Measuring the mediating role of quality education for ensuring employability skills: An analysis of higher education student perception in Bangladesh

determine the minimal sample size, one should employ power analysis and consider the model constructs with the greatest number of significant predictors. According to [44], the rule of thumb established by [45] helps assess the statistical power of multiple regression models and determine the appropriate sample size. This rule considers the path model’s complexity, a min- imum R2 value, and 80% statistical power. A minimum sample size of 103 was necessary in the PLS path model employed in this investigation to identify a minimum R2 value of 0.10 at a sig- nificance level of 5% and a suggested statistical power of 80%. Measures From the literature, the measurement items for all constructs were adapted. The purpose of this study is to identify the relationship among Course Structure (CS), Institutional Policy (Pol), Physical Aspects (PA), Quality