Cranfield Online Research Data (CORD)
Tache, Irina_Poster_DSDS17_2017-11-30.pdf (645.72 kB)

Considering Culture When Using Unexpected Questions in Detecting Deception

Download (645.72 kB)
posted on 2017-11-15, 12:04 authored by Irina Tache, Lara Warmelink, Paul Taylor, Lorraine Hope
Poster presented at the 2017 Defence and Security Doctoral Symposium.

Researchers and practitioners must be careful not to interpret differences in how people respond to questions as indicative of deception without first considering cultural background. This research encourages discussion on how cultural backgrounds might best be considered to guide interviews.

Asking unexpected questions during interviews is an effective method of detecting deception. However, most of the existing evidence for this comes from studies of people from individualist cultures (e.g., Sweden, UK). Studies of cultural norms and interpersonal beliefs suggests that what counts as unexpected may differ between individualist and collectivist cultures (e.g., China, Ukraine). Our study compared the responses of individualist and collectivist truth-tellers and liars to expected and unexpected questions.

Second-language English-speaking participants (N = 120; 58 individualists; 61 liars) were interviewed, in English, about a future intention. They responded to three questions grouped into four types: expected by all cultures, unexpected by all, individualist-expected, collectivist-expected. These were derived from a pilot. Analyses were run on the language participants used, comparing liars and truth tellers as a function of culture and question expectedness.

This study is the first to suggest that question expectancy may be culturally dependent. Moreover, we find that differences in language use, known to be indicators to deception (e.g. negations, pronoun use), were explained much better by cultural background. Whether someone is lying or not explains some of the differences in people’s language but is not meaningful when accounting for people’s cultural background.