The Stickiness of Category Labels: Audience Perception and Evaluation of Producer Repositioning in Creative Markets
成果类型:
Article
署名作者:
Kovacs, Balazs; Hsu, Greta; Sharkey, Amanda
署名单位:
Yale University; University of California System; University of California Davis; Arizona State University; Arizona State University-Tempe
刊物名称:
MANAGEMENT SCIENCE
ISSN/ISSBN:
0025-1909
DOI:
10.1287/mnsc.2021.02070
发表日期:
2024
页码:
6315-6335
关键词:
repositioning
categorization
audiences
book publishing industry
reviews
Deep learning
Natural Language Processing
摘要:
Market producers often seek to position themselves in different categories over time. Successful repositioning is difficult, however, as audiences often devalue offerings that depart from a producer's past creations. Prior research suggests that this penalty arises as evaluators withhold opportunities for producers to reposition because of presumptions of a lack of competence in different categories. In this paper, we develop understanding of a novel evaluator-driven challenge to producers' repositioning efforts: evaluators are prone to categorical stickiness, by which the categories they have come to associate with a producer through its prior offerings shape their perceptions of the producer's subsequent offerings. The result is a systematic mismatch between what producers claim and what evaluators perceive when a producer repositions. We further propose that audience members who have the greatest prior experience with a producer are the least likely to recognize its repositioning efforts. We examine evidence for our theory using data from Goodreads.com on authors within the book publishing industry, 2007-2017. We first build a novel deep-learning framework to predict categorization of a given book based solely on an author's description of its content. We then use data on how Goodreads users categorize and evaluate books as well as their past reading behavior to test for evidence of our proposed mechanism. Overall, our results extend understanding of the evaluative processes that generate categorical constraints and how these may differ among various types of audience members.