Investigating Mechanisms of Social Desirability Bias in Self-Administered Surveys
Mohamed, Abdirahman Saeed
xmlui.mirage2.itemSummaryView.MetaDataShow full item record
This dissertation addresses social desirability bias, also known as socially desirable responding (SDR), a tendency of some respondents to provide responses they consider to be socially favorable. Though SDR declines in self-administered modes, the mechanisms behind SDR in self-administered modes are not fully understood. To improve the understanding of SDR in a self-administered setting, this dissertation aims to empirically investigate the effects of survey attributes and has three-fold objectives: First, to examine whether introduction interest, survey topic, survey sponsor, and perceived topic sensitivity prompt a systematic tendency to choose socially desirable response categories; second, to examine whether the order in which the response categories are presented precipitates the tendency to choose socially desirable response categories; and finally, to investigate whether the increment of item-related sensitivity prompts the tendency to SDR. To achieve the first two objectives, primary data were collected through a 2×2×2 online survey experiment from Hacettepe University students, with a sample size of 386. To achieve the third objective, a combination of primary and secondary data was used. The secondary dataset used in the dissertation is from the 2012 Argentina Global School-Based Students Health survey with a sample size of 28,368. Respondents who perceived the survey topic as sensitive marginally tended more to choose socially desirable response categories but the survey topic and sponsor are insignificant. Intriguingly, respondents who rated the survey introductory statement interesting demonstrated more tendency to SDR. The response order effect was significant and socially desirable response categories were more frequently chosen when they appeared first on the list. Also, responding in the socially desirable direction intensified as the sensitivity of survey items increased, and respondents with more missing responses tended to respond in the socially desirable direction.