The influence of online survey design on data quality in social science and consumer behavior studies

• Ano de defesa: 2025
• Autor(a) principal: Yoda, Fernanda Sayuri
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Chatbot; Online surveys; Anthropomorphism; Antropomorfismo; Data quality; Design do questionário; Qualidade de dados; Questionnaire design
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/12/12139/tde-31102025-163025/

New questionnaire designs are being developed to improve the quality of online survey responses in consumer research. One emerging data collection strategy, increasingly adopted by companies and researchers, involves the use of automated chat-based questionnaires (or chatbots) to enhance response quality in both academic and managerial studies. These tools incorporate conversational design elements (e.g., informal communication and emoji use), which function as social cues that may lead to anthropomorphization and potentially improve response quality. However, previous studies have yielded conflicting findings, depending on the quality indicators used and the specific design features implemented in chat-based surveys. It is therefore necessary to investigate the effects of chat interfaces and social cues on various response quality indicators. Additionally, individual respondent characteristicssuch as thinking style and self-efficacy in completing online surveysmay moderate these effects. Given the growing use of chat-format questionnaires in consumer behavior research, it is essential to assess their impact on response quality. This study aimed to evaluate the factors influencing responses in online consumer behavior research. A literature review was first conducted to define and measure response quality. Subsequently, an online survey experiment was carried out in which participants were randomly assigned to one of four questionnaire versions, identical in content but differing in format (chat vs. traditional form) and in the presence or absence of social cues. Both ad hoc and post hoc response quality indicators were analyzed across conditions, along with the influence of individual variables such as thinking style and self-efficacy. Results showed that anthropomorphic elements significantly increased perceived social presence in traditional interfaces but had less effect in chat interfaces. Overall, chat-based interfaces proved effective without compromising data quality, although participants with lower cognitive style or self-efficacy achieved better outcomes with traditional formats.