Računalna primjena testa S

Marinović, Mladen. (2009). Računalna primjena testa S. Diploma Thesis. Filozofski fakultet u Zagrebu, Department of Psychology. [mentor Ljubotina, Damir].

PDF (Croatian) - Registered users only Download (372kB) | Request a copy

Abstract

The paper and pencil version of Test S has been adapted for computer administration. The computer form has been administered to 131 participants, students of the Psychology Department of the Faculty of Humanities and Social Sciences and students of the Faculty of Electrical Engineering and Computing, for the duration of 45min. The test results have had a normal distribution with an average difficulty index of p=0.75 which implies that the test wasn’t very difficult for the participants. The Cronbach alpha index indicates that the test reliability is relatively high (=0.84). The test sensitivity, expressed with the Ferguson’s index, is also high (δ=0.968). Prerequisites for the factor analysis haven’t been fulfilled, but the analysis results point to one general factor which explains 11% of the total variance. The described data indicate general compliance with the paper and pencil test data. A correlation of r=−0.645 between the difficulty index and the average time to the first answer implies that the more difficult the items are, the more time is requested for their solving. Most participants change up to 6 answers in their tasks, although there are great individual differences among them. The analysis showed no influence of the correct answer position on response speed. Different categories of test successfulness (9 categories) have been obtained by categorizing participants according to the speed (3 categories) and correctness of response (3 categories). Different forms of curves relating to the time of the first answer and the test result have been established; however, not all of them can be theoretically explained. The data obtained from the computer administration can be used during test construction and in specific selection procedures where, besides correctness, working speed is important, too. Methodological issues which diminish generalization of the conclusions have been discussed. Additional research is needed on the usefulness of the time related variables obtained from the computer administration and the results of this research have to be verified on a representative sample.

Actions (login required)

View Item