Empirical results are presented as regards the implementation of a latent-trait psychometric model by means of conditional maximum likelihood estimation. Items are scored polychotomously into varying numbers of nominal categories and the test and item characteristic curves and information functions are examined. It is concluded that scoring items in four or more categories, as opposed to the usual dichotomous scoring, can increase information gain by a factor of two or more in the lower range of ability. Thus, the error of measurement is decreased to an extent equivalent to doubling the test length in this range. Alternatively, one can sample the range of ability in the target population with far rewer items. This latter property addresses itself directly to the empirical constraints on time and resources which are encountered in psychological testing. (Author)