Abstract
A profusion of recent research on consumer responses to individualized messages points to an enduring interest in the psychology of web-based customization. Across disciplines, the consensus is that highly individualized messages will generate more favorable outcomes than less individualized ones because they match message recipients' need for unique self-identity. We challenge this popular notion by arguing that highly individualized messages will not be as effective in collectivistic cultures that discourage unique self-identity. We test this proposition in a 2 × 3 experiment by randomly assigning participants (N = 120) representing either an individualistic or collectivistic culture (American, Chinese, respectively) to one of two levels of a customized message (highly individualized vs. less highly individualized) or a noncustomized, generic message (control condition). The experimental findings suggest a significant mediated moderation effect: Culture moderates the relationship between message type and attitude such that American participants report a more favorable attitude toward highly individualized messages and Chinese participants report a more favorable attitude toward less highly individualized messages, and this moderation is mediated by a psychological sense of community. Besides theoretical implications, we also highlight an original methodological procedure for manipulating customized messages in online environments.
Notes
DISCLOSURE: This manuscript was submitted to Media Psychology before the second author assumed responsibility as a co-editor of the journal. He was not involved in the editorial handling of the submission and did not have access to the correspondence in the peer-reviewing process.
*p < .05.
**p < .01.
***p < .001.
1. Two participants who were assigned to the tailored message condition both reported that movie star Vince Vaughn best represented their unique interests. Thus, they read the same news story in the experiment. As a result, the number of experimental websites in the tailored condition was 40 but the number of stories was 39 (instead of 40).
2. Many participants in the targeted message condition reported the same group membership (e.g., Chinese student community at the university) and the same common news interest shared by all group members (e.g., the university basketball team), thus, they viewed the same experimental website. As a result, the number of experimental websites was 11 (instead of 40) and the number of stories was 10 (instead of 40).
3. A pilot study was conducted with 30 Chinese participants using the same research methodology described in the current study, to ensure the efficacy of customized message manipulation, measurement reliability, and experimental procedure.
4. We followed CitationMuller et al. (2005) on this claim since the overall moderation effect was significant in the original regression model (Equation 1). Moreover, we tested the idea of moderated mediation by creating interaction variables between the moderator and the mediator. However, no significant result was detected.