Does Satisfaction With Multi-Attribute Products Vary Over Time? a Performance Based Approach

ABSTRACT - A multi-attribute approach to study satisfaction is suggested. Linkages between an attribute's performance, goal-fulfillment (Swan 1988), and satisfaction are drawn. Preliminary exploration of automotive industry data suggests the viability of the approach. Results indicate that performance on different attributes contribute differentially to CS, and that an attribute's salience seems to change between pre-purchase evaluations and post-purchase judgement of satisfaction. Research and managerial implications are discussed.

Citation:

Vikas Mittal, Jerome M. Katrichis, Frank Forkin, and Mark Konkel (1994) ,"Does Satisfaction With Multi-Attribute Products Vary Over Time? a Performance Based Approach", in NA - Advances in Consumer Research Volume 21, eds. Chris T. Allen and Deborah Roedder John, Provo, UT : Association for Consumer Research, Pages: 412-417.

Advances in Consumer Research Volume 21, 1994      Pages 412-417

DOES SATISFACTION WITH MULTI-ATTRIBUTE PRODUCTS VARY OVER TIME? A PERFORMANCE BASED APPROACH

Vikas Mittal, Temple University

Jerome M. Katrichis, Temple University

Frank Forkin, Research Data Analysis, Inc.

Mark Konkel, Consumer Attitude Research, Inc.

ABSTRACT -

A multi-attribute approach to study satisfaction is suggested. Linkages between an attribute's performance, goal-fulfillment (Swan 1988), and satisfaction are drawn. Preliminary exploration of automotive industry data suggests the viability of the approach. Results indicate that performance on different attributes contribute differentially to CS, and that an attribute's salience seems to change between pre-purchase evaluations and post-purchase judgement of satisfaction. Research and managerial implications are discussed.

INTRODUCTION

The dominant paradigm of expectation disconfirmation (Oliver 1980) as an explanation of customer satisfaction (CS) has been extended and modified recently by many theorists (c.f Yi 1990). Findings suggest that disconfirmation, expectations and performance affect satisfaction differentially based on factors such as individual response tendencies (Oliver and Desarbo 1988) and product durability (Churchill and Surprenant 1982). While the role of disconfirmation and expectations in satisfaction judgement has been widely studied, relatively little attention has been devoted to the effect of performance on satisfaction (Swan 1988; Yi 1990). Studies examining the impact of performance on satisfaction seem to indicate that product performance has a separate and distinguishable impact on satisfaction irrespective of expectations (cf. Churchill and Suprenant 1982, Swan 1988); however, these studies generally treat performance as a uni-dimensional construct. That is, they either examine performance of a single attribute, or use an overall "index" of performance and relate it to satisfaction. On the other hand, most research considering consumer evaluation of decision alternatives takes the approach that products have multiple attributes with varying performance levels (cf. Wilkie and Pessemier 1972; Bettman, Johnson and Payne 1992). More recently, such a multi-dimensional approach has been taken in a satisfaction context. Oliva et al (1992), have proposed a catastrophe model of consumer satisfaction with service, where specific service attributes are related to overall service satisfaction in order to determine the impact of a particular attributes on satisfaction. This points to both a conceptual and methodological similarity of approach between multi-attribute attitude research and satisfaction research.

This study explores the structure of attribute contributions to overall satisfaction and compares this structure to the self reported importance of attributes in the formation of initial purchase preference attitudes. The purpose of such an exploration is to emphasize the temporal variation in the importance of product attributes in the satisfaction judgement process. That is, this study proposes that there may be a vital distinction between attributes consumers consider important in product choice versus satisfaction evaluation. This is accomplished utilizing a data set from the automotive industry.

PERFORMANCE AND SATISFACTION

Although earlier studies used product performance ratings as a proxy for CS (e.g., Olson and Dover 1976; Anderson 1973), later studies (Oliver and Desarbo 1988; Swan 1988; Churchill & Surprenant 1982) have conceptually distinguished between performance and other moderators of CS such as disconfirmation and expectations. Churchill and Surprenant (1982) demonstrated discirminant validity between performance, disconfirmation, and expectation. An interesting finding has been that performance has an effect on satisfaction that is separate and distinct from, and sometimes stronger than, disconfirmation or expectations (Churchill and Surprenant 1982; Swan 1988). In fact, for a complex durable good (Video disk product) Churchill and Surprenant (1982) found that performance had the largest impact of all variables used to explain satisfaction. They speculated that performance was the dominant factor in determining CS for durable goods.

