Using the short subscales of a questionnaire to assess subcomponents with low reliabilities: a cautionary note

Some authors of personality questionnaires use factor analysis with the aim of constructing subcomponents to assess subsidiary traits within the broad domain tapped by the scale. However, these short subscales suffer from low reliabilities in most cases. The findings of using the Maudsley Obsessional-Compulsive Inventory in American (n=73) and Kuwaiti (n=87) college students justified the caution against using such short subscales with low reliabilities, inasmuch as 7 out of the 8 alpha reliabilities were less than .69, ranged from .35 to .68, and the median was .56.