Guests’ perceptions of hotels’ (un)fair treatment of staff: their impact on service recovery

Limited previous research has examined the impact that third-party guests’ observations of (in)justice for others can have on a hotel. This study deals with this issue by testing whether the way guests who raised a complaint perceive management treatment towards hotel staff influence their satisfaction with service recovery (SSR), when controlling for perceptions of justice for the self: distributive, procedural, and interactional. The paper first verifies the distinctiveness of all the justice variables included in this study. Data were collected from 288 guests who had experienced a service failure in seven sampled hotels in the Canary Islands (Spain). Structural equation modeling (SEM) results indicated that guests who raised a complaint and were unhappy on the management treatment towards hotel staff, showed significantly lower satisfaction with service recovery (SSR). These findings suggest that, like justice for the self, justice toward staff also predicts satisfaction with service recovery (SSR), thus earning a place in the service recovery literature. Suggestions for future research and practical implications are also presented.


Introduction
Service failure is a prevalent and costly problem for hotels. Each year, for example, Hampton Inn spends approximately 0.5% of its total room revenue on trying to make customers satisfied through service recovery (Stoller, 2005), that is, the actions and process through which a hotel attempts to rectify service failure (Kelley and Davis, 1994). Additionally, if satisfaction with service recovery (SSR) is not achieved, it has been suggested to be related to a wide range of guest responses that can substantially impair the hotel's effectiveness (see, for a review, Gelbrich and Roschk, 2011;Matos et al., 2007;Kim et al, 2012) such us decreased customer loyalty or revisit intention (e.g., de Ruyter and Wetzels, 2000;Karatepe, 2006), customer complaint behavior (Kim et al., 2010), negative word-of-mouth (WOM; Mattila, 2001) communication about the hotel, and perceived service quality (Moreno-Gil et al., 2006), just to name a few.
However, the above-mentioned theory and research linking justice theory and SSR have over-emphasized the effects of justice toward the self on SSR (i.e., when guests themselves are targets of justice), neglecting the potential impact that guests' observations of incidents of (in)justice toward staff (i.e., those perpetrated by the hotel toward their own staff) can exert on SSR. This significant imbalance in favor of the victim's perspective raises an important gap in the justice literature in hospitality (e.g., Karatepe, 2011). Based on prior research suggesting that third-party customers can also make fairness judgments and react to the way employees are treated (e.g., Cowley, 2005;Skarlicki et al., 1999), this study attempts to fill this gap by examining whether perceptions of justice toward staff still display significantly increased effects on SSR with unique variance, when controlling for justice dimensions for the self. The argument is made that the way guests perceive the staff's treatment by hotel management may be associated with the service recovery process and attributed to it, ultimately affecting the level of SSR. The authors are unaware of any prior empirical study of perceptions of justice toward staff as the origin of SSR in the hospitality industry context.
To summarize, first this paper plans to justify and formulate the study's hypothesis based on the theory and research described above. Then, structural equation modeling (SEM) will be used to test the distinctiveness of all the justice constructs, among themselves and with SSR, as well as main effects of justice toward staff on SSR when three justice dimensions for the self (distributive, procedural, and interactional justices) are controlled. Finally, the paper will discuss the results, along with their implications for theory and for effectively promoting SSR in hotel organizations.

