The nature of the service quality and satisfaction relationship

Empirical evidence for the existence of satisfiers and dissatisfiers

The Authors

Birgit Leisen Pollack, College of Business Administration, University of Wisconsin Oshkosh, Oshkosh, Wisconsin, USA

Abstract

Purpose – The aim of this paper is to challenge the assumption that the relationship between service quality attributes and customer satisfaction follows a classic linear path. It suggests the existence of a threshold level after which the strength of the relationship between the two constructs, for some service attributes, adjusts. Depending on the type of adjustment, service attributes can be classified as satisfier or dissatisfier. Each requires a distinct managerial action.

Design/methodology/approach – The data were collected via self-administered questionnaire from customers of three service industries (i.e. banking service, hairdresser service, phone service). The type of impact of nine individual service quality attributes on customer satisfaction was investigated using multiple regression models with interaction terms.

Findings – The empirical findings provide evidence for the existence of satisfiers and dissatisfiers. Satisfiers exhibit initially no relationship with satisfaction, but after the acceptable level of quality (i.e. inflection point) has been reached, become positively related. Dissatisfiers follow initially a positive relationship path with satisfaction but after the inflection point exhibit no relationship, or at best a significantly weakened one, with satisfaction. The relationship patterns were found to be service attribute as well as service type dependent.

Originality/value – The findings suggest that, for dissatisfiers, increasing service quality ad infinitum may not be the most prudent approach. Here, gains in satisfaction can no longer be achieved after the acceptable quality level has been reached. However, for satisfiers, service firms must leap over the threshold before gains in satisfaction can be realized. This may require large investments in quality improvements.

Article Type:

Research paper

Keyword(s):

Customer services quality; Customer satisfaction; Critical success factors; Supplier relations; Expectation.

Journal:

Managing Service Quality

Volume:

Number:

Year:

2008

pp:

537-558

Copyright ©

Emerald Group Publishing Limited

ISSN:

0960-4529

Introduction

For over two decades, researchers have been studying service quality and its relationship to critical business outcomes. To this end, a number of studies have addressed the relationship between service quality and customer satisfaction and it is generally believed that higher levels of service quality lead to higher levels of customer satisfaction (Gotlieb et al., 1994; Kang and James, 2004; Oliver, 1997). While, most commonly this relationship is viewed and treated as linear, evidence from two distinct streams of literature suggests alternative relationship patterns. The first stream evolved from the zone-of-tolerance theory (see Zeithaml et al., 1993; Parasuraman et al., 1994) and suggests the existence of two threshold levels after which the strength of the relationship between the two constructs adjusts (Zeithaml et al., 1996). However, given that empirical tests seem to only support the threshold at the minimum acceptable level of service quality (Teas and DeCarlo, 2004; Yap and Sweeney, 2007), the theory in its current form inadequately describes the linkage. But, even a revised theory that includes only one significant breakpoint, casts doubt on the linear nature of the relationship between quality attributes and satisfaction.

The second stream of literature is largely based upon qualitative investigations and springs from Herzberg's two-factor theory of employee satisfaction with work place attributes (Herzberg, 1968). According to this theory, quality attributes can be classified into two categories: those that, if present, create satisfaction and those that, if not present, create dissatisfaction. Herzberg's theory has been extended into the product quality literature (e.g. Swan and Combs, 1976) and used to gain initial qualitative insights in the service quality field (e.g. Johnston, 1995). This stream of literature has several shortcomings. First, Herzberg's theory is largely based upon qualitative insights gained from critical incidence and, as such, has not been subjected to empirical tests (Vargo et al., 2007). Second, the fact that a number of incidences cannot be classified into the dichotomy has not been adequately addressed. Finally, the identified service attributes are usually service specific and, hence limited in usefulness for the finding of an overarching theory.

Given that “nonlinearity reflects the idea that there may be diminishing returns from progressive units of improvement” (Maddern et al., 2007, p. 1001), the present study begins to combine the two streams of literature into a more unifying framework for empirically studying the potential nonlinear nature of the quality – satisfaction linkage. As was pointed out by Vargo et al. (2007), such nonlinear models are currently under-researched on not well developed.

This research briefly reviews the two theories that question the linearity of the service quality/satisfaction relationship and synthesizes the findings. The study then seeks to extrapolate these findings into the service quality domain via empirical assessment. To this end, the research suggests that the relationship between service quality and satisfaction follows three distinct patterns. It is proposed that some service quality attributes follow initially a horizontal path (i.e. increased levels of quality on this attribute do not lead to increased levels of satisfaction), but after an inflection point (i.e. threshold) exhibit a positive linear relationship. Such service quality attribute is classified as a satisfier. Conversely, some service attributes follow initially a positively sloped linear path, but after an inflection point become unrelated to customer satisfaction. Here, the service quality attribute is classified as a dissatisfier. And finally, for some service attributes the path follows the classic linear pattern over the entire data range. Service attributes following this path are classified as criticals. This research further explores whether or not the nature of the relationship varies by the type of service.

