Two contradictory perspectives exist concerning extreme responses in survey data. In the literature of attitude strength, extreme responses indicate extreme and strong attitudes. From the perspective of the extreme response style (ERS), however, extreme responses do not indicate strong attitudes. Furthermore, some research has found that people scoring higher on the Extraversion personality scale are significantly more likely to have the ERS. Survey interviewers, however, may rely on the characteristics of extraversion, such as happiness, curiosity friendliness, and sociability, to complete an interview. Thus the relationship between extraversion and the ERS brings into question the quality of survey data. This study examines how the manifestations of extraversion in social survey interviews are related to the ERS and attitude strength, using the “Survey Research on Attitudes toward Death Penalty and Related Values in Taiwan” of 2014. Attitude strength is indexed by stability of attitudes towards the death penalty. Measurement of the ERS is based on responses to six questions that have similar distributions but have low associations with one another. Based on the conceptualization of extraversion in the Big Five personality scale, respondents’ participation eagerness and friendliness during the interview are argued to be two manifestations of extraversion. In the survey research literature, however, participation eagerness is also positively related to interest in the survey topic. Therefore, more eager participants may also be more likely to have strong attitudes. This study tests five research hypotheses, as follows. 1. Both manifestations of extraversion are positively associated with the ERS. 2. Respondents with higher levels of participation eagerness are more likely to have stable attitudes, but not those with higher levels of friendliness. 3. Extreme responses to the initial attitude question are positively related to attitude stability. 4. Initial extreme responses of respondents higher in neither of the two manifestations do not predict attitude stability. 5. Initial extreme responses involve more ERS, and less attitude strength. The data are analyzed in HLM models. AIC is used to examine Hypothesis 5. Results support the hypotheses.