Validation of Career Choice Goals Scale on Indian Senior Secondary School Students using Network Psychometrics Approach

Abstract

The 15 items Career goals scale developed by Graco (2016) was extended in the Indian context on 353 secondary school students and validated using network approach. While the default “glasso” approach of exploratory graph analysis (EGA) using “EGAnet” package of R 4.2.3 version, extracted two clusters, the “TMFG” approach extracted three factors, with both the networks having the same entropy estimate of -7.204. Following the principle of parsimony (Vandekerckhove et al., 2015), the two clusters network structure was retained, and also since it was in line with the original work of Seibert et al., (2013). Unique variance analysis did not report any node redundancy. The structural consistency estimates obtained of the “extrinsic” and “intrinsic” clusters, using “bootnet” package, were acceptable at 0.622 and 0.948 respectively. The package “mgm” was used to compute node predictability and “qgraph” package plotted the same. The ordinal confirmatory factor analysis using “WLSMV” estimator found acceptable goodness of fit estimates (cfi.robust=0.949; tli.robust=0.939; srmr_bentler=0.038; rmsea.robust=0.061). The “LASSO and EBIC” techniques based regularized network structure and centrality indices plots were obtained. The edge weight accuracy confidence interval plot was found to be non-significant indicating trustworthy ordering of the edge weights in the network, obtained through the comparison of sample and bootstrapped data for 500 iterations. The correlation stability CS coefficient was above the minimum benchmark strength-wise at 0.283, obtained using “qgraph” package. The package “psychTools” provided the edge difference and the node difference plots. The educational and psychometric implications of
UtilitasMathematica