Date of Award

Spring 2004

Document Type


Degree Name

Doctor of Philosophy (PhD)


Business Administration-Marketing

Committee Director

John B. Ford

Committee Member

Earl D. Honeycutt, Jr.

Committee Member

Edward Markowski


Some researchers (Abrams and Paese, 1993; Bitner, Booms, and Tetreault, 1990; Etzel and Silverman. 1981; Feinberg et al., 1990; Folkes and Kotsos, 1986; Gilly and Gelb, 1982; Hart, Heskett, and Sasser, 1990; Hocutt, Chakraborty, and Mowen, 1997; Kelley and Davis, 1994; Kelley, Hoffman and Davis, 1993; McCollough and Bharadwaj, 1992; Michel, 2001; Chrage, 2001; Smith and Bolton, 1998; Spreng, Harrell, and Mackoy, 1995; Tax, Brown, and Chandrashekaran, 1998) support the notion of a ‘recovery paradox’ which states that the occurrence of a failure may, if the recovery is effective, offer an opportunity to acquire higher satisfaction ratings from customers than if the failure had never happened. While a number of researchers have provided evidence in support of the recovery paradox, several recent studies (Andreassen, 2001; Maxham, 2001; Maxham and Netemeyer, 2002; McCollough et al. 2000) have failed to find such support.

This dissertation theoretically and empirically examines factors which moderate the occurrence of a ‘recovery paradox’ in the event of a service failure. The research findings indicate that, under appropriate conditions, a customer can experience a paradoxical satisfaction increase after a service failure. One such condition entails the severity of the failure. That is, results indicate that it is unlikely that a first-rate redress initiative can return the satisfaction of a severe failure recipient to par. The findings of this investigation also reveal that a customer who has experienced a prior failure with the firm is less likely to be impressed by a superb recovery than a customer who has never encountered a problem with the service provider. In addition, customers are more forgiving of failures that occur during a process than mistakes that occur as part of the outcome. Furthermore, both control and stability intervene to affect the likelihood of increases in post-failure customer satisfaction. That is, people are more forgiving if they feel that the failure was not reasonably foreseeable to the service provider. Likewise, customers are more apt to exonerate the firm if they assess that the failure is unlikely to happen again. Lastly, this research found that control and relationship type interact to influence the probability of a recovery paradox. Specifically, customers in a true relationship are more likely to accept a low control explanation of the failure than customers in a pseudo-relationship with the firm.