Theoretical Background and Hypothesis
Perceived justice in service encounters, rooted in the equity theory by Adams (1963), is a multi-faceted concept comprising three dimensions: distributive, procedural, and interactional justice. Smith et al. (1999, pp. 358-359) defined distributive justice as "the allocation of costs and benefits in achieving equitable exchange relationships." Procedural justice refers to "the perceived fairness of policies, procedures, and criteria used by decision makers to arrive at the outcome of a dispute or negotiation" (Blodgett et al., 1997, p. 189), whereas Tax et al. (1998, p. 62) conceptualized interactional justice as "the perceived fairness of interpersonal treatment that people receive during the enactment of procedures." As mentioned above, numerous studies have supported the positive effects of perceived justice on SSR (Blodgett et al., 1997;Chang et al., 2012;Chebat and Slusarczyk, 2005;del Río-Lanza et al., 2009;Mattila and Patterson, 2004;Ha and Jang, 2009;Karatepe, 2006;Kim et al., 2009;Kuo and Wu, 2012). Thus, perceived justice has been adopted as a powerful vehicle in predicting SSR.
Despite the practical importance of this connection and the abundant literature supporting it, there is a lack of models examining whether perceived justice from the dissatisfying guest on management treatment towards hotel staff predicts SSR. Previous third-party literature has suggested that an observer who witnesses (in)justice and is not an organizational member is likely to respond to perceived justice for others in a manner similar to that of an actor-victim in the situation, although perhaps less intensely (e.g., Lind et al., 1998;Sheppard et al., 1992;Tyler and Smith, 1998;Walster et al., 1978). Due to this functional similarity between perceived justice for others and for the self, perceived justice toward staff could also perform as a source of satisfaction for guests when immersed in a service recovery process in a hotel (SSR). Thus, as in perceived justice for the self, perceived justice toward staff in a hotel may be "… a source of both satisfaction and positive evaluations of the [hotel] organization" (Lind and Tyler, 1988, p. 191) and, therefore, of the process and actions performed by the hotel organization regarding service recovery. However, how can justice toward guests build this SSR? Oliver (1999) states that during the service encounter, a customer passes through cognitive and emotional phases. First, in the cognitive phase, customers develop attitudes toward the hotel based on prior knowledge or experience-based information. If perceived justice toward staff occurs, it could provide guests with positive knowledge or information about attributes of the hotel service, leading them to the rationale that the hotel is efficient in resolving service failure. As a result, on the one hand, cognitive self-interest could cause justice toward staff to lead guests to become satisfied with the service recovery (SSR).
In an emotional phase, on the other hand, the guests could also develop dis/like or other affective feelings toward the hotel as cumulative events of (un)fair treatment of staff occur (Oliver, 1999). Although perceived justice toward staff does not elicit a self-interest concern, Folger (2001) proposed that people could also be emotionally concerned with just treatment toward others simply because they believe it is the "right way to act." Thus, if guests perceive fair treatment for staff, since they are morally compelled to admit that the hotel is behaving correctly toward the staff, positive emotions could cause guests to feel satisfied with service recovery even though they suffer personal harm (Turillo et al., 2002). In fact, perceived justice has been found to make individuals more willing to subordinate their own short-term individual interests to the interests of others or the organization (Lind and Tyler, 1988). To the extent that guests perceive the hotel's fair treatment of staff to be part of the service recovery actions, guests will probably make an effort to contribute to bettering service recovery and be willing to have a positive and tolerant view of their hotel's performance in this regard. Therefore, perceived justice toward staff should be able to predict SSR in hotel guests by developing emotional liking toward the hotel.
Another reason to expect that perceived justice toward staff will predict SSR involves the phenomenon of legitimacy. Tyler's (2006) previous research shows how individuals are motivated by their evaluations of the legitimacy of organizational policies and processes (see also, Tyler and Blader, 2005). Scholars in the field agree that the way the organization allocates duties can be perceived as legitimate when it is inspired by widely agreed-upon principles or values (e.g., Biel and Thoegersen, 2007). Since justice is one of the most important principles to legitimize organizational conduct, when justice perceptions inspire the way service recovery process is conducted by the hotel, it is likely to contribute to guests' SSR. Since the service recovery process includes not only hotel managers and staff, but also guests as co-producers (Vargo and Lusch, 2008), guests' perceptions of the staff's treatment by hotel management may be directly assigned to the service recovery itself, favoring its legitimacy and ultimately shaping guests' feelings of SSR.
Perceived justice for the self and for others may overlap. Although the concept of justice toward staff clearly differs from justice toward the self (in its three dimensions, i.e., distributive, procedural, and interactional justices), as the focus lies outside (employees) the figure of the guest, perceived justice for the self captures strong justice perceptions that are able to outshine the effects of justice toward staff on SSR. This situation implies that some part of the effects of justice toward staff on SSR may already be included in the impact of justice dimensions for the self on SSR. Therefore, in order to test the unique variance of justice toward staff on SSR, the general hypothesis in this study is formulated as follows: Hypothesis: When positive perceptions of the three justice dimensions for the self (distributive, procedural, and interactional justices) among guests are controlled for, their justice perceptions toward staff will be positively associated with higher levels of SSR.

