Information loss occurs when continuous data are grouped in discrete intervals. After calculating the squared correlation coefficients between continuous data and corresponding grouped data for four population distributions, the effects of population distribution, number of intervals, and interval width on information loss and recovery were…

An example demonstrates and explains that summary statistics commonly used to measure test quality can be seriously misleading and that summary statistics for the whole test are not sufficient for judging the quality of the test. (Author/LMO)

This response to the Baglin paper (1986) points out the fallacy in inferring that inappropriate scaling procedures cause apparent discrepancies between medians and means and between means calculated using different units. (LMO)

Approximations were compared with exact statistics obtained on 85 different classroom tests constructed and administered by professors in a variety of fields; means and standard deviation of the resulting differences supported the use of approximations in practical situations. (Author)

This study was designed to examine the level of dependence within multiple true-false test-item clusters by computing sets of item correlations with data from a test composed of both multiple true-false and multiple-choice items. (Author/LMO)

Two methods of constucting equal-interval scales for educational achievement are discussed: Thurstone's absolute scaling method and Item Response Theory. Alternative criteria for choosing a scale are contrasted. It is argued that clearer criteria are needed for judging the appropriateness and usefulness of alternative scaling procedures.…

The effect of reducing the number of score groups in the matching criterion of the Mantel-Haenszel procedure when screening for differential item functioning was investigated with a simulated data set. Results suggest that more than modest reductions cannot be recommended when ability distributions of reference and focal groups differ. (SLD)

For the purpose of obtaining data to use in test development, multiple matrix sampling plans were compared to examinee sampling plans. Data were simulated for examinees, sampled from a population with a normal distribution of ability, responding to items selected from an item universe. (Author/LMO)

Linear conversions were developed relating scores on recent forms of the Graduate Record Examinations. Conversions based on specially selected subpopulations were compared with total-group conversions and evaluated. Conclusions indicated that the data clearly support the assumption of population independence for homogenoeous tests, but not quite…

This study was designed to determine whether adjustments for the differential prediction observed among sex, racial/ethnic, or age subgroups in one freshman class at a college could be used to improve prediction accuracy for these subgroups in future freshman classes. (Author/LMO)

The precision of the group-based item-response theory (IRT) model applied to school ability estimation is described, assuming use of Bayesian estimation with precision represented by the standard deviation of the posterior distribution. Similarities with and differences between the school-based model and the individual-level IRT are explored. (SLD)

Rasch model equatings of multilevel achievement test data before and after the deletion of misfitting persons were compared. Rasch equatings were also compared with an equating obtained using the equipercentile method. No basis could be found in the results for choosing between the two Rasch equatings. (Author/LMO)

Using simulated and real data, this study compares the performance of three methodologies for assessing unidimensionality: (1) DIMTEST; (2) the approach of Holland and Rosenbaum; and (3) nonlinear factor analysis. All three models correctly confirm unidimensionality, but they differ in their ability to detect the lack of unidimensionality.…

Four methods are outlined for estimating or approximating from a single test administration the standard error of measurement of number-right test score at specified ability levels or cutting scores. The methods are illustrated and compared on one set of real test data. (Author)

Equal percentile (EP) and partial credit (PC) equatings for raw scores from performance-based assessments with free-response items are compared through the use of data from the Maryland School Performance Assessment Program. Results suggest that EP and PC methods do not give equivalent results when distributions are markedly skewed. (SLD)