This research provides a significant contribution to the services literature. It extrapolates findings from various fields of study into the service literature and, if confirmed, provides empirical evidence for the existence of the three distinct relationship patterns between service quality attributes and customer satisfaction. From a managerial perspective, the present research provides evidence that allows us to question whether service quality enhancements are wise investments under all circumstances. In fact, recommendation to do so may be misleading. The research allows service providing organization to identify service attributes were minimum levels have to meet in order to avoid dissatisfied customers. Here, however, additional investments, above and beyond the acceptable level, may provide diminishing returns of the service quality enhancements. To contrast, the research also allows for the identification of attributes were service providing organization have the opportunity to delight customers by providing high levels of quality on these attributes.

Theoretical framework

Service quality and customer satisfaction

Service quality is “the delivery of excellent or superior service relative to customer expectations” (Zeithaml and Bitner, 1996, p. 117). Service quality is recognized as a multi-dimensional construct. While the number of dimensions often varies from researcher to researcher, there is some consensus that service quality consists of three primary aspects: outcome quality, interaction quality, and physical service environment quality (Rust and Oliver, 1994; Brady and Cronin, 2001). Outcome quality refers to the customer's assessment of the core service which is the prime motivating factor for obtaining the services (e.g. hair cut, money received from ATM). Interaction quality refers to the customer's assessment of the service delivery process, which is typically rendered via a physical interface between the service provider, in person or via technical equipment, and the customer (Grönroos, 1984). It includes, for instance, the consumer's evaluation of the attitude of the service providing staff. The physical service environment quality dimension refers to the consumer's evaluation of any tangible aspect associated with the facilities or equipment that the service is provided in/with. It includes, for example, the architectural design features of a hair salon or the physical conditions of an ATM machine.

A number of researchers further elaborate on sub-aspects of these three broad dimensions. For example, the most popular conceptualization of service quality – SERVQUAL – features five dimensions: tangibles, reliability, responsiveness, empathy and assurance (Parasuraman et al., 1988). The tangibles dimension correspond to the aforementioned physical environment aspect, the reliability dimensions corresponds to the service outcome aspect, and the remaining three represent aspects of interaction quality.

Building upon and synthesizing previous work, Brady and Cronin (2001) advanced the hierarchical conceptualization of service quality. Here, service quality is a multilevel construct consisting of the aforementioned three sub-dimensions. The interaction quality dimension, in turn, is comprised of attitude, behavior, and expertise of the service provider. The physical environment quality dimension is comprised of ambient conditions, design, and social factors of the service facilities. The outcome quality dimension is comprised of waiting time, tangibles, and valence.

Service quality and customer satisfaction are distinct but related constructs (Spreng and Mackoy, 1996). Oliver (1997, p. 13) defines satisfaction as “the consumer's fulfillment response”, a post consumption judgment by the consumer that a service provides a pleasing level of consumption-related fulfillment, including under- or over-fulfillment. Service quality is an antecedent of the broader concept of customer satisfaction (Gotlieb et al., 1994; Buttle, 1996; Zeithaml and Bitner, 1996; Lee et al., 2000) and the relationship between service quality and loyalty is mediated by satisfaction (Caruana, 2002; Fullerton and Taylor (2002). Most commonly, the nature of this service quality and satisfaction link is viewed as linear, indicating that higher levels of service quality lead to higher levels of satisfaction. However, two streams of literature suggest potential nonlinear paths for at least some service quality attributes. The streams of literature are reviewed and synthesized next.

Zone-of-tolerance theory

The zone-of-tolerance (ZOT) theory conceptualized by Parasuraman et al. (1991) suggests that service quality perceptions are dependent upon consumers' expectations. The authors identify two levels of expectations: adequate and desired (Figure 1A). Both represent service expectation levels falling on a continuum extending from low expectation on one end to high expectation on the other extreme. The adequate service level represents what “the customer” finds acceptable. It is in part based on the customer's assessment of what the service “will be”, that is, the customer's “predicted service” (Parasuraman et al., 1991, p. 42). The desired service level represents the “service the customer hopes to receive. It is a blend of what the customer believes “can be” and “should be” (Parasuraman et al., 1991, p. 42). Between the desired and adequate level is a zone of tolerance. The ZOT is said to vary for the various service attribute. The authors suggest that service quality perception falling below the adequate level result in a competitive disadvantage. Perceptions falling in the ZOT result in a competitive advantage and offer a potential for differentiation. Perceptions falling beyond the desired level lead to customer franchise.

The initial conceptualization of the ZOT model is based on focus group results and on an analysis of varying views and definitions of expectations in the customer satisfaction literature (Zeithaml et al., 1993; Parasuraman et al., 1994). It identifies desired and adequate service levels but does not address the implications of the ZOT on outcomes of service quality to include its most prominent consequence of customer satisfaction.

