Exploring the application of a text-to-personality technique in job interviews

Abstract

ABSTRACT This research’s purpose was to develop a valid and transparent text-to-personality technique to fit the requirements for personnel selection assessments. In this research we developed an advanced word-counting technique, the HEXACO text-to-personality (HTTP) technique, based on prior lexical personality research to assess personality from job interviews. To evaluate the technique’s construct and criterion-related validity we conducted three studies and analysed the transcripts of asynchronous (n = 102 and 72) and face-to-face (n = 155) interviews. These studies provided four key insights. First, the HTTP technique showed small to medium correlations with self-reported and interviewer-rated personality. Second, the technique showed mixed, but generally favourable, evidence for criterion-related validity. Third, the technique produced a more construct valid personality score when the interview questions activated the predicted personality trait. Fourth, the technique’s additional features (i.e., having weighted keywords and adjusting the keywords’ weight for adjacent quantifiers) did not improve its validity; unit-weighing was approximately equally effective. Altogether, the results show that a word-count text-analysis technique can discover traces of personality in interview transcripts. Still, significant improvements are needed before these types of automatically computed text-to-personality ratings can be used to replace or supplement interviewer ratings.