A causal relationship model of ICT project managers' competencies: A multi-group analysis

This study applies competency theory to project management. The purpose of this study is to test for multi-group invariance of ICT project managers' competencies across two groups (low and high experience). The baseline model consisted of four latent variables: knowledge areas, technical and managerial skill, personal attributes, and efficiency of project management. The survey sample consisted of 335 project managers in companies within the ICT (Thailand) industry. SEM was used to analyze the causal relationship model; SPSS was employed for descriptive statistics. The baseline models were considered optimal for representing the data for low (χ
²
= 114.15; df = 124; p = 0.73; GFI = 0.93; CFI = 1.00; RMSEA = 0.00; R
²
= 0.77) and high (χ
²
= 140.50; df = 124; p = 0.15; GFI = 0.92; CFI = 0.99; RMSEA = 0.02; R
²
= 0.84) experience groups. The results for the overall model show that all indicators used to form the low and high experience models complied with goodness-of-fit criteria. The adjusted models were consistent with the empirical data. A summary of fit indices for five nested models for the multi-group analysis showed that the baseline model did not lead to a statistically significant change in the model fit. Thus, (1) the causal structural relationships model separately in each group (low and high experience) were consistent with the empirical data; (2) the baseline model complied with the criteria for goodness-of-fit and supported across the samples for both groups; (3) the chi-square test differences demonstrated the invariance of the factor loading, the regression weights among the latent variables, the variances of the latent variables, the variances of the structural latent variables, and the measurement residuals across the two groups; (4) a causal relationship model of ICT project managers' competencies fits the data equally well for the low and high experience samples; and (5) the assumption of an equivalent factor structure was supported across the low and high experience groups. Implications of the findings are discussed, and further studies are suggested.