Customers/ Respondents
Data were collected from guests at four sampled four-star hotels and three sampled five-star hotels in Gran Canaria (The Canary Islands, Spain). The researchers chose upscale hotels for the sample because successful service recovery after failure is a high priority for luxury hotel managers in maintaining long-term guest relationships. The four-to five-star differentiation in The Canary Islands stems from Hotelstars' criteria (set by HOTREC -Hospitality Europe; www.hotelstars.eu), which are primarily based on the quality and quantity of facilities, communications, customer area, general service and staff area (e.g., the minimum size for the rooms and bathrooms is 15% lower in 4star hotels than in 5-star hotels). Guests were chosen who met the criteria of being 18 years of age or older and staying 4 nights or more (in that particular hotel and that visit), so that they had a certain socialization period at the hotel and were more likely to experience service failure. Based on each 4-and 5-star hotel's capacity, an average occupancy of 80% and 76% for 4-and 5-star hotels, respectively, and an average guest stay of 6.8 nights, a population size of 1,093,540 guests was estimated, and a sample of 384 guests was established for a 5% sampling error. The sample comprises guests who stay in halfboard (includes breakfast and dinner), full board (includes breakfast, lunch, and dinner) and all-inclusive packages (all the food and beverage services and other needs are met by the hotel). Details about characteristics of the guests in the sample are shown in Table 1.

Research settings
The Canary Islands is a leading sun-and-beach tourism destination in Spain. The importance of tourism to Spain is unquestionable; with 56.7 million international tourists and tourism revenue of $59.9 billion a year, Spain is fourth in the destination rankings, behind France, the United States, and China, but only behind the United States in international tourism revenue (World Tourism Organization, WTO 2012). Gran Canaria receives about 3.23 million foreign tourists a year, with European countries being its principal markets: British and German tourists jointly represent 41% of the total, Scandinavians 28%, and other nationalities the remaining 31%. There are 56,841 hotel beds and 72758 non-hotel beds in the tourism municipalities in Gran Canaria (Canarian Government, Department of Tourism, 2012). Four-and five-star hotels represent 50% of the hotels in Gran Canaria. Specifically in the sample, the number of hotel rooms ranges from 88 to 676, with an average number of rooms per hotel of 343. According to type of property, of the seven sampled hotels, international chains own four.

Questionnaire Design and Procedure
In constructing the justice toward staff assessment scale for this study, the Hinkin (1998) guidelines for adequately developing these measures for use in survey questionnaires were followed. Although justice toward staff is generally missing in the existing hospitality research, item generation was first deductively undertaken based on the strong theoretical and empirical base available in prior literature on justice for the self (for metaanalytic reviews see, for example, Cohen-Charash and Spector, 2001;Colquitt et al., 2001). Number of items per scale (4 to 6 per scale) and item scaling used matching requirements for "content adequacy" (Schriesheim et al., 1993), whereas recommendations for item-to-response (range from 1:4 (Rummel, 1970) to at least 1:10 (Schwab, 1980)) were also fulfilled. Following Hinkin's (1998) suggestions, the authors checked that inter-item correlations of the variables and factor loadings were 0.4 or greater and the variance explained by each of the items was greater than 60%. The resulting measures, together with the justice for the self and SSR scales, comprise the final questionnaire used in this paper, once the language in the German and Spanish questionnaires had been pretested.