Swan (1988) proposed an argument about the performance-satisfaction link. He argued that consumers buy products to fulfil their needs/values and the degree to which the needs/values are fulfilled depends on how well the product performs. As product performance increases, the needs/values are better met leading to increased satisfaction. In addition, holding disconfirmation constant, higher satisfaction would result from higher performance (Swan 1988). The relationship was tested for restaurants where satisfaction with food and service was studied separately. Results showed that performance was a significant predictor of satisfaction compared to disconfirmation and expectation. For service satisfaction the results were mixed. The first performance component, time, was not significant while the second performance component, attention, was highly significant. These findings raise the possibility that different attributes of a product impact CS differentially; this is especially true because even for food satisfaction, the "performance index" had a low reliability alpha suggesting the possibility that performance on various attributes regressed separately may have had more explanatory power.

Swan's conceptualization specifies a positive monotonic relationship between performance on an attribute and satisfaction, but one can easily think of situations when a decrease in performance or a fixed level of performance on an attribute leads to increased satisfaction. For instance, in the case of an automobile, weak acceleration may cause higher satisfaction for a customer who is concerned with the control of the vehicle or who has young drivers in the household. Powerful acceleration, on the other hand, may cause higher satisfaction to a driving enthusiast. Similarly, in the case of prescription glasses, the more precise (not higher) the performance, the higher the satisfaction. The reason is that satisfaction may not be linked directly to performance, but to the degree with which performance leads to goal fulfillment (Johnson 1984). Thus, any level of performance that maximizes goal fulfillment, should lead to higher satisfaction. This notion is consistent with Johnson's (1984) comparison of non-comparable alternatives; since the performance of different attributes (e.g. trunk space versus. mpg) cannot be directly compared, customers may use a higher level of abstraction (i.e. goal-fulfillment) to determine satisfaction. Thus, the level of performance that maximizes satisfaction varies with the attribute's contribution to customer's goal fulfillment.

MULTI-ATTRIBUTE MODELS

The literature on the multi-attribute nature of products has been developed in diverse marketing areas such as choice modelling (Danes and Cattin 1980), attitude modelling (Wilkie and Pessemier 1973), and consumer information environments (Johnson and Katrichis 1988). Within the literature on choice and attitude modelling an important consideration has been the determination of the salience of attributes in terms of their impact on a choice or formation of an attitude.

Within this literature the issue of the impact of particular attributes on the judgement or choice decision has gone by many names. Myers and Alpert (1968) referred to it as determinance to distinguish it from importance of an attribute to a consumer. For instance, a steering wheel might be an important attribute for your car to have, but because all cars have them, it will not be determinant, or have an impact on brand choice. The term salience is more often used (cf. Day 1972; Wilkie and Pessemier 1973) to make the same distinction, while others simply use the term importance, or decision importance (Bettman et al 1992). Operationally, the difference is generally one of inserting the phrase "to your decision" after the word important in measurement items (Wilkie and Pessemier 1973) when direct methods are used.

Most of the research concerning attribute salience has focussed on pre-purchase decision making (c.f. Bettman et al 1992). Rendering satisfaction evaluations, is , however, a post-purchase phenomenon. Here we assert that salience of an attribute varies temporally as the customer uses the product, and thus attenuates satisfaction judgements over time.

SATISFACTION AND MULTI-ATTRIBUTE PRODUCTS

The process of expectations formation about a product is a learning experience based on previous experiences (Woodruff, Cadotte, and Jenkins 1983) with the particular product or similar products. Part of the learning involves recognizing salient attributes and forming expectations about them (Day 1977). Such an attribute by attribute learning process occurs for not only complex products but also simple products (Day 1977). Bolton and Drew (1991) found that customers accord different weights to the core, facilitating, or supporting attributes while evaluating service satisfaction.