Later, Zeithaml et al. (1996) investigate the link between the ZOT model and a set of behavioral consequences (Figure 1B). The extended theory suggests that customer sensitivities, expressed in a set of behavioral consequences, to service quality variations change across the data range. This change in sensitivity occurs at two thresholds; the first reflects the customer's acceptable level of service quality and the second reflects the desired quality level. Hence, the service quality/behavioral consequences relationship can be broken down into three zones with three distinct slopes. At first, the relationship is positive, the second zone may have a zero slope or a slightly positive one, and the third zone again exhibits a positive relationship. The ZOT is located between the two thresholds and reflects a range of service quality performances that the consumer finds acceptable. The authors affirm that a firm operates at a competitive disadvantage if the quality perceptions are below the acceptable level and, further, that improvement has a greater impact on the customers if their perceptions are currently outside the ZOT, that is if firms seek to exceed the desired or below adequate level of quality.

From a theoretical perspective, the three zones exhibit differing sensitivities with respect to key service quality outcomes. The relationship between service quality and its consequences is said to be significantly greater above and below the ZOT compared to within (Zeithaml et al., 1996; Yap and Sweeney, 2007). However, empirical tests seem to support only the breaking point at the acceptable level (Figure 1C). Zeithaml et al. (1996) investigated the ZOT for four different industries. Although they did not link service quality to satisfaction but used the correlated loyalty construct as one of their dependent variables, for three of the industries they found only one threshold level, for the other industry they did not find any. Thus, while they did not find empirical support for the two-threshold model, they found support for a one-threshold model. Similarly, Teas and DeCarlo (2004) testing the ZOT model by linking service quality to satisfaction also found only one threshold. The empirical evidence suggests that this threshold level rests at the acceptable service quality level. They conclude that:

[…] no evidence was found indicating that, when compared to the situation in which performance is perceived to be adequate, the relationship between quality and satisfaction is different when perceived performance is above the desired service level (Teas and DeCarlo, 2004, p. 279).

Yap and Sweeney (2007), in their investigation of the ZOT model, also found only one threshold, separating the slope below the zone of tolerance, and the slope within the theoretical zone. Their findings suggest the existence of a significant positive relationship below the acceptable level for the two service quality dimensions they investigated. They also found that this relationship is not quite as strong above the acceptable level of service quality. They concluded that:

[…] the relationship of service quality with satisfaction … did not vary significantly from the case when a customer's perceptions were within the ZOT to when they exceeded desired service expectations (Yap and Sweeney, 2007, p. 143).

Thus, while the empirical evidence to date fails to support the two-threshold ZOT model, the findings support the existence of one threshold. The evidence for even this one significant breakpoint casts doubt on the linear nature of the relationship between quality attributes and satisfaction.

Herzberg's two-factor theory

The second stream of literature springs from Herzberg's (1968) theory on work place motivation and is based upon qualitative data. Herzberg suggests the existence of two distinct types of work place attribute/job satisfaction relationships. One set of work place quality factors is associated with satisfaction (e.g. recognition, achievement). The stronger the perceived presence of these attributes, the higher the satisfaction. Herzberg termed these factors motivators. The absence of these factors will not lead to dissatisfaction but only to no satisfaction. The second set of factors termed hygiene factors, are associated with dissatisfaction (e.g. company policy and administration, work conditions). The lower the quality rating on these attributes, the higher the dissatisfaction. To contrast, high quality ratings on those factors, according to Herzberg only result in no dissatisfaction but not in satisfaction. Herzberg also found a number of overlapping factors that did not fit the strict motivators and hygiene factor dichotomy but did not further elaborate on those quality attributes.

Since its inception, Herzberg's theory has been extended into the product quality literature and used to gain initial qualitative insights in the service quality field. Based upon findings using the critical incident technique (CIT), Swan and Combs (1976) suggests the existence of two types of product quality attributes which they termed satisfactory items which correspond to Herzberg's motivators and dissatisfactory items which correspond to Herzberg's hygiene factors. While Herzberg's work is based upon a rather discrete felt presence or absence of a quality attribute, Swan and Combs introduced the concept of a more continuous level of customer evaluations of quality attributes. Here, satisfactory items are associated with performance outcome equal or above expectations rather than just the presence of an attribute. Similarly, dissatisfactory items are associated with outcomes below expectations rather than just with the mere absence of the quality aspects. Swan and Combs also recognized that customer evaluation of product quality is a multi-attribute evaluation on salient items. Similar to Herzberg, Swan and Combs also found a number of factors that did not follow the strict satisfactory/dissatisfactory dichotomy. Maddox (1981), in replicating Swan and Combs work, comments on the number of attributes that could not be classified as either satisfactory nor dissatisfactory by suggesting the existence of another category of factors that follow the classic linear path with satisfaction and dissatisfaction as the anchors on a bipolar continuum.

