Bimodal score distributions and the Myers-Briggs Type Indicator: fact or artifact?

We examined Myers-Briggs Type Indicator (MBTI) score distributions computed using item response theory (IRT) to assess the generalizability of earlier bimodality reports that have been cited in support of the "type" versus "trait" view of personality. Using the BILOG IRT program to score a sample of approximately 12,000 individuals who participated in leadership development programs, theta score distributions for the 4 dimensions of the MBTI computed using 10 (the BILOG default) versus 50 quadrature points were compared. Results indicated that past reports of bimodality were artifacts caused by BILOG's default use of a small number of quadrature points; when larger numbers of points were used, score distributions became strongly center-weighted. Although our findings are not supportive of the "type"-based hypothesis, the extremely high correlations between theta scores (rs > .996) suggest that no practical differences would be expected as a function of the number-of-quadrature-points decision.