Construct and response bias correlates in summated scale definitions of personality traits

Graphic rating scales are frequently used in the collection of self-report data. Although the specification of these response scales is common, they have been identified as being particularly susceptible to several response biases - most notably acquiescence, central tendency, and extremity. The possibility that these response styles may be more or less prominent in individuals of different trait standing has been acknowledged, but has resulted in conflicting conclusions and recommendations. The current chapter posits that response biases are particularly problematic in the assessment of personality, because they may be characterized as sources of true construct variance (i.e., extreme option endorsement or avoidance, central tendency, and yeah/nay-saying are facet dimensions of some FFM constructs). Unfortunately, the source of true construct variance is scattered across different traits. This is problematic for establishing trait orthogonality - if "other" construct variance is introduced into FFM measurement because of response bias, scale correlations would be expected not solely because of construct association, but also because of measurement method/response bias shared across trait specifications. We used a large archival dataset to estimate relationships between FFM trait (and subfacet) standing and response styles along a 5-point graphic rating scale (ranging from response options of Very Inaccurate to Very Accurate). In addition to identifying personological content associated with acquiescence and extremity, implications of this investigation: 1) point toward some of the observed FFM trait correlations potentially being attributable to response style confounds, and 2) suggest adjective checklists or forced-choice formats may be preferable to graphic rating scale specification in personality assessment.