Extrapolating Herzberg's theory into the service quality domain, Johnston (1995) analyzed qualitative critical incidence statements of bank customers. The incidence were initially grouped into 18 predetermined service quality categories and then classified into what he termed satisfiers and dissatisfiers. The satisfiers correspond to Herzberg's motivators and the dissatisfiers correspond to Herzberg's hygiene factors. His findings suggest the presence of both – satisfiers and dissatisfiers – for the sampled industry. For instance, the commitment of the service staff to their work and the cleanliness of the service facility were classified as satisfiers, whereas the integrity with which the customer is treated and the aesthetics of the service environment were classified as dissatisfiers. Similar to Herzberg, Johnston, with only four attributes that could be exclusively classified as either satisfier or dissatisfier, also found a number of overlapping factors.

Chowdhary and Prakash (2005) identified vantage factors, which correspond to Herzberg's motivators, and qualifying factors, which correspond to the Herzberg's hygiene factors. The authors describe the vantage factors as order winning or sought after factors. Increasing the level of those has the ability to delight the customer. To contrast, the presence of qualifying factors is considered an essential part of services (e.g. cleanliness) and, if absent, have the greatest potential to drive customers away. However, Chowdhary and Prakash (2005) further suggest that the qualifying factors only have to be at a level perceived as adequate by the consumer and that ratings of above this adequate level do not translate into additional gains in overall quality.

Expanding the number of factors to account for the attributes that did not fit the satisfier/dissatisfier dichotomy, Schvaneveldt et al. (1991) identified four major categories of factors which they termed attractive quality elements, one-dimensional quality elements, must-be quality element, and indifferent quality elements. Corresponding to a satisfier, an attractive quality element is a quality attribute that if present leads to satisfaction, but, if absent, does not lead to dissatisfaction. According to the authors, such elements primarily focus on extras that go beyond the core service such as a nice interior design. The authors suggest their use in differentiation from competing offerings. The one-dimensional quality elements give satisfaction when present and dissatisfaction when absent and fit the conventional linear model. Corresponding to a dissatisfier, the must-be-quality element does not create satisfaction when present but dissatisfaction when absent. The authors view the delivery of these attributes without failure as essential to preventing dissatisfaction. The indifferent quality element does not lead to either satisfaction or dissatisfaction when absent and seems to be not important in a consumer's evaluation of quality. According to the authors, the presence of these attributes may suggest either poor targeting and/or a misdirection of effort. Such an element is similar to the one-dimensional quality element; it follows a linear path but without extreme values. Although, the authors conclude that the classification into one of the four categories is attribute specific and target segment specific. They further observed that quality attributes that are related to performance or service outcome dimension are frequently judged as one-dimensional, whereas attributes more associated with the service process dimension such as assurance are judged as must-be-quality elements (i.e. dissatisfier). Attributes that are considered to go beyond the core service are judged as either attractive (i.e. satisfiers) or indifferent. The authors most frequently observed the one-dimensional relationship.

Cadotte and Turgeon (1988), in applying CIT in studying the hospitality industry, also list four factors, which they termed satisfiers, dissatisfiers, criticals, and neutrals. Criticals are similar to the aforementioned one-dimensional quality elements in that, depending on the performance level, they affect both satisfaction and dissatisfaction. Again, these are reflective of the classic linear relationship. Neutrals are similar to aforementioned indifferent quality elements in that they seemingly have no impact on satisfaction regardless of performance level. Similarly, Kueh (2006), also using the CIT identified the identical four types of relationships across a range of service types. The author classified an attribute as critical if it was identified as both a top satisfier and top dissatisfiers. Again neutrals, according to the author are items with relative little impact on either satisfaction or dissatisfaction. The author attributed this lack of impact to the non-saliency of these attributes to the studied industry.

In sum, this stream of literature identifies, based upon critical incidence, three types of service quality relationships for salient service quality attributes. Based upon the nature of the relationship, service quality attributes are referred to as satisfiers, dissatisfiers, and criticals.

Synthesized findings

Based upon the above review of the two streams of literature, three distinct relationship patterns between service quality attributes and customer satisfaction can be identified. Empirical tests of the ZOT theory provide evidence for the existence of satisfiers (Figure 1C). The stream of literature that evolved from Herzberg's theory exclusively relies on the rather subjective qualitative critical incidence technique (Maddox, 1981) and as such unearthed interesting preliminary findings as opposed to empirically tested ones. These preliminary findings suggest:

the existence of at least three types of relationship: satisfiers, dissatisfiers, and criticals;
the pivotal role the acceptable level of service quality plays in distinguishing between satisfiers and dissatisfiers; and
that classification into anyone of the categories is service type specific.

Thus, combined the theories suggests the existence of one threshold level after which the strength of the relationship between quality and satisfaction adjusts. Further, the strength before and after this threshold depends on the quality attribute. Therefore, depending on whether the strength decreases or increases after the threshold, different relationship patterns between the various quality attributes and satisfaction emerge. These relationships are graphically portrayed in Figure 2.

In support of the existence of dissatisfiers, Swan and Combs (1976) argue that:

[…] consumers judge products on a limited set of attributes, some of which are relatively important in determining satisfaction, while others are not critical to customer satisfaction but are related to dissatisfaction when performance on them is unsatisfactory.

