Neither agree nor disagree: use and misuse of the neutral response category in Likert-type scales

The Likert-type scales are among the most widely implemented instruments in social sciences, nonetheless, it is not clear so far whether such scales should or should not employ a mid-point "neutral" response option. While a mid-point category might improve the psychometric properties of survey instruments when appropriately applied, it has been argued that respondents often tend to use it in several invalid ways. This study aims to examine how a neutral response modality may influence the scales' psychometric properties. We conducted two types of survey experiments employing a between-subjects and a within-subjects design, comparing psychometric properties of twelve personality scales in both cases i.e., with and without the neutral response category. Our findings show that the scales presenting the neutral category allow to some extent for better psychometric characteristics, both in terms of their reliability and with respect to the proportion of accounted variance by the first factors. Results also suggest that most respondents seem to use the neutral category validly. However, there are also indications that a minority of respondents actually employ the neutral answer as an "escape" option, especially when asked socially sensitive questions.