LaTour and Peat (1979) propose a model of satisfaction where overall satisfaction is a linearly additive function of the outcome of attribute level comparisons and each outcome is weighted by its stated importance. Thus, the more important an attribute, the higher its impact on overall satisfaction. The model includes salient attributes, but does not discriminate between different degrees of salience among attributes. Thus in their model, importance remaining constant, all salient attributes should be equally relevant in satisfaction evaluation. However, empirical evidence (Wilkie and Weinreich 1972) shows that importance is a poor criteria for judging the relevance (or salience) of an attribute in the evaluation process. Furthermore, linear compensatory models of satisfaction judgement formation, such as the one proposed by LaTour and Peat, are not only difficult for customers to execute but also require the explicit resolving of difficult value trade-offs (Bettman et al 1992). Additionally, linear additivity of attribute level satisfaction may not hold under the "elimination by aspects" or "frequency of good and bad features" models of judgement formation (Bettman et al 1992). In these respects, the weighted average rule appears to be more of a normative strategy than a realistic strategy for judgement formation.

One can easily conceive of situations where "unimportant" attributes become salient through use of a product and therefore relevant in satisfaction evaluation. A young car driver may be very enthusiastic about the styling of her car (high importance) but be continuously bothered by the leaky engine (salient); more likely this customer will be dissatisfied with her car. Thus, salience of an attribute should be the relevant criteria for a customer's satisfaction judgement. The logic here is that, in the context of satisfaction, more salient attributes will be more "active" in memory and more readily available for processing (Anderson 1990); in other words, memory "nodes" regarding salient attributes will have a higher activation potential and therefore influence the satisfaction judgement.

In addition, the performance (hence contribution to goal attainment) of certain attributes can be known with reasonable certainty (eg. trunk space) before the purchase process; on the other hand, the performance of other "sensory" or "experiential" attributes is uncertain (eg. wind noise) and may only be "discovered" during the post-purchase consumption process. Expectation-disconfirmation regarding attributes that were evaluated with certainty before the purchase process must already be positive or else it would result in strong cognitive dissonance (Festinger 1957); these attributes, though still salient, may not be instrumental in the subsequent satisfaction judgement task.

PROCEDURE

Data from a survey of new-car buyers for the year 1992 was used to explore the above ideas. The survey was conducted among buyers of new domestic (all brands) 1992 model year vehicles purchased in the months January - March 1992. Respondents had used the vehicle for 3-5 months ensuring that they had experience with the various attributes under investigation. Of the 134,833 surveys mailed out, 43,908 (33.5%) were usable returns. For the sake of efficiency, a small random sampling of completed responses yielded a working sample of 1,342 responses for the current study. The mean age of these respondents was between 40-44 years with an average of 4,854 miles on the new car purchased. The average price of the car bought was 19,000 dollars. These numbers indicate that the respondents included in this study are not only representative of the industry, but also that they had driven the vehicle long enough to "explore" the various attributes of the product.

Measurement

Overall satisfaction was measured on a 5 point scale (5 = completely satisfied and 1 = very dissatisfied); attribute level performance was measured on a 5 point scale (5 = excellent and 1 = poor). Pre-purchase salience was assessed via a 5 point scale (5= extremely important, 1=not at all important) in answer to the question "How important was each of the following in your decision?" Attribute level performance was measured for 27 attributes. Of those 27 attributes, pre-purchase importance ratings were available for 9 attributes.

Analysis and Results

Descriptive statistics for attribute level performance and pre-purchase importance are reported in Table 1. Correlations between attribute-level performance and overall satisfaction, and purchase decision importance and overall satisfaction are reported in Table 2. The data show that performance on various attributes was rated differentially with a range of 4.47 to 3.43. All relationships between attribute level performance ratings and overall satisfaction are positive and significant, as are the relationships between pre-purchase importance ratings and overall satisfaction. Further, performance on different attributes was differentially related to overall satisfaction.

TABLE 1

DESCRIPTIVE STATISTICS

Table 3 shows the rankings of performance ratings in column one, and decision importance ratings in column 2. Column 3 shows the rankings of the correlations between attribute level performance and overall satisfaction, and column 4 shows the relationship between an attribute's decision salience and satisfaction.

