Abstract:
To develop, validate, and test the invariance of a proposed multilevel structural equation model of academic dianship effectiveness. The samples were stratified randomly, and consisted of 20 deans, 379 faculty members and 280 supporting staff in three public autonomous universities in Thailand. The instruments were on five-point Likert scales, with Cronbach's alphas ranging from 0.585 to 0.962. Statistical analyses were made based on descriptive statistic Pearson's product moment correlation and MANOVA, using SPSS version 13.0 for Windows. The confirmatory factor analysis was carried out using LISREL version 8.35, whereas the multilevel confimatory factor analysis and the multilevel structural equation model analysis were performed using Mplus version 3.13. The model of academic deanship effectiveness was described by six a priori dimensions namely, (1 ) management of the unit (2) vision and goal setting (3) communication skills (4) interpersonal relationships (5) quality of the unit's education and (6) research, professional and community endeavors. The main research results showed that (1) The perceptions of faculty members towards the academic deanship effictiveness were moderate for nearly all variables, except for communication skills which was rather high. The perceptions of supporting staff were quite high for all variables, except for management of the unit which was moderate. Comparison between groups of variables showed that, the perception of both faculty members and supporting staff on research, professional and community endeavors, and communication skills were higher than others, whereas the perceptions on management of unit were lower than others (2) The proposed multivel structural equation model of academic deanship effectiveness fit quite well with the empirical data set (x[supercript]2 = 107.679, df = 80, x[superscript]2 / df = 1.346, CFI = 0.995, TLI = 0.993, RMSEA = 0.023, SRMR [subscript]B = 0.096, SRMR [subscript]w = 0.013). The statistical analysis showed further that, the individual-level variables, such as the leader-member exchange, the performance of evaluators, and associate/assistance deans significantly affected the perceptions of the academic deanship effectiveness. Whereas for the institutional-level variables, only female dean and effectiveness of the unit significantly affected the academic deanship effectiveness. the predictor variables at the individual and institutional levels accounted for variance of the academic deanship effectiveness of about 79% and 56%, respectivety (3) The test of the invariance of the multilevel structural equation model of academic deanship effictiveness showed that model was not invariance in form and structure, the proposed model might not be suitable for supporting staff.