On the Consistent Construction of Forced-Choice Ipsative Tests

Abstract The need for consistent construction of forced-choice ipsative tests to avoid the introduction of hard-coded ‘errors-by-design’ is highlighted. A set of necessary criteria for consistent construction, apparently widely overlooked in the literature, is found and recommended. It is shown that for ipsative tests involving P scales with more than M = 2 items per question block, the consistent combinations of P and M are in fact few and far between for tests with a manageable number of item blocks. The introduction of errors through badly designed scoring methods is also illustrated and explained via analysis.