As such, dissatisfiers are determinants of dissatisfaction and may create complaining behavior. These service attributes follow initially a positive relationship path with satisfaction but after an inflection point exhibit no relationship, or at best a diminished positive one, with satisfaction (see Figure 2). As suggested by qualitative evidence, such empirically verifiable dissatisfiers may also exist in the quality of services. For instance, as aforementioned, Johnston (1995), in a qualitative study concludes that for banking customers the integrity of service employees are dissatisfiers. Once the customer perceives the integrity to be at the acceptable level, satisfaction levels ff. Johnson (1997, p. 113) adds that:

[…] the main source of dissatisfaction appear to be cleanliness, aesthetics, integrity, functionality, reliability and security which are associated with either the more tangible aspects of service or systemic issues.

Rahman (2004) identified the safe handling of customer information as a dissatisfier in the banking sector.

In support of the existence of satisfiers, Swan and Combs (1976) state that “satisfaction will tend to be associated with expressive outcomes (i.e. satisfier) above or equal to expectations”. Maddox (1981) states that “low values on an expressive attribute will reduce satisfaction, but will not lead to dissatisfaction”. Indicative of the absence of a relationship, satisfiers follow initially a horizontal path but after an inflection point exhibit a positive linear relationship. Satisfiers do not create dissatisfaction when absent, but when provided create positive gains in satisfaction (Figure 2). Such satisfiers may also exist in the quality of services. Here, again Johnston (1995), based upon qualitative evidence, found that the commitment of banking service employees to be a satisfier. Johnston (1997, p. 113) states that “satisfiers …tend to be concerned more with the intangible nature of the service, commitment, attentiveness, friendliness, care and courtesy”. Rahman (2004) identified the willingness to help customers by bank employees is viewed as a satisfier.

The current literature on service quality suggests the existence of criticals. These are service attributes that follow the classic linear path. That is higher levels on these service quality aspects lead to higher levels of customer satisfaction (Figure 2). In studying a linear linkage between quality attributes and satisfaction, Levesque and McDougall (1996) confirmed that the more abstract dimensions of outcome and process quality are important components of quality in banking. Stafford (1994) lists friendliness, courtesy, accuracy of transactions and records along with convenient hours as the top four service quality aspects in banking. At the more abstract level these translate into process (courtesy, friendliness), outcome (accuracy), and tangibles (e.g. convenient hours) dimensions.

In sum, the literature suggests:

the existence of a threshold level in the relationship between service quality and satisfaction after which the intensity of the relationship changes;
the nature of the service quality/satisfaction relationship before and after the threshold level follows either one of two distinct patterns – satisfiers or dissatisfier;
at least one of the service's attributes that make up service quality follows the path of a satisfier and at least one follows the path of a dissatisfier; and
at least one service attribute follows the critical relationship path.

This research seeks to empirically classify service quality attributes as criticals, satisfiers, and dissatisfiers. As such, the research seeks to empirically address the following general proposition:

P1. Not all of the service quality dimensions follow a linear path (i.e. are criticals) and, consequently, some dimensions can be classified as satisfier or dissatisfier.

The intent of the present study is to empirically verify grounded speculations on the existence of satisfiers, dissatisfiers, and criticals across service industries. However, the study does not speculate which service quality aspects follow the three paths. In fact, such paths are likely service type specific. Chowdhary and Prakash (2007) firmly concluded that the relative importance of service quality aspects is service dependent. They observed that the tangibles service quality dimension seems to be more important to services with some tangible outcomes such as beauty saloons and reliability seems to be more important for services with rather intangible outcomes such as telephone companies. Lee et al. (2000) found that tangibles were found to be more important to facility/equipment-based industry and aspects of interaction quality were found to be more important to people-based services. Johns and Howard (1998) using the critical incident technique identified satisfiers and dissatisfiers in the fast food service. Their findings suggest that the identified attributes differ from those identified by Johnson's banking study. Lassar et al. (2000) report that importance of quality dimensions to satisfaction vary and that therefore, certain service quality conceptualization may be better suited for specific services. They suggest that interaction quality (i.e. functional quality) and outcome quality (technical quality) model is better suited for banking because both dimensions were found to significantly influence satisfaction. Whereas most dimension of an alternative conceptualization failed to be significant predictors. Maddern et al. (2007), in their investigation of the impact of outcome (technical) and process (functional) quality found empirical support only for the outcome quality – satisfaction relationship in the financial services sector in the UK.

