mojtaba habibi; balal izanluo; ebrahim khodai
Volume 3, Issue 9 , October 2012, , Pages 81-104
Abstract
According to Kerlinger (1P86), factor analysis is a statistical method serving “scientific parsimony principle”. This technique can be used to reduce multiplicity between variables, purify the relationships among variables, and maximize simplification. In fact, this technique determines The ...
Read More
According to Kerlinger (1P86), factor analysis is a statistical method serving “scientific parsimony principle”. This technique can be used to reduce multiplicity between variables, purify the relationships among variables, and maximize simplification. In fact, this technique determines The dependency among measures and variables and facilitates scientific interpretation. The purpose of the present study is to introduce the application of factor analysis for assigning weight to the variables, exploring the simplest and the most feasible indices with highest weight and significance of evaluation of the sample group, and determining the factor loading for each of the scale items on every one of the explored factors. In the present descriptive study, the data related to the psychological and physiological symptoms of stress among 430 high school teachers of West Azerbaijan Province (105 teachers of physical education and 298 teachers of other fields) were utilized. In the analysis of the data using method of factor analysis, weight was assigned to items and in the next stage, the weight of each item based on the weight of that indicator in the related factor would be used instead of constant weight equal one, in order to come up with final scores of subjects. Results show that there is a significant difference between the two methods in determining the weight of the items by Friedman. Referring to the results, it could be concluded that efficacy of factor analysis in comparing with other ranking methods of indicators was in optimum level.
m Habibi; fatemeh moradi; balal Izanlo
Volume 2, Issue 6 , January 2012, , Pages 1-27
Abstract
Background: Discussion about invariance of questions and tests is an important issue in assessment.
Objectives: The present study was conducted to compare the invariance of the parameters in item-response theory and confirmatory factor analysis.
Methods: After reviewing the related basics of each approach, ...
Read More
Background: Discussion about invariance of questions and tests is an important issue in assessment.
Objectives: The present study was conducted to compare the invariance of the parameters in item-response theory and confirmatory factor analysis.
Methods: After reviewing the related basics of each approach, the researcher compared the invariance of the parameters in each approach based on empirical data result from International Reading Literacy Study (PIRLS) Test. The sample was 5000 Iranian students (half female and half male) in 2006 who responded to six questions which were related to the scale of attitude toward reading.
Results: Data analysis showed that question 6 is biased using both item-response theory and confirmatory factor analysis. The results, however, were different considering questions 1, 3 and 4. Question 1 was found to be biased based on item-response theory only; questions 3 and 4, on the other hand, were found to be biased based on confirmatory factor analysis.
Conclusion: It is suggested that both approaches be employed when deciding on the invariance of the parameters, since making decisions otherwise will be misleading. Also, it is offered that intercept and differences in the distribution of the ability of the groups and their effects on the invariance be considered as primary.