Introduction

Marketers constantly strive to achieve loyal, committed customers, but few companies are ultimately successful. Thus, a better understanding of loyalty, its effects, and an improved measurement model is needed. The current research creates unique measures of loyalty based on Oliver’s (1997, 1999) conceptualization of loyalty as differing cognitive-affective-conative-action levels. Specifically, the conceptualization identifies consumers as either cognitive, affective, conative, or action loyal to a firm in relation to competitor service providers. Items related to specific service benefits determine the cognitive loyalty phase. Affective loyalty reflects whether a customer “likes” a service provider. Commitment and purchase intention items are used to determine if customers are conative loyal. Finally, action loyalty items reflect deeply held motivations to purchase from a service provider.

In the study, we first assess the measures of service loyalty across three unique service industry samples: auto repair, dry cleaners, and physicians. The data from self-reported questionnaires and structural equation modeling methodology is used to assess the reliability and validity of the loyalty model, and the results identify strong support for the internal consistency and validity of the scale. We then discuss the behavioral outcomes of loyalty, which transcend current conceptualizations and models of service loyalty. A detailed discussion of the managerial implications of the model of service loyalty follows. Next, we consider whether gender moderates the relationships between the antecedents to loyalty and service loyalty.

The moderation effects are assessed using multigroup LISREL analyses (cf. Jaccard and Wan 1996). Specifically, median splits of the full sample are calculated for the complexity moderator in order to provide a series of subsamples (high and low) for input into the multigroup analyses. Following this, baseline models are tested where equality constraints are imposed on all beta and gamma parameters across the groups. Then a series of comparison models are analyzed by allowing each path identified in the hypothesis to be freely estimated for the comparison group. The sample is used as input into a multigroup structural equation modeling analysis. In the first step of the multigroup analysis, all the structural paths are constrained to be equal across both samples providing a baseline chi-square value. Then, a second model is estimated where the path between the antecedent and loyalty is freed so that it can be uniquely estimated for the sample and thus provide a second chi-square value.

In order for moderating effects to be supported, the chi-square value for the first model must be significantly higher (.05 = Δχ 2 ≥ 9.49, .01 = Δχ 2 ≥ 13.30) than the chi-square value produced in the second model.

Analysis of the Hypothesis

The hypothesis suggests that a person’s gender moderates the relationships between the antecedents and loyalty and service loyalty. Additional structural analyses are conducted to determine if the beta coefficients for the two groups differ as hypothesized. The sample is split and a separate structural model is examined for the two groups (male and female).

The theoretical model is tested with a scale-level structural equation model. Fit indices indicate that the model fits the data well, χ 2 (205, n = 2187) = 3907.57, CFI = .96, NNFI = .95, NFI = .96, IFI = .93, and SRMR = .10. Results (OSQ → Loyalty, Δχ 2 = 10.14, p < .01; SAT → Loyalty, Δχ 2 = 39.14, p < .01; VALUE → Loyalty, Δχ 2 = 10.50, p < .01; TRUST → Loyalty, Δχ 2 = 9.75, p < .01; DJ → Loyalty, Δχ 2 = 17.62, p < .01, PJ → Loyalty, Δχ 2 = 27.58, p < .01; IJ → Loyalty, Δχ 2 = 19.72, p < .01) of the chi-square difference tests reveal that the gender moderator has a significant effect on seven of the eight paths and thereby provides support for the hypothesis.

Hypothesis: Gender moderates the strength of the relationships between overall service quality, satisfaction, service value, trust, justice, and risk on loyalty.

The hypothesis is based on the assumption that gender moderates the relationship between loyalty and its antecedents. This hypothesis is tested on the across industry sample in order to achieve a sample of sufficient size to perform the analysis. This hypothesis is strongly supported by the data. The analysis demonstrates that the path coefficients from service quality, satisfaction, value, and trust to the four types of loyalty are all stronger for the females in the sample. This finding is consistent across all four types of loyalty (cognitive, affective, conative, and action) for each of the aforementioned antecedents of loyalty. Thus, these four traditionally studied drivers of service loyalty are more salient for females. This means that service providers must be aware that women are more likely to reward service providers who offer higher service quality, satisfaction, value, and trust than men by becoming cognitive, affective, conative, and action loyal customers. On the other hand, the results of the moderation tests indicate that women are also more likely to punish firms who provide substandard service.

The results for the hypothesis are most likely due to the conventional roles of women in the purchase of goods and services. Specifically, women have traditionally been the “family purchasing agent” (Slama and Tashchian 1985); thereby they have higher involvement with purchase decisions and subsequent product usage evaluations. Thus, women feel a greater sense of achievement when they make a positive purchase decision. This sense of achievement coupled with higher levels of motivation and involvement creates the moderating effects identified by the analysis of the hypothesis. Thus, even though the composition of the American family is changing with more women now entering the workplace than ever before, it appears that traditional familial roles have remained. Therefore, managers are advised to reaffirm females’ purchase decisions and also make every effort to limit potential negative cognitive dissonance among their female clientele. This can be accomplished through post-purchase contact such as follow-up phone calls, letters, and/or emails. In recent years, the automobile industry has undertaken this form of post-purchase dissonance-reducing strategy with much success.