctoral Dissertation Hofstra University, Hempstead, New York [Publication number AAT3001588].

Punj, G. & Stewart, D.W. 1983, 'Cluster analysis in marketing research: Review and suggestions for application', Journal of Marketing Research, vol. 20, no. 2, pp. 134-48.

Smith, S.M. 1988, PC-MDS: A Multidimensional Statistics' Package, Brigham Young University, Provo, Utah

Soutar, G.N. & McNeil, M.M. 1991, 'A benefit segmentation of the financial planning market', International Journal of Bank Marketing, vol. 9, no. 2, pp. 25-9.

Soutar, G.N. & McNeil, M.M. 1995, 'Information search for a professional service', Journal of Professional Services Marketing, vol. 11, no. 2, pp. 45-60.

Soutar, G.N., McNeil, M.M. & Maisey, G. 1995, 'Community satisfaction with policing: The Western Australian experience', The Police Journal, vol. 58, no. 4, pp. 351-67.

Soutar, G.N., Ramaseshan, B. & Molster, C.M. 1994, 'Determinants of pro-environmental consumer purchase behaviour: Some Australian evidence', Asia Pacific Advances in Consumer Research, vol. 1, pp. 28-35.

Sweeney, J.C., Hausknecht, D. & Soutar, G.N. 2000, 'Measuring cognitive dissonance: A multidimensional scale', Psychology and Marketing, vol. 17, no. 5, pp. 369-86.

Sweeney, J.C. & Soutar, G.N. 2001, 'Consumer perceived value: The development of a multiple item scale', Journal of Retailing, vol. 77, no. 2, pp. 203-20.

Sweeney, J.C., Soutar, G.N. & Johnson, L.W. 1999, 'The role of perceived risk in the quality-value relationship: A study in a retail environment', Journal of Retailing, vol. 75, no. 1, pp. 77-105.

Thompson, S.C., Pitts, J.S. & Schwankowsky, L. 1993, 'Preferences for involvement in medical decision making: Situational and demographic differences,' Patient Education and Counselling, vol. 22, no. 1, pp. 133-40.

Vernarec, E. 1997, 'The consumer as a health care manager', Business and Health, Special Issue, pp. 51-4.

Willert Jr., M.G. 1995, Coping Strategies and Emotional and Physical Status of Family Members of Mental Health Consumers. Doctoral Dissertation, The University of North Dakota [Publication number AAT9605485].

Yoon, S.J. 2002, 'The antecedents and consequences of trust in online purchase decisions', Journal of Interactive Marketing, vol. 16, no. 2, pp. 47-63.

Zeithaml, V.A., Berry, L.L. & Parasuraman, A. 1996, 'The behavioral consequences of service quality', Journal of Marketing, vol. 60, no. 2, pp. 31-46.

Zeithaml, V.A. & Bitner, M.J. 2000, Services Marketing: Integrating Customer Focus across the Firm, second edition, McGraw-Hill Book Company, New York.

Geoffrey N. Soutar ([dagger])

Jillian C. Sweeney ([dagger])

([dagger]) UWA Business School, University of Western Australia, Perth WA 6009. Email: [e-mail] [e-mail]

The helpful comments of the two reviewers and the editor are gratefully acknowledged, although responsibility remains with the authors.

(Date of receipt of final transcript: April 16, 2003. Accepted by Mark Uncles, Area Editor.)