1 Paper 352-2011 Scale Measurement: Comparing Factor Analysis and Variable Clustering Albert Liau, Teck Kiang Tan and Angeline Khoo Psychological Studies, and Centre for Research in Pedagogy and Practice Nanyang Technological University, Singapore ABSTRACT In the social sciences, factor analytical methods are commonly used in the scale measurement in examining the structure of scales. This paper provides an alternative way of examining the structure of scale measurement using variable cluster analysis. Three analytical methods are illustrated in this paper, namely exploratory factor analysis, confirmatory factor analysis and variable clustering analysis. As variable cluster analysis is rarely used and not widely discussed in the literature, the advantages, benefits and limitations of using this method are discussed. Comparison of variable clustering to the factor analytical approach is illustrated using the personal strength scale. PROC VARCLUS, PROC FACTOR, and PROC CALIS are the three SAS procedures for these three analytical methods. The syntaxes of the three procedures are given and the results are interpreted and compared. INTRODUCTION In measurement literature, exploratory and confirmatory factor analyses are two common standard statistical tools for developing measurement scales. SAS® PROC FACTOR and PROC CALIS are the two main procedures cater for these two statistical approaches. The SAS® VARCLUS procedure, borrows from the factor analysis literature, combined with hierarchical clustering, gives an alternative approach to the measurement procedure. The advantages of the PROC VARCLUS approach over the traditional factor analysis approach are discussed in details in the paper. Scale Measurement Social science research generally creates scales as constructs by using several items in a questionnaire to capture the various aspects of a construct in order to detain its essence. For example, an educational test might contain 50 items that intends to measure student’s academic performance of a subject domain. Psychological measurement almost always relies on using items to create psychological scales. For instance, an inventory of depression that makes up of 40 items to create the depression inventory. This paper uses the personal strength scale proposed by Liau (2007) and Liau et al (2010) as an example to illustrate the scale measurement of the personal strength of students. This scale is consistent with the positive psychology movement that focusing on positive qualities of students, an instrument intends to measure the well-being and development of students and it is of wide practical application (Liau et al 2010). The current personal strength inventory is a shorter version of Liau et al (2010) which makes up of 18 items subdivided into four subscales. The four subscales are emotion awareness (5 items), social competence (4 items), empathy (4 items), and goal setting (5 items). Data Source and Sample The data are obtained from a three-year longitudinal project examining the effects of video gaming on children and adolescents’ development.