5.3. Limitations and future research
While this study provided useful theoretical and managerial insights, it is not free from limitations. First, while the sample was drawn to be representative, all online panels are non-probability samples, suggesting some caution in making generalizations. Second, although the model was assessed in two developed and two developing markets, it would be advisable to replicate the study in other service settings. It seems that, since cognitive trust is based on shared values, its effect is more pronounced in collectivist cultures, where relational norms are more prominent, which is consistent with Chen et al.'s (2002) suggestion. The service setting (e.g., hotel) used in the study might also have encouraged such a mindset. Past research also supports distinct pathways (e.g., cognitive vs affective) to persuasion for marketing communications in developing countries (Zarantonello, Jedidi, & Schmitt, 2013). Further, it is possible that some of the responses were motivated by people's personal culture orientation rather than national culture. For example, Sharma (2010) argued individuals from a collectivist society can demonstrate personal individualism. Some of our findings, like similar patterns of results for the USA and India support this. It is possible that, despite being a collectivist country, some Indian respondents were driven by personal individualism. Hence, future research might examine the impact customers' personal values, age, gender and personal cultural orientations have on their perceptions of service fairness, trust, and customer engagement behavior relationships. Finally, a limitation of the current study was the use of crosssectional design. A longitudinal design could be used to assess CEBs at different touch points to see if this provides additional insights (Bijmolt et al., 2010).
