Methods of ability parameter estimation in educational testing are subject to the biases inherent in various estimation procedures. This is especially true in the case of tests whose properties do not meet the asymptotic assumptions of estimation procedures like Maximum Likelihood Estimation. The item weighting procedures in this study were developed as a means to improve the robustness of such ability estimates. A series of procedures to weight the contribution of items to examinees' scores are described and empirically tested using a simulation study under a variety of reasonable conditions. Item weights are determined to minimize the contribution of some items while simultaneously maximizing the contribution of others. These procedures differentially weight the contribution of items to examinees' scores, by accounting for either (1) the amount of information with respect to trait estimation, or (2) the relative precision of item parameter estimates. Results indicate that weighting by item information produced ability estimates that were moderately less biased at the tails of the ability distribution and had substantially lower standard errors than scores derived from a traditional item response theory framework. Areas for future research using this scoring method are suggested.

As the content and format of educational assessments evolve, the need for valid and workable standard setting methods grows as well. Although there are numerous standard setting methods available for multiple-choice items, there is a much smaller pool of methods from which to choose when constructed-response items or performance assessments are considered. In this study, four standard setting methods were evaluated. Two of the methods were used with the simulation component of a licensing examination, and two were used with the multiple-choice component. The two methods used with the simulations were the Work Classification method and the Analytic method. With the multiple-choice items, the Item Cluster method and Direct Consensus method were employed. The Item Cluster and Direct Consensus methods had each been the subject of research on two previous occasions, and the aims of the current study were to make modifications suggested by earlier findings and to seek replication of trends found earlier. The Work Classification and Analytic methods, while bearing some similarity to existing methods, are seen as new approaches specially configured to reflect the features of the simulations under consideration in the study. The results for each method were evaluated in terms of three sources of validity evidence—procedural, internal, and external—and the methods for each item type were contrasted to each other to assess their relative strengths and weaknesses. For the methods used with the simulations, the Analytic method has an advantage procedurally due to time factors, but panelists felt more positively about the Work Classification method. Internally, interrater reliability for the Analytic method was lower. Externally, the consistency of cut scores between methods was good in two of the three simulations; the larger difference on the third simulation may be explainable by other factors. For the methods used with the multiple-choice items, this study's findings support most of those found in earlier research. Procedurally, the Direct Consensus method is more efficient. Internally, there was less consistency across panels with the Direct Consensus method. Externally, the Direct Consensus method produced higher cut scores. Suggestions for future research for all four methods are given.

No description available.

The purpose of this study was to investigate the abilities of students to fake good on the MMPI, to identify particular MMPI clinical scales most vulnerable to faking good and to determine the efficacy of the MMPI validity scales to detect that faking. A dual administration design was used. Subjects took the MMPI once under standard instructions and once under instructions to fake good. Faking good was measured by comparing the two profiles and calculating the change in $T$-score points in going from the standard to the fake-good condition across the clinical and validity scales. A counterbalanced design was used with 76 subjects in each condition to control for re-test effects. The quantification of faking-good behavior allowed for multiple regression equations to be calculated. The regression factors were found by correlating the clinical and validity scales with faking good performance measured to $T$-score changes. $T$-score reductions in going from standard to fake-good conditions were defined as successful faking good. The results showed that students were able to fake good to a significant degree overall on the MMPI. The average $T$-score decrease over all ten clinical scales was 21.5 $T$-score points scored with the $K$-correction factor and 30.8 $T$-score points scored without $K$-correction. Five of the ten clinical scales were significantly altered by the fake good condition. Those scales were D, Hs, Pd, Pt and Sc and the addition of the $K$-correction factor tended to mask measured fake-good behavior on these scales. Measured faking-good behavior was found to correlate at $r$ =.57 with regression equation (2). Equation (2) yielded Phi coefficients of.48 and.61 at faking-good levels of 25 and 50 $T$-score points, respectively. Equation (2) was able to correctly identify 82 percent of the fake-good profiles and 89 percent of the honest profiles using a fake-good criterion of 50 or more $T$-score points. Regression equation (2) is as follows: (2) Faking score without $K = 154.7 - 2.5F + .95L - 1.1K.$ The use of the dual administration design allowed for regression equations that yielded higher validity scale screening scores than those found in past research. This finding may have significance in testing situations with fake-good demand characteristics.

This study developed and validated the Berke Prenatal Anxiety Questionnaire, created to measure prenatal anxiety. Scores from 301 pregnant women in various stages of pregnancy were used to determine the internal consistency of the Berke Prenatal Anxiety Questionnaire. These women were recruited in obstetric offices and hospitals across California. Coefficient alpha for the total scale was.87. Factor analysis showed that the items did not really cluster together in three groups as expected. Two approaches were undertaken to develop a new definition, in order to revise the scale. Three one-way ANOVA with trimester of pregnancy as the between subjects variable were calculated on the total scale and revised scales scores from the questionnaire and no significant differences were found. This questionnaire needs additional validation though the creation of new items characterizing the revised definitions. Eventually, in a revised form, this questionnaire can be used in additional research and in a clinical setting to identify pregnant women experiencing high levels of maladaptive prenatal anxiety. Efforts could then be made to reduce their anxiety, which can benefit both the mother and her infant.

No description available.

No description available.

No description available.

No description available.

Thesis (M.S.)--University of Michigan, 1978. / "A research report submitted in partial fulfillment of the requirements for the degree ..."