Tonković, Maša.
(2007).
The Psychometric Validation of the Test S.
Diploma Thesis. Filozofski fakultet u Zagrebu, Department of Psychology.
[mentor Ljubotina, Damir].
Abstract
Experience of psychologists in practice releveal that, since the content of tests often becomes publicly known with time, there is a constant need for new intelligence tests. The aim of this research was to make a complete analysis of the preliminary form of the Test S, a new abstract thinking test, and its 61 items, so that the final, more economical version of the test could be made. The test was administered to 172 participants - senior high-school students and students of different professions, for the duration of one hour. Results of the test are distributed normally. The average difficulty index (p=0,675) shows that the participants didn't find the test to be too difficult. The Cronbach alpha index points to high test reliability (�±=0,908). Test sensitivity is also high, according to Ferguson's index ('=0,980). The factor analysis gave more factors, but from the sizes of their eigenvalues it can be concluded that there is only one general factor, which describes 17% of the variance. A short version of the test is suggested, containing 44 items, with estimated test administration of 30-35 minutes. Normal distribution of results has been preserved and the test difficulty increased towards the optimal one. By deleting 17 items, the reliability and the sensitivity of the test have even increased, and in the factor structure, the general factor is even more emphasized with a variance explanation of 22%. We can conclude that by reducing its size, the Test S became more effective. Replication of this study on a representative sample and test standardization are recommended. It would be highly useful to check how the short version of the test functions in practice.
Actions (login required)
 |
View Item |
Statistics
Statistics