Statistical Analysis
Data analyses for this study include descriptive analyses, confirmatory factor analysis (CFA), and structural equation modeling (SEM). The collected data were analyzed using the statistical package for social sciences (SPSS). Structural equation modeling (SEM) was used to assess the validity of the measures, and SEM was employed to test the hypothesized relationships through AMOS 19.0. In order to test the construct validity of the justice toward staff, justice for the self and SSR measures, following Hinkin's (1998) guidelines, an exploratory factor analysis (EFA) was performed for item reduction with orthogonal rotation, followed by the planned confirmatory factor analysis (CFA). The indices used include comparative-fit (CFI), normed-fit (NFI), Tucker-Lewis (TLI), incremental-fit (IFI), and root mean square error of approximation (RMSEA). Calculations of the convergent/discriminant validity of the constructs were also performed.
All items were scored on a 7-point scale ranging from (1) 'Strongly Disagree' to (7) 'Strongly Agree,' and they are presented in Table 2. Reliability was assessed by using the Cronbach's alpha. Alpha values appear on the main diagonal of the correlations table, as well as in Table 2.
Distributive Justice for the Self. Distributive justice was assessed with a four-item measure developed by Karatepe (2006), who adapted it from Smith et al. (1999).
Procedural Justice for the Self. A five-item procedural justice scale was adapted from Blodgett et al. (1997) and Karatepe (2006).
Interactional Justice for the Self. Guests' interactional (mis)treatment was assessed with a seven-item measure developed by Karatepe (2006), after a review of the literature. Specifically, he combined five items from Severt (2002) and two items from Smith et al. (1999).

Justice toward Staff.
A thirteen-item scale was constructed by the authors, adapting scales from the literature on organizational justice (e.g., Moorman, 1991). It included aspects of distributive, procedural and interactional justice. However, it is unlikely that guests acting as mere observers are in a position to make a clear distinction between these different aspects of the hotel's (mis)treatment of employees. Therefore, a one-dimensional scale is expected.

Satisfaction with Service Recovery (SSR).
A five-item scale was used, adapted from Brown et al. (1996) and Maxham III and Netemeyer (2002). Table 3 shows the scale means, standard deviations, reliabilities and correlations (r) among all the variables. Results show that justice toward staff has a significant positive inter-correlation with SSR in the expected direction. These results appear to be a good starting point to support the hypothesis in this study. The exploratory factor analysis (EFA) performed for item reduction, which includes justice for the self, justice toward staff, and SSR, suggests that one procedural justice for the self item, X06 'They were resolved as quickly as they should have been', should be rejected and consequently dropped. The factor loadings are very similar to those in the interactional justice factor, less than .4 (e.g., Hatcher, 1994;Tabachnick and Fidell, 2001). However, the remaining items loaded as predicted in the expected factors.

Results
Next, a CFA was calculated to test the fit of the hypothesized 5-factor model and ensure that the variables were five separate constructs. In the testing model for CFA, the factor loading for item X08 of procedural justice for the self 'The procedure the hotel uses for handling my complaints is complicated' was less than 0.5 and, therefore, deleted (see Hatcher, 1994;Tabachnick and Fidell, 2001). Once item X08 had been dropped, the AVE and Cronbach's alpha were recalculated, changing from 0.498 to 0.579 and from 0.795 to 0.801, respectively. In addition, the modification indices were analyzed, showing that a significantly better fit would be achieved if a number of the error terms were allowed to be correlated (see Figure 2). /df< 3, GFI>.9, IFI>.9, TLI>.9, NFI>.9, CFI>.9, RMSEA<.05) and, hence, providing support for the distinctiveness of the four variables under study. Table 2 shows that the composite reliability ranged from 0.802 to 0.968, which is greater than the standard of 0.6 (Hair et al., 2006). A set of established procedures was also employed to check for the convergent and discriminant validity of the scales. The average variance extracted (AVE) for each construct was between 0.579 and 0.798, supporting convergent validity (Bagozzi and Yi, 1988;Fornell and Larcker, 1981). Discriminant validity was measured by calculating the AVE for all the pairs of constructs and comparing this value to the squared correlation between the two constructs of interest. The results show that, in all cases, the squared correlation between any given pair of constructs was less than the respective AVE of each of the constructs in the pair (Fornell and Larcker, 1981), thus supporting discriminant validity.
Next, SEM was used to test the relationships among the variables in the study. Figure 2 is a path diagram that shows the relationships between the observed variables (survey answers, in rectangles) and the unobserved latent variables (circles). The items provided in Table 2 define the variables of the observed model. The various fit indices used, shown in Figure 2, reveal an acceptable fit of the model. Support for the Hypothesis is provided by the significant path between justice toward staff and SSR (β=0.177; p<0.01). In addition, unlike distributive justice for the self (β=0.012; n.s.), procedural justice (β=0.042; p<0.001) and interactional justice (β=0.019; p<0.05) showed significant links with SSR.

