Using velicer's map test and horn's parallel analysis for determining component number

Abstract

In structural validity studies, determining number of dimension of a measurement tool requires critical decisions and variety techiques, methods from those decisions' basis. Rather than arbitrary decisions if researchers use the methods, techniques which have got statistical basis, the quality of study improves. For this reason instead of using some flawed exploratory factor analysis's techniques; Velicer's MAP test, Parallel analysis which are validated procedures have been used in study. For confirming structure, confirmatory factor analysis have been used. Data consist of 4645 students, gathered from PISA 2006 attitude items. Cronbach alpha have been found: 0.916, 0.912, 0.812, respectively. At the end of the confirmatory factor analysis it has seen that the error indices are low, fit indices are more than 0.90. By using O'Connor (2000)'s program, MAP test and the parallel analysis are applied on the same data, it is found that the structure has three dimensions with both methods.