Several points are of interest in this table. First, a comparison of columns 2 and 3 seems to indicate that there is some difference between how salient an attribute is in the purchase decision and how much performance on that attribute contributes to overall satisfaction. Most notably, the exterior styling of the vehicle dropped from 5th place in pre-purchase decision importance to 8th place in terms of its relationship with overall satisfaction. This may be because, regardless of the level of exterior styling of the vehicle, it can be assessed with a relatively high degree of certainty prior to the decision to purchase. Once the purchase decision is made, styling remains exactly what it was prior to purchase and is no longer salient in judgements about satisfaction. A similar case could be made for ease of handling. Riding comfort, power/pickup, quietness, and value for the money all increased in salience after the purchase. For the first three attributes it could be argued that this was due to the "experiential" or hedonic nature (Kahneman and Snell, 1990) of the attributes; that is, because they are attributes that the consumer experiences on a fairly regular basis after purchase, they have increased in salience. The increase in value for money is harder to account for but may be due to a difference in a pre-purchase and post-purchase comparison group for consumers. Pre-purchase comparisons of value for the money may be between different price ranges of vehicles. In the pre-purchase choice task, consumers may be asking themselves whether a $25,000.00 vehicle represents as much value for the money as a $15,000.00 vehicle. After the purchase, consumers, if they bought a $15,000.00 vehicle, may be making judgements about the value for the money on $15,000.00 vehicles only.

The simple correlation coefficients between each attribute's performance rating and satisfaction were used as a measure of the attribute's contribution to satisfaction. A simple correlation coefficient was calculated between these correlations and the self reported pre-purchase importance scores yielding a relationship of .77 (p<.05), indicating a much stronger relationship than expected between pre-purchase importance of an attribute and its relationship to overall satisfaction. Visual inspection of the data reveals that the data-point for "Cargo-capacity" is indeed an influential outlier (Everitt and Dunn 1992). As explained earlier and as evidenced by the fact that it ranked last in all categories (Table 3) "Cargo-capacity" appears to be an unimportant and non-salient attribute. We therefore, recomputed the correlation coefficient which was .66 (p<.05), indicating that post purchase salience may be related to pre-purchase salience but that they are somewhat different. The strength of this relationship may also be due in part to a couple of methodological difficulties. First, the 3-5 months driving experience, after purchase may not have given salience the opportunity to change dramatically. Second, and probably more likely, pre-purchase importance was assessed at the same time as performance and satisfaction, and was assessed post-purchase. So some of the salience shifting might already be part of the pre-purchase importance ratings utilized.

TABLE 2

SIMPLE CORRELATION COEFFICIENTS BETWEEN OVERALL SATISFACTION AND ATTRIBUTE PERFORMANCE/PRE-PURCHASE IMPORTANCE

TABLE 3

RANKS

One puzzling observation is that despite ranking fourth in performance "Durability/Reliability" (DR) has the highest correlation with overall satisfaction. One possible explanation, is that in recent years the auto-industry has consistently advertised DR (sometimes under the rubric "quality") as a yardstick of satisfaction. Thus, by making DR salient and, in some cases, leading customers to automatically (automatically is used here in an information-processing manner) equate DR (attribute performance) with satisfaction (goal-fulfillment) we may be obtaining a biased result. Another explanation is that information regarding DR was processed at a "deeper" level (Craik and Lockhart 1972).

DISCUSSION

The direct link between goal-fulfillment and satisfaction on an attribute level has not been explored within the CS literature. The current findings, provide justification for exploring this link. Despite the methodological limitations, the notion that an attribute's salience in the purchase decisions is different than salience for satisfaction judgements is supported. This has several implications.

First, we find that information-processing and multi-attribute choice modelling approaches (Bettman et al 1992) can be fruitfully applied to CS research within a goal-fulfillment (Johnson 1984) model. Second, the results offer support for arguments that CS research should not treat complex, multi-attribute products as simple products (Day 1977) or the satisfaction judgement decision as a uni-dimensional decision. Increasingly, scholars are acknowledging the multi-dimensional nature of satisfaction as an affective construct (Westbrook and Oliver 1991). Our findings suggest that satisfaction may also be multi-dimensional vis-a-vis performance and goal-fulfillment of salient attributes.