Discussion
The aim of this research was to test the impact of justice toward staff on SSR, when controlling for the three dimensions of justice for the self (distributive, procedural, and interactional) by entering them together with justice toward staff in the same model. The supported association between justice toward staff and SSR offers several useful theoretical and practical implications for the hospitality management literature and the field of consumer behavior.

Figure 2. Accepted Model of Satisfaction with Service Recovery (SSR), Justice for the Self (Distributive, Procedural and Interactional) and Justice toward Staff
injustice for others. The present study does not claim that the guests in our sample do not feel fear or vulnerability about subsequent mistreatment after observing incidents of injustice for staff (see, for a review, Skarlicki and Kulik, 2005). Instead, it is likely that they do feel intimidated, but these feelings do not keep them from taking (in)justice for staff into account. Thus, the idea that guests react to justice for staff by rejecting self-interested calculations like "it is someone else's problem, not mine" gains strength, suggesting that thirdparty guests who raised a complaint are concerned with injustice for staff largely due to a moral imperative. The performance of perceived justice for the self, on the other hand, was consistent with the findings by Ok et al. (2005) and del Río-Lanza et al. (2009), who concluded that procedural justice has a greater influence on hotel guests' SSR than distributive and interactional justices. In contrast, the study results challenge the Smith et al. (1999) and Kim et al. (2009) findings that distributive justice has a stronger impact on SSR than the other two types of justice. Nor do they agree with Blodgett et al. (1997) and Karatepe (2006), who suggested that interactional justice is the most important determinant of SSR, followed by distributive and procedural justices.
The results of this paper open up a new avenue in the study of the impact of perceptions of (in)justice on SSR, specifically in the context of the hospitality industry. Unless the actions and process through which a hotel attempts to rectify a service failure (Kelley and Davis, 1994) adequately elicit favorable guest perceptions of (in)justice for both the self and others, these service recovery efforts may be unsuccessful. Neither type of (in)justice seems innocuous, and the presence of both appears to greatly enhance the effects of the service provider's recovery efforts on guests' feelings of SSR. Nevertheless, this study highlights that a justice strategy for achieving SSR should strengthen healthy guest perceptions about the way the hotel organization treats its own staff. Moreover, since perceptions of distributive justice for the self failed to predict SSR, and interactional justice did so to a lesser extent than justice toward staff, the above-mentioned justice strategy seems critical because the results point to justice toward staff as a SSR driver that has similar potential to that of justice for the self. Without a doubt, justice toward staff seems to warrant a significant place within the service recovery literature in the hospitality industry context.