Methodology

Samples and data collection

The data for empirically addressing the research proposition were collected from convenience samples of 250 US customers of hairdresser/barber services representing a more people-based service, 300 US customers of local phone service subscribers representing a more equipment-based service, and 150 US customers of banking services representing a mixed service. Although the sample size for the banking service falls below the typically suggested number of 200 (i.e. Bearden et al., 1982) for evaluating structural equation models, it lies at the minimum suggested by Anderson and Gerbing (1988) for evaluating measurement models which is the purpose here. Students at a Mid-western University were given extra credit for distributing self-administered questionnaires to customers of hair dresser/barber services and local phone service subscribers. As such, several firms within each of the three service sectors are represented in the samples. The students were instructed to provide the names and phone numbers of the survey participants so that random verification checks can occur. A similar procedure was also used by Brady and Cronin (2001). For the banking service a student sample was used. The hairdresser/barber service sample consists of 47.6 percent males and 52.4 percent females with an average age of 29.0 years. The phone service sample consists of 53.6 percent males and 46.4 percent females with an average age of 30.8. The banking services sample consists of 51.4 percent males and 48.6 percent females with an average age of 22.5.

Measures

The questionnaire included 27 service quality items adopted from Brady and Cronin (2001). This measure assesses nine service quality attributes representing the three sub-dimensions of interaction quality, outcome quality, and physical environment quality. It measures the study participants' perceptions of the actual service performances along the various attributes. It did not include an assessment of service expectations, which is needed to yield the acceptable service level. The acceptable level of service was determined via focus group studies, which are described in the following section of this article.

The questionnaire also contained three questions related to the respondents' overall satisfaction (α=0.93 for hairdresser, α=0.92 for banking service, α=0.95 for phone service). All items were measured with a seven-point Likert-type scale (1=strongly disagree, 7=strongly agree). The results of computations of Cronbach alpha reliability coefficients along with measurement model fit statistics as well as basic descriptive statistics for the service quality items are presented in Table I.

Indicative of an acceptable fit of the hypothesized model and the observed data, the fit statistics for the measurement model generally follow the cut-off values suggested by Hu and Bentler (1999). That is the NNFIs, CFIs, IFIs are generally above 0.95, the RMSEAs are close to 0.06, and the SRMRs are close to 0.08. Most alpha coefficients are at acceptable levels and fall between 0.8 and 0.9 with only a few, primarily for the banking service hovering around or fall below the lower bound of acceptable values suggested by Nunnally (1978). In general, the Cronbach alpha coefficients are higher for the hairdresser/barber service than for the mixed service and the phone service. For some of the physical environment quality sub-dimensions the discrepancies in alpha coefficients for the three services are quite large. For example, the alpha values for the social factors attribute extends from only 0.41 for the banking services to 0.72 for the phone service. This may indicate that the reliabilities of the various sub-dimensions of the Brady and Cronin scale are service dependent. More specifically, it may be indicative of varying levels of importance or saliency of the individual quality aspect and, therefore, should be further investigated in subsequent studies.

Qualitative assessment of acceptable service quality levels

To establish a typical acceptable level of service quality for each of the nine service quality attributes, three focus group sessions with eight customers of each sampled service industry were conducted. The group addressing the phone service consisted of three females and five males. One person was an older retired individual, two were in their twenties, and the remaining five were middle aged. The group addressing the hairdresser/barber service consisted of four males and four females, three were in their early twenties, four were middle aged, and one neared retirement. The group addressing the banking service consisted of four female and four male students in their early twenties. The compositions of the focus groups for the three industries were fairly similar in gender and age to the above described larger samples for these industries.

The participants were presented with the nine service quality dimensions and asked to specify on a seven-point Likert-type scale the level of service they deem acceptable. For the hairdresser/barber service and the phone service, the acceptable levels of quality on the nine dimensions fell after the mid-point of the scale indicating that ratings of four and below would not be acceptable. For the banking service, the acceptable levels on the nine sub-dimensions fell at the above five rating indicating that ratings of five and below would not be acceptable. The focus group participants were fairly consistent in their ratings of acceptable service levels. The acceptable service level findings for this study are consistent with those identified by Teas and DeCarlo (2004) who also found the typical acceptable level of quality to lie after the mid-point of their scale.

Results and discussion

To test the advanced hypotheses, the following general regression model was formulated: Equation 1 where:

Sat =customer satisfaction score.

SQ =service quality sub-dimension score.

d =dummy variable with a value of 1 if above acceptable level of quality on sub-dimension, 0 otherwise.

βs =standardized regression coefficients.

ɛ =error term.

The β ₁ coefficient represents the slope below the acceptable quality level and can be viewed as the average effect of the service quality dimension on satisfaction (Cohen and Cohen, 1983; Jaccard et al., 1990). The β ₃ coefficient represents the additional contribution to satisfaction as the value for the service quality attribute rises above the acceptable level. Thus, β ₁ ⁺ β ₃ represents the slope above the acceptable quality level. A total of 81 regression models were formulated. The standardized beta coefficients, adjusted R ² values, and the classification as criticals, satisfiers, and dissatisfiers are reported in Table II.