Third, the approach utilized suggests a need to address the temporal variance inherent in the satisfaction judgement process. Schematic representations (Matlin 1983) of products may evolve over time; judgement heuristics (Bettman et al 1992) or experience based norms (Woodruff, Cadotte and Jenkins 1983) may change over time. The ratings in this study were obtained after only 3-5 months of driving. It is likely, that the relative performance of various attributes and hence their contribution to goal-fulfillment will change over the course of many more months or years. Several managers in the auto-industry have asked the question: Which satisfaction rating should I worry about: the one customers report after using the car for 6 months, or the one customers report with the disposed vehicle, and just before they are ready to buy their next car? Obviously, satisfaction at each point in time is important and has strategic implications for researchers and managers alike. For instance, the satisfaction rating after disposal of the vehicle may be very salient and relevant to the next purchase decision (LeBarbera and Mazursky 1983) and has market share implications (Fornell and Wernnerfelt 1987). The satisfaction after 6 months is crucial because it may well be the summary judgement about a product that is spread via word-of-mouth (c.f. Richins 1983); this has obvious implications for new brands introduced in a market. Fourth, an attribute level view of customer satisfaction is better able to guide managerial resource allocation under a TQM approach. For instance, managers frequently ask the question: What attributes of the product should I enhance to improve CS? This approach provides guidance to managers regarding this issue.

Fifth, the findings suggest that pre-purchase choices may not lead the customer to buy products that will be most satisfying during consumption. Kahneman and Snell (1990) found that there was little or no correlation between the predictions of "hedonic change" that individuals made and the changes they actually experienced after using the product. They phrased the issues as follows: (1) Do people know what will be good for them and (2) do people choose what will be good for them? In particular, they note people's predictions about their own "hedonic experiences" were no better than the predictions made about a random stranger. Such findings, in the context of satisfaction, raise some serious policy and ethical implications. For instance, how might the pre-purchase decision environments be designed to sensitize customers to this sort of "miscalculation?"

LIMITATIONS

The reader should view the results as extremely preliminary and exploratory. Although the dataset used was obtained from "real" customers purchasing real products, it was originally collected for managerial purposes and has drawbacks associated with the practical limitations of such a data set. For instance, all measures utilized were single indicators, so no assessment of measures could be performed. A more complete matching of importance scores and performance scores would have been desirable. This would have enabled us to compare more attributes on their pre and post purchase importance. Also, because measures of pre-purchase importance were taken after purchase, they represent beliefs about pre-purchase importance and as such may not be fully representative of pre-purchase importance. Because the dataset was developed for managerial purposes, the attributes that were included were ones that had been found to be of interest to managers in previous studies. Thus, the attributes represent a mixed bag in the sense that they may not be truly comparable (Johnson, 1984). For instance, some attributes (cargo capacity) are more concrete while other attributes are really consequences of other attributes (eg. riding comfort may a consequence of wind noise, comfort of lap belt, ease of handling etc.). Finally, all the results are based on simple bivariate correlations and as such are somewhat tenuous.

Although the results are supportive, more conclusive empirical evidence for the propositions await. However, the discussion does point to examining CS as a construct that varies temporally across product attributes.

REFERENCES

Anderson, John R. (1990) Cognitive Psychology and its Implications. 3rd ed., New York: W.H. Freeman and Company.

Anderson, Rolph E. (1973) "Consumer Dissatisfaction: The Effect of Disconfirmed Expectancy on Perceived Product Performance," Journal of Marketing Research, 10, Feb., 38-44.

Bettman, James R., Eric J. Johnson, and John W. Payne (1992) "Consumer Decision Making," in Handbook of Consumer Behavior, Thomas S. Robertson & Harold H. Kassarjian (eds.) New Jersey: Prentice Hall.

Bolton, Ruth N. and James H. Drew (1991) "A Multistage Model of Customers' Assessments of Service Quality and Value," Journal of Consumer Research, 17, March, 375-384.

Churchill, Gilbert A., and Carol Surprenant (1982) "An Investigation into the Determinants of Customer Satisfaction," Journal of Marketing Research, 19, November, 491-504.

Craik, F.I.M. and R.S. Lockhart (1972) "Levels of processing: A framework for memory research," Journal of Verbal Learning and Verbal Behavior, 11, 671-684.

Danes, Jeffrey E. and Philippe Cattin (1980) "Multiattribute Choice Models: A Critical Review," in Advances in Consumer Research, vol. 8, Kent B. Monroe ed., Association for Consumer Research, Ann Arbor, MI.