Managerial implications
As supported by our findings, events related with justice toward staff can be a threat for SSR. However, from the same findings can be derived that hotel management (mainly responsible for guests' satisfaction) could take actions to achieve guests' perceptions of staff fair treatment by the hotel as an opportunity to maintain, and even enhance, the levels of SSR. In other words, negative episodes of justice toward staff are not events against which (as a kind of 'given variable') hotel managers are defenseless. Managers can and must counteract events of staff mistreatment by designing actions that communicate to guests how the organization values and supports its staff, and that they are not staying in a hotel that treats its employees unjustly. In light of this study result, in order to prevent incidents of injustice, it would undoubtedly be necessary to arrange the workplace based on serious, solid, and believable principles and norms that truly cultivate proper hotel treatment of staff. This situational practice leading to a kind of 'just hotel toward itself' might involve reaching employer-worker agreements about the general arrangement of the workplace and its processes. Although the general rule is that staff should always be treated well, guests who have experienced a service failure are at "high risk" and should receive the utmost care in order to ultimately achieve SSR. Managers must discuss this fact with employees and supervisors. They should include the idea that episodes of mistreatment toward subordinates by bosses (or by co-workers toward their peers) in the presence of guests are not innocuous, and they can significantly erode the quality of the service recovery provided. Hotel managers, in addition, can make erroneous decisions in this regard. For example, trying to handle service failures of guests with a reduced staff might seem to be an efficient decision at first glance, but it may ultimately cause guests to witness how the staff barely copes with their tasks. These events can become an embarrassment for guests and be counterproductive, particularly if they involve the same staff that usually handles the service failure. Finally, the mistreatment observed by guests during their stay at the hotel may act in parallel to other mistreatments they experience 'in their own flesh.' The results of this study already provide warnings in this regard, by supporting how justices for the self and toward staff independently and additively affect SSR with unique variances.

Future research
Some questions remain unanswered that could be the basis for future research. First, there is a need to extend the span of what has been defined here as (in)justice "for others," which can also include (in)justice for "other guests" in addition to staff. Second, there is a need for research on the different impacts that guests' perceptions of (in)justice can have on SSR depending on the different situations in which guests find themselves when observing (un)fair treatment. Nationality could play a intervening role in this regard, since not all country cultures may be equally inclined to "protest" against rough treatment of service employees. These situations can induce significant differences in the performance of the constructs used in this study. For example, episodes of (in)justice toward staff in the hotel reception during the guests' check-in may be especially influential, since first impressions during the service exchange are widely accepted as more potent. Second, whether or not episodes of (in)justice toward staff are directly involved in the service recovery process could make them more or less influential. Finally, the significant results of this study regarding hotel guests' reactions to (in)justice from the third party's perspective should elicit new research focusing on employees as third-party observers of (un)fair treatment. Hotel employees' reactions to (in)justice for guests or/and for co-workers, for instance, could lead to very suggestive research.

Limitations and Conclusions
The present study has strengths, but it also has several weaknesses. First, it might suffer from mono-method/source bias. Second, the surveyed guests are all Europeans; hence, they are influenced by certain cultural conditions that can offer a narrow view of justice and the hotel exchange. Additionally, the questionnaires were distributed in 4-and 5star hotels, raising concerns about whether the results would be the same in 2-and 3-star hotels. Consequently, the performance of the constructs used in the present study, as well as their implications, could vary in other cultures, hotel categories and countries. Lastly, the data set stems from a limited universe, also raising concerns about the generalizability of the findings.
This paper, on the other hand, contributes to a better understanding of how different (in)justice perceptions among guests can be relevant to the effective functioning of service recovery in a hotel. The almost exclusive focus on victims of perceived (in)justice, and inattention to thirdparty reactions to (in)justice in studies involving the hospitality industry, have yielded patterns that do not sufficiently explain the role that guests' perceptions of (in)justice toward staff can play in their feelings of satisfaction with hotels, particularly those concerning service recovery after service failure. By uncovering main effects of guests' perceptions of (in)justice toward staff on SSR, when controlling their perceptions of justice toward themselves, this study contributes to the success of justice strategies in satisfactorily recovering service failure.
In sum, hotel guests' perceptions of fair treatment toward staff (in addition to themselves) can become a useful tool to rectify service failure satisfactorily. Only hotel managers who build a 'just hotel', from both the victims' and third parties' perspectives, will be able to deploy the full potential of equity theory in satisfactorily handling service failure.