Not available / text

Thesis (Ph. D.)--University of Washington, 2005. / Vita. Includes bibliographical references (p. 87-91).

The fallibility of human judgment in the making of decisions requires the use of tests to enhance decision-making processes. Although testing is surrounded with issues of bias and fairness, it remains the best means of facilitating decisions over more subjective alternatives. As a country in transition, all facets of South African society are being transformed. The changes taking place within the tertiary education system to redress the legacy of Apartheid, coincide with an international trend of transforming higher education. One important area that is being transformed relates to university entrance requirements and admissions procedures. In South Africa, these were traditionally based on matriculation performance, which has been found to be a more variable predictor of academic success for historically disadvantaged students. Alternative or revised admissions procedures have been implemented at universities throughout the country, in conjunction with academic development programmes. However, it is argued in this dissertation that a paradigm shift is necessary to conceptualise admissions and placement assessment in a developmentally oriented way. Furthermore, it is motivated that it is important to keep abreast of advances in theory, such as item response theory (IRT) and technology, such as computerized adaptive testing (CAT), in test development to enhance the effectiveness of selecting and placing learners in tertiary programmes. This study focuses on investigating the use of the Accuplacer Computerized Placement Tests (CPTs), an adaptive test battery that was developed in the USA, to facilitate unbiased and fair admissions, placement and development decisions in the transforming South African context. The battery has been implemented at a university in the Eastern Cape and its usefulness was investigated for 193 participants, divided into two groups of degree programmes, depending on whether or not admission to the degree required mathematics as a matriculation subject. Mathematics based degree programme learners (n = 125) wrote three and non-mathematics based degree programme learners (n = 68) wrote two tests of the Accuplacer test battery. Correlations were computed between the Accuplacer scores and matriculation performance, and between the Accuplacer scores, matriculation performance and academic results. All yielded significant positive relationships excepting for the one subtest of the Accuplacer with academic performance for the non-mathematics based degree group. Multiple correlations for both groups indicated that the Accuplacer scores and matriculation results contribute unique information about academic performance. Cluster analysis for both groups yielded three underlying patterns of performance in the data sets. An attempt was made to validate the cluster groups internally through a MANOVA and single-factor ANOVAs. It was found that Accuplacer subtests and matriculation results do discriminate to an extent among clusters of learners in both groups of degree programmes investigated. Clusters were described in terms of demographic information and it was determined that the factors of culture and home language and how they relate to cluster group membership need further investigation. The main suggestion flowing from these findings is that an attempt be made to confirm the results with a larger sample size and for different cultural and language groups.

Please read the abstract in the section 00front of this document / Thesis (PhD (Educational Guidance and Counseling))--University of Pretoria, 2005. / Educational Psychology / unrestricted

It is not uncommon to use unidimensional item response theory (IRT) models to estimate ability in multidimensional data. Therefore it is important to understand the implications of summarizing multiple dimensions of ability into a single parameter estimate, especially if effects are confounded when applied to computerized adaptive testing (CAT). Previous studies have investigated the effects of different IRT models and ability estimators by manipulating the relationships between item and person parameters. However, in all cases, the maximum information criterion was used as the item selection method. Because maximum information is heavily influenced by the item discrimination parameter, investigating a-stratified item selection methods is tenable. The current Monte Carlo study compared maximum information, a-stratification, and a-stratification with b blocking item selection methods, alone, as well as in combination with the Sympson-Hetter exposure control strategy. The six testing conditions were conditioned on three levels of interdimensional item difficulty correlations and four levels of interdimensional examinee ability correlations. Measures of fidelity, estimation bias, error, and item usage were used to evaluate the effectiveness of the methods. Results showed either stratified item selection strategy is warranted if the goal is to obtain precise estimates of ability when using unidimensional CAT in the presence of two-dimensional data. If the goal also includes limiting bias of the estimate, Sympson-Hetter exposure control should be included. Results also confirmed that Sympson-Hetter is effective in optimizing item pool usage. Given these results, existing unidimensional CAT implementations might consider employing a stratified item selection routine plus Sympson-Hetter exposure control, rather than recalibrate the item pool under a multidimensional model.