Day, Ralph L. (1977) "Extending the Concept of Consumer Satisfaction," in Advances in Consumer Research, vol. 6, William D. Perreault, Jr., ed., Association for Consumer Research, Ann Arbor, MI, 149-154.

Day, George (1972) "Evaluating Models of Attitude Structure," Journal of Marketing Research, 9, August, 279-286.

Everitt, Brian S. and Graham Dunn (1992) Applied Multivariate Data Analysis. New York: Oxford Univ. Press.

Festinger, L. (1957) A Theory of Cognitive Dissonance. New York: Harper and Row.

Fornell, Claes and Birger Wernnerfelt (1987) "Defensive Marketing Strategy by Customer Complaint Management: A Theoretical Analysis," Journal of Marketing Research, 24, November, 337-346.

Johnson, Michael D. (1984) "Consumer choice strategies for comparing non-comparable alternatives," Journal of Consumer Research, 11, 741-753.

Johnson, Michael D. and Jerome M. Katrichis (1988) "The Existence and Perception of Redundancy in Consumer Information Environments," Journal of Consumer Policy, 11, 131-157.

Kahneman, Daniel and Jackie Snell (1990) "Predicting Utility," in R.M. Hogarth (ed.), Insights in decision making. Chicago: Univ. of Chicago Press.

LaTour, Stephen A. and Nancy C. Peat (1979) "Conceptual and methodological issues in consumer satisfaction research," in Advances in Consumer Research, vol. 6, William D. Perreault, Jr., ed., Association for Consumer Research, Ann Arbor, MI, 431-437.

LeBarbera Priscilla A. and David Mazursky (1983) "A Longitudinal Assessment of Consumer Satisfaction/Dissatisfaction: The Dynamic Aspect of the Cognitive Process," Journal of Marketing Research, 20, Nov., 393-404.

Matlin, Margaret W. (1983) Cognition. New York: CBS College Publishing.

Myers, James H. & Mark I. Alpert (1968) "Determining Buying Attitudes: Meaning and Measurement," Journal of Marketing, 32, Oct., 13-20.

Oliva, Terence A., Richard L. Oliver, & Ian C. MacMillan (1992) "A Catastrophe Model for Developing Service Satisfaction Strategies," Journal of Marketing, 56, July, 83-95.

Oliver, Richard L. (1980) "A Cognitive Model of the Antecedents and Consequences of Satisfaction Decisions," Journal of Marketing Research, 17, November, 460-469.

Oliver, Richard L. and Wayne S. Desarbo (1988) "Response Determinants in Satisfaction Judgements," Journal of Consumer Research, 14, March, 495-507.

Olson, Jerry C. and Philip Dover (1979) "Disconfirmation of Consumer Expectations Through Product Trial," Journal of Applied Psychology, 64, April, 179-189.

Richins, Marsha L. (1983) "Negative Word-of-Mouth by Dissatisfied Consumers: A Pilot Study," Journal of Marketing, 47, Winter, 68-78.

Swan, John E. (1988) "Consumer Satisfaction Related the Disconfirmation of Expectations And Product Performance," Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behavior, vol. 1, 40-47.

Westbrook, Robert A. and Richard L. Oliver (1991) "The Dimensionality of Consumption Emotion Patterns and Consumer Satisfaction," Journal of Consumer Research, 18, June, 84-91.

Wilkie, William L. and Edgar A. Pessemier (1973) "Issues in Marketing's Use of Multi-Attribute Attitude Models," Journal of Marketing Research, 10, November, 428-441.

Wilkie, William L. and Rolf P. Weinreich (1972) "Effects of the Number and Type of Attributes Included in an Attitude Model: More is not Better,' Proceedings. 3rd Annual Conference, Association for Consumer Research, 1972, 325-340.

Woodruff, Robert B., Ernest R. Cadotte, and Roger L. Jenkins (1983) "Modelling Consumer Satisfaction Processes Using Experience Based Norms," Journal of Marketing Research, 20, August, 296-304.

Yi, Youjae (1990) "A Critical Review of Consumer Satisfaction," in Review of Marketing 1990, Valarie A. Zeithaml, ed., Chicago: American Marketing Association, 68-123.

----------------------------------------