Indicators of confirmation bias in the investigative interview with suspects

Abstract

The aim of the research conducted for this thesis was to investigate indicators of confirmation bias within the police-suspect investigative interview. Confirmation bias in the investigative interview generally occurs when the interviewer holds a presumption of guilt about the suspect, and then sets out to find evidence of that guilt. The interview then becomes less about information gathering and more about finding support for the suspect’s involvement in the crime. In the present thesis, indicators of confirmation bias were explored through a series of experimental and applied studies. The nonverbal effects of assumed guilt expectation on the behaviour of truth-telling interviewees (N = 52) were examined in Study 1 (Chapter 2). The influence of guilt presumption on the interviewer’s behaviour was investigated in Study 2 (N = 107; Chapter 3) and in Study 3 (n = 33; Chapter 4). Evidence in the extant literature suggests that pre-existing beliefs are an underlying mechanism of bias. The novel contribution of this research to the discourse on bias and guilt presumption can be found in Studies 4 and 5. In Study 4 (Chapter 5), indicators of biased language were identified in a sampling of police-suspect interviews (N = 9). In Study 5 (Chapter 6), a sample of police-suspect interviews in a murder case (N= 6) were analysed by question type and the utterances used by interviewers. Linguistic tools (e.g., content and pragmatic analysis) were used to demonstrate the richness of information that could be found when the questions and utterances are evaluated within the context of the interview. The amount of information obtained, and suspect cooperation was measured for any effects of language on those variables. The research findings for the overall thesis suggests that indicators of confirmation bias are detectable through guilt presumptive language. Accusations and insinuations of guilt seem to be the most consistent indicator that an interviewer believes the suspect is guilty and may be attempting to find support for those beliefs. Incidences of bias are subtle; however, the influence on the interview and the suspect’s behaviour is clear (denial, defensiveness, and non-cooperation). Thus, the observable nature of biased behaviour may allow for the development of interventions prior to, and during, the investigative interview to reduce biased behaviour in interviewers.

Date of Award	May 2018
Original language	English
Awarding Institution	University of Portsmouth
Supervisor	Aldert Vrij (Supervisor), Peter J. Van Koppen (Supervisor), Sam Mann (Supervisor), Zarah Vernham (Supervisor) & Robert Horselenberg (Supervisor)