<p>Doktorska disertacija razmatra kori&scaron;ćenje računarskog adaptivnog testa u evaluaciji znanja programskih jezika C++ i Java. Za potrebe istraživanja realizovana je aplikacija koja omogućava adaptivno testiranje i njena funkcionalnost proverena je u praksi. Izvr&scaron;eno je anketiranje studenata radi evaluacije, dobijeni podaci su statistički obraĎeni i protumačeni. Na osnovu rezultata istraživanja utvrĎeno je da postoji statistički značajna razlika u korist studenata eksperimentalne grupe, naspram studenata koji su test radili na konvencionalan način (papir i olovka).</p> / <p>The doctoral dissertation discusses the application of the computer adaptive test in knowledge evaluation of the programming languages C++ and Java. For the purpose of this research an application was realized which enables adaptive testing. Its functionality was checked in practice. A survey was carried out among students for the purpose of evaluation, the attained data was statistically analyzed and interpreted. Based on the results of the research it was determined that there is a statistically significant difference in favor of the students in the experimental group as opposed to the students who had taken the test in the conventional way (paper-pencil tests).</p>

Testing analog circuits is more difficult than digital circuits. The reasons for this difficulty include continuous time and amplitude signals, lack of well-accepted testing techniques and time and cost required for its realization. The traditional method for testing analog circuits involves measuring all the performance parameters and comparing the measured parameters with the limits of the data-sheet specifications. Because of the large number of data-sheet specifications, the test generation and application requires long test times and expensive test equipment. This thesis proposes an implicit functional testing technique for analog circuits that can be easily implemented in BIST circuitry. The proposed technique does not require measuring data-sheet performance parameters. To simplify the testing only time domain digital input is required. For each circuit under test (CUT) a cross-covariance signature is computed from the test input and CUT's output. The proposed method requires a training sample of the CUT to be binned to the data-sheet specifications. The binned CUT sample cross-covariance signatures are mapped with a supervised machine learning classifier. For each bin, the classifiers select unique sub-sets of the cross-covariance signature. The trained classifier is then used to bin newly manufactured copies of the CUT. The proposed technique is evaluated on synthetic data generated from the Monte Carlo simulation of the nominal circuit. Results show the machine learning classifier must be chosen to match the imbalanced bin populations common in analog circuit testing. For sample sizes of 700+ and training for individual bins, classifier test escape rates ranged from 1000 DPM to 10,000 DPM.

Although not universally accepted, much of the field has converged upon the Five Factor Model (FFM) of personality as constituting a comprehensive taxonomy of normal personality functioning. A weakness common to all personality inventories is excessive length, which can result in examinee fatigue, and ultimately, poor data quality. Computer adaptive testing offers a solution to the test fatigue problem by administering only the items that are informative for that examinee on a given scale. A new test based upon the FFM of normal personality administered in a computer adaptive fashion was constructed. Reliability and validity evidence were obtained, with favorable results. New approaches to the detection of intentional response distortion were explored with mixed results including some promising findings in need of cross-validation. Response latencies were able to discriminate between honest and faking subjects, but the findings were unable to clarify the issue of whether faking is an easier or more laborious cognitive process than honest answering. New directions in computer adaptive personality testing research are proposed. / Ph. D.

The purpose of this research was to investigate the effect that mode of test administration could have on computerised assessment results involving proctored and unproctored test conditions. Two South African test instruments, the Learning Potential Computerised Adaptive Test (LPCAT) and the Career Preference Computerised Adaptive Test (CPCAT) were used in the study. A quantitative, quasi-experimental design was used, and a convenience sample for LPCAT (N=82) and CPCAT (N=81) consisted of employees in the hospitality industry. Using a within-participants design, the dependent t-test was used for statistical analysis. For the total group the LPCAT results yielded no statistically significant differences between the mean scores for the two different modes of administration. For the total group the CPCAT results yielded statistically significant differences in the mean scores per mode of administration for five out of 34 dimensions, however, for the majority of the CPCAT sub-dimensions, the mode of administration did not impact on results. It was concluded that mode of administration did not impact on the cognitive test scores and only to a very limited degree on the non-cognitive test scores. Based on the results the null hypotheses for the effect of mode of administration were not rejected. / Industrial & Organisational Psychology / M.A. (Industrial and Organisational Psychology)

The purpose of this research was to investigate the effect that mode of test administration could have on computerised assessment results involving proctored and unproctored test conditions. Two South African test instruments, the Learning Potential Computerised Adaptive Test (LPCAT) and the Career Preference Computerised Adaptive Test (CPCAT) were used in the study. A quantitative, quasi-experimental design was used, and a convenience sample for LPCAT (N=82) and CPCAT (N=81) consisted of employees in the hospitality industry. Using a within-participants design, the dependent t-test was used for statistical analysis. For the total group the LPCAT results yielded no statistically significant differences between the mean scores for the two different modes of administration. For the total group the CPCAT results yielded statistically significant differences in the mean scores per mode of administration for five out of 34 dimensions, however, for the majority of the CPCAT sub-dimensions, the mode of administration did not impact on results. It was concluded that mode of administration did not impact on the cognitive test scores and only to a very limited degree on the non-cognitive test scores. Based on the results the null hypotheses for the effect of mode of administration were not rejected. / Industrial and Organisational Psychology / M.A. (Industrial and Organisational Psychology)