Do specific personalized recommendations cause more harm than good to social identity? A moderated mediation model

Basit Öğe Kaydı
dc.contributor.authorSelcan, Burcu
dc.date.accessioned2025-10-13T18:36:01Z
dc.date.available2025-10-13T18:36:01Z
dc.date.issued17/06/2025
dc.departmentEnstitüler, Lisansüstü Programlar Enstitüsü, İşletme Ana Bilim Dalı
dc.description.abstractThis dissertation investigates how personalized product recommendations—whether from human or artificial intelligence recommenders (AI-R)—can unintentionally trigger adverse psychological reactions such as social identity threats, particularly when the product or its presentation is associated with dissociative reference groups. The scope of this research explores the influence of the signaling content of a reference group (associative vs. dissociative), the recommender type (AI vs. humans), and identifying moderators such as collective self-esteem (CSE), stereotype-related fear of negative evaluation (SR-FNE), and the possibility of price discounts overriding identity threat concerns and perceived price (PP) that shape consumers' reactions. A pretest and two studies with 510 participants from a diverse Turkish sample showed that even accurate product recommendations can lead to negative feelings when the product is linked to a group from which someone wants to dissociate. While the recommender type (H1) did not yield a significant main effect, subgroup analyses revealed differential responses based on age and gender. A significant effect was found for reference group imagery (H2): dissociative visuals led to lower product liking and higher SIT, supporting H3’s mediation model. PROCESS Model 4 confirmed SIT as a significant mediator in the relationship between visual cues and consumer response. Further, PROCESS Models 1 and 14 demonstrated that CSE marginally moderated the effect of reference group type on SIT_DC, while SR-FNE did not significantly moderate the indirect path. Hypothesis H7 tested a three-way interaction (PROCESS Model 18); CSE and SR-FNE did not consistently moderate these effects, although partial effects were observed in specific recommendation contexts. Further, three-way interactions involving SIT_DC, SR-FNE, and PP, H8 (tested via PROCESS Model 37), showed that limited but suggestive evidence that identity threats have stronger negative impacts when consumers are highly sensitive to social evaluation and perceive higher prices. Overall, the results highlight the importance of social-psychological cues in shaping consumer reactions to recommendation systems. This research contributes to the marketing and human–AI interaction literature by demonstrating that even accurate recommendations may backfire if they evoke identity threats or fears of stereotyping, especially when consumers are highly evaluative or perceive the product as high-value.
dc.identifier.urihttps://hdl.handle.net/11411/9731
dc.institutionauthorSelcan, Burcu
dc.language.isoen
dc.publisherİstanbul Bilgi Üniversitesi, Lisansüstü Eğitim Enstitüsü
dc.relation.publicationcategoryTez
dc.rightsinfo:eu-repo/semantics/openAccess
dc.titleDo specific personalized recommendations cause more harm than good to social identity? A moderated mediation model
dc.typeDoctoral Thesis

Dosyalar

Orijinal paket
Listeleniyor 1 - 1 / 1
Küçük Resim
İsim:
Do specific personalized recommendations cause more harm than good to social identity A moderated mediation model.pdf
Boyut:
4.47 MB
Biçim:
Adobe Portable Document Format
İndir
Lisans paketi
Listeleniyor 1 - 1 / 1
Küçük Resim Yok
İsim:
license.txt
Boyut:
1.17 KB
Biçim:
Item-specific license agreed upon to submission
Açıklama:
İndir

Koleksiyon

Graduate Programs Institute Thesis Collection