¡§Recovery Paradox¡¨ (RP) means that customer who experienced a service failure and a superior service recovery were more satisfied than the customer who didn¡¦t experience a service failure. Furthermore, ¡§Double Deviation¡¨ (DD) means that customer who experienced a service failure and a inferior service recovery were more dissatisfied than the customer who didn¡¦t experience a service failure or just experience a service failure. These two phenomena imply that customers pay more attention on the recovery than on the failure and that the evaluation of recovery is boosted by the initial failure. This study aims to discuss the boundary conditions in which the RP and DD are more likely to occur. This study uses experimental design to examine the boundary conditions of RP and DD. Experiment scenarios were composed of a series of service event in a restaurant. And the experiment wants to examine whether the customer who experienced the service failure would magnify the evaluation of recovery. Moreover, the experiment also wants to examine whether there are other factors affect the occurrence of these two phenomena. The previous failure experience, the strength of service and the measurement of satisfaction were included in this study as potential boundary conditions. The result supports RP, but doesn¡¦t support DD. And the result shows that there is no significant effect between the satisfaction and the previous failure experience. Beside, the strength of service and the measurement of satisfaction have effect on recovery satisfaction.

La satisfaction client et la création d’expérience de service positives sont des enjeux forts pour les entreprises. Il faut désormais envisager ces concepts comme des processus évolutifs au fur et à mesure que se construit la relation client. Cette thèse s’inscrit qualitativement et quantitativement dans cette démarche. Elle propose, dans le cas d’expériences de services hédoniques, d’étudier l’évolution de la structure de la satisfaction client au fur et à mesure qu’il multiplie ses expériences. Le secteur spécifique des jeux vidéo est choisi ; un accord a été passé avec une entreprise privée. La partie qualitative, par des entretiens de joueurs, d’experts et des données secondaires, permet de définir l’expérience client autour de quatre dimensions (cognitive, affective, sensorielle, sociale). Ces dimensions sont confirmées par une première étude quantitative portant sur un jeu de quêtes et d’aventures auprès de 2198 joueurs américains. Une seconde étude quantitative est menée auprès de 2208 joueurs dont la satisfaction est mesurée à trois moments de la relation client afin d’observer l’évolution dans le temps du poids des dimensions de l’expérience. Les résultats démontrent qu’au cours du temps il y a bien une évolution du poids des dimensions sur la satisfaction avec notamment un rôle croissant de la dimension affective. Nous avons également pu conclure à l’importance des éléments cognitifs de l’expérience dont le poids sur la satisfaction devient constant dans une phase de relation longue avec le service. Ces apports permettent d’envisager la mise en place d’actions managériales permettant d’améliorer la gestion de l’expérience de service hédonique au cours du temps. / Customer satisfaction and the creation of positive service experiences form part of the main stakes of companies nowadays. It is now necessary to consider these concepts in a time dynamic perspective observing their changes during the different stage of the customer relationship. It is the approach followed quantitatively and quantitatively in this thesis. We propose in the case of hedonic service experiences, to study the evolution of the structure of customer satisfaction as experiences accumulate. We focus on the specific sector of video games – an agreement has been signed with a private company. Thanks to semi-structured interviews, experts’ interviews and secondary data, the first qualitative part aims to define the customer experience around four dimensions (cognitive, affective, sensorial, social). These dimensions are confirmed in a first quantitative study with a panel of 2198 American gamers of a quest and adventure video game. A second quantitative analysis based on 2208 American gamers has been done to observe the evolution of the weight of experience dimensions on satisfaction over time. Measuring their satisfaction at three different stages of the customer relationship, the results demonstrate that, over time, there is an evolution of experience dimensions weight on satisfaction. We underline a growing role of the affective dimension as experiences accumulate. We also conclude about the importance of the cognitive dimension even if it becomes constant as the customer goes into a long-term relationship. In a managerial perspective, this thesis offers some cues to better manage the customer hedonic service experience over time.

本研究提出一個四階段的消費者重購承諾心理評價程序模式，認為消費者將依據知覺品質和期望品質的比較形成品質失驗，然後藉由內在的價值函數將品質失驗轉換成為同樣是失驗觀點的個別交易滿意度（滿意度失驗）。在形成個別交易滿意度之後，消費者將依循累積滿意度基礎的個別交易滿意度邊際微調模型來形成當期的累積滿意度，並且依據當期的累積滿意度和當期的忠誠門檻值比較的結果，進而產生品牌忠誠或品牌轉換的行為意圖。 整體而言，失驗觀點的個別交易滿意度（滿意度失驗）和品質失驗之間的關係大致服膺期望理論價值函數的特性。此外，本研究提出的累積滿意度基礎的個別交易滿意度邊際微調模型也獲得支持，並進一步確定累積滿意度的調整過程主要是採行性線的調整模式。最後，本研究提出的累積滿意度忠誠門檻模型亦獲得實證證據的充分支持。因此，一旦當期的累積滿意度決定之後，消費者將以他心目中對競爭品牌的累積滿意度做為比較的基礎、當作忠誠門檻值，進行焦點品牌累積滿意度和競爭品牌累積滿意度的比較，並且依據比較的結果形成品牌忠誠或品牌轉換的行為意圖。 / This study proposes a four-stage model about the process of consumers’ repurchase commitment. The model postulates that consumers would result in quality disconfirmation based on the result of comparison between perceived quality and expected quality, then transform it through intrinsic value function into transaction-specific satisfaction which is defined as satisfaction disconfirmation. Once transaction-specific satisfaction is formed, consumers would result in current cumulative satisfaction according to cumulative satisfaction-based transaction-specific satisfaction marginally adjusting model that depicts current cumulative satisfaction as a function of previous cumulative satisfaction and current transaction-specific satisfaction. And finally, according to loyal threshold model of cumulative satisfaction, intention to repurchase or switch buying would be formed based on the relative magnitude between cumulative satisfaction and loyal threshold which is defined as cumulative satisfaction towards a main competing brand. Overall speaking, the relation function between quality disconfirmation and transaction-specific satisfaction exhibits most traits of value function depicted in the prospect theory. Furthermore, the cumulative satisfaction-based transaction-specific satisfaction marginally adjusting model proposed by this study was supported by experimental data, and the result shows that the adjusting process of cumulative satisfaction complies mainly with linear adjusting model. And finally, the loyal threshold model of cumulative satisfaction was fully supported by experimental data, and the result implies that consumers would take the cumulative satisfaction towards a main competing brand as the comparing basis, i.e., loyal threshold, and then compare cumulative satisfaction towards a focal brand with this comparing basis to decide whether to repurchase the products of the focal brand or instead switching to buy the goods of the competing brand.