A service quality sub-dimension was classified as critical if the standardized beta coefficient for the dimension is significantly different from zero and positive and the standardized beta coefficient for service quality sub-dimension/dummy variable interaction is not significantly different from zero. A service quality sub-dimension was classified as satisfier if the standardized beta coefficient for service quality sub-dimension is not significantly different from zero and the standardized beta coefficient for service quality sub-dimension/dummy variable interaction is significantly difference from zero and positive. Since this study is exploratory in nature, service quality dimensions that exhibit a positive and significant beta coefficient for the service quality sub-dimension and a positive and significant beta coefficient for the service quality sub-dimension/dummy variable interaction are also classified as satisfiers. Here, although the relationship is significant before the acceptable level has been reached, it is significantly more pronounced afterward. A service quality sub-dimension was classified as a dissatisfier if the standardized beta coefficient for the dimension is significantly different from zero and positive and the standardized beta coefficient for service quality sub-dimension/dummy variable interaction is negative. Here, the relationship between quality and satisfaction is significantly less pronounced after the acceptable level of service quality is reached.

The findings for the more people-based service (i.e. hairdresser/barber service) suggest that all three sub-dimensions of interaction quality are criticals. The main effects of attitude, behavior, and expertise are significant and positive whereas the interaction term is non-significant. Data plots generated for attitude–satisfaction, behavior–satisfaction, and expertise-satisfaction substantiated a linear relationship extending over the entire data range. The physical environment quality sub-dimensions produced mixed results. Ambient conditions seem to follow the path of a satisfier. The main effect was non significant but the interaction effect was significant and positive. This suggests that consumers' satisfaction ratings are not affected at lower levels on this sub-dimension but increase with increasing levels of perceived quality after the acceptable level has been reached. The average ambient conditions seemingly do not lead to either satisfaction or dissatisfaction but an above par ambience increases the satisfaction level. To contrast, the design sub-dimension seems to follow the path of a dissatisfier. Here, the main effect was significant and the interaction effect was significant and negative. This suggests that after an acceptable level of quality is reached on the design dimension, the gains in satisfaction when increasing the quality perception is diminished. The outcome quality sub-dimensions exhibits only the presence of criticals. The waiting time, tangibles and valence attributes were found to be criticals. In general, higher levels on these critical core service features lead to higher levels of customer satisfaction.

The results for the banking service suggest that all three sub-dimensions of interaction quality follow the path of a satisfier. Here, the main effects were non-significant, suggesting the absence of a relationship up to the acceptable level of quality. The interaction effects were significant and positive; suggesting a positive relationship after the acceptable level of quality has been reached. The findings were also confirmed by data plots of the individual sub-dimensions and satisfaction. Thus, lower perceived levels of quality seemingly do not affect satisfaction, but customer satisfaction increases by performing beyond the threshold level. The results for the physical environment sub-dimensions suggest that ambient conditions and design follow the classic linear path. The main effect was significant and positive whereas the interaction effect was non-significant. These two sub-dimensions were classified as criticals. The social sub-dimension revealed the absence of a relationship between the social aspects of quality and satisfaction for the banking service. The findings for the outcome quality dimensions suggest that waiting time and valence are criticals. Here, the main effect is significant and positive but the interaction effect is non-significant. The tangibles sub-dimension follows a somewhat unique path; the main and interaction effect are significant and positive. This suggests a positive relationship between tangibles and satisfaction up to the acceptable level and an even more pronounced positive association thereafter. Although this attribute is not prototypical of a satisfier, it resembles most closely the satisfier relationship and was therefore classified as such.

The findings for the more equipment-based service (i.e. phone service) suggest that two of the three sub-dimensions of interaction quality can be classified as criticals. These are attitude and expertise. The main effect is significant and positive and the interaction effect is non significant. Thus, the higher the perceived quality on these dimensions the greater the satisfaction level. The behavior sub-dimension follows the path of a dissatisfier. Here, the main effect is significant and positive and the interaction effect is significant and negative. This suggests that at below acceptable service levels increases in quality lead to higher satisfaction levels. However, this relationship is not as strong after the acceptable level has been reached. The findings for the physical environment and outcome quality dimension suggest linear relationships over the entire data range.

In sum, the findings provide evidence for the existence of alternative relationship patterns between the various service quality sub-dimensions and satisfaction. About 26 percent out of 27 investigated relationships across three service industries exhibit patterns that divert from the classic linear path.

Conclusion and implications

This study provides empirical support for the existence of non-linear relationships between some service quality attributes and satisfaction. Four general conclusions can be drawn from the present findings. First, two types of non-linear patterns are identified. Based upon the intensity of the quality/satisfaction relationship before and after the acceptable quality level, following conventional terminology, the quality attributes are classified as satisfier or dissatisfier. Second, a nonlinear path exists for some service attributes but not for others and as such the path is service attribute specific. Third, whether or not a specific service quality attributes follows a non-linear path is service type specific. Kueh (2006) also suggests that the nature of service quality dimension – satisfaction relationship is service type dependent. However, each of the investigated service types was found to have at least one “non-linear” attribute. Four, a large number of service quality attributes are criticals. The qualitative evidence springing from Herzberg's theory maybe misleading in that it suggests the existence of primarily satisfiers and dissatisfiers. To this end, the present study raises serious doubts, since many aspects where found to follow the classic linear path often assumed in quality/satisfaction research. Criticals exhibit no plateau and do not suffer from diminishing returns as a result of increasing their respective performances. As such, the existence of criticals suggests that minimum acceptable service levels are irrelevant to a customer's satisfaction rating on these attributes. That is the strength of the satisfaction link does not change after the minimum level has been reached. This suggests that for many service attributes satisfaction and dissatisfaction lie on a continuum with both extremes as anchors.

These findings have significant managerial implications for service quality management. Instead of advising firms to increase the quality levels on all attributes they should first investigate the nature of the relationship for each service attribute. Then they should adopt different strategies for satisfiers and dissatisfiers. To this end, the present research suggests that investment in service quality does not necessary lead to higher levels of customer satisfaction. In case of the satisfier, the findings reveal, that if a firm finds itself operating at the lower end of the quality spectrum on a satisfying attribute, a little improvement will not affect customers' satisfaction levels. Here substantial investments in quality improvements need to be made in order to move consumers beyond the threshold level after which satisfaction gains can be made. For example, the present research finds the expertise of employees in the banking sectors to be satisfiers. If a customer perceives this expertise to be well below the acceptable level, satisfaction gains cannot be made with only small improvements. In fact, substantial investments in training programs or hiring more qualified personnel at a higher salary level may become necessary.

For certain attributes, the return on quality may diminish after the acceptable level is reached. After this point, service firms may not achieve great gains in customer satisfaction. Given scarce resources, service firms may want to rather allocate their monies for quality improvements to the satisfiers. For instance, the present study identified the design of the hairdresser/barber shop facility as a dissatisfier. If a customer perceives the design as lacking, satisfaction will be greatly affected. However, moving beyond the acceptable level by, for example, installing Italian marble floors will not leave the customer more satisfied. However, Johnston (1997) advises that if ratings for dissatisfiers fall below acceptable levels, they should be addressed first before improving on satisfiers.

As with any research, this study is subject to several limitations. First, a rather generic non-industry specific scale for assessing service quality was adopted. The 27-item tool taps into three overarching service quality dimensions, which are said to be universal to all services. While this was deemed necessary for identifying potential empirical generalizations across industries, a generic scale somewhat neglects attribute saliency. While this is appropriate for theory development and enhanced understanding of the intricacies of the quality/satisfaction relationship, individual service firms may want to adjust the scale to fit their more service specific quality features. For instance, it is unlikely that a consumer uses twenty plus salient attributes in his or her evaluation of a service. Studies collecting qualitative critical incidence statements found on average three or four positive ones and two to three negative ones. Thus, the use of five or six salient attributes is perhaps more realistic.

Related to the aforementioned, future studies may want to investigate the relationship between more industry specific service attributes falling under the three umbrella dimensions and customer satisfaction. This helps to assure that salient service specific attributes are being investigated rather than the more global constructs which may lack relevance in specific industry contexts. In fact, a number of more service specific assessment tools (e.g. banking and information systems) already exist. From a managerial perspective, it may be more meaningful to apply these scales when identifying satisfiers and dissatisfiers for specific industries.

Additionally, the present study does not take into consideration target market specific service attributes. The importance of service attributes to satisfaction may vary by customer segment. For one segment, an attribute may be a satisfier while for a second, the same attribute may be a critical or even irrelevant:

In the retail-banking sector, customers who use particular products (e.g. loans or mortgages) may focus on service features, such as competitive interest rates, more than customers who do not hold these product. Thus, the determinants of customer satisfaction towards the service provider may vary depending on customer or segment characteristics (Levesque and McDougall, 1996, p. 14).

Also, the present study does not address the behavior of satisfiers, dissatisfiers, and criticals over time. Rahman (2004) noted that due to changing customer expectations, a classification into one of the three categories is likely to change over time. For example, criticals may become satisfiers and vice versa or satisfiers may become dissatisfiers. A similar observation was also made by Chowdhary and Prakash (2005).

Another limitation is the use of a student sample for the banking sector. The experiences and expectations of student, however, may not necessarily reflect those of the population as a whole. The question of being representative of that larger population arises also for the hairdresser and local phone service samples since both exhibit a relative low average age. Therefore, subsequent studies may want to replicate this research with additional samples.

Finally, this study's purpose was to identify possible deviation from the classic linear relationship between quality attributes and satisfaction. Future research may enlarge the model by incorporating financial performance indicators as outcome variables. For example, future studies may want to address the relationship between the three service quality – satisfaction linkages and revenue or profit measures.

Equation 1

Figure 1ZOT theory and empirical support

Figure 2Alternative service quality/satisfaction relationships deerived from Herzberg's theory

Table IDescriptive statistics of HSQM items and measurement model results

Table IIRegression results and classification of service quality sub-dimensions

References

Anderson, J.C., Gerbing, D.W. (1988), "Structural equation modeling in practice: a review and recommended two-step approach", Psychological Bulletin, Vol. 103 No.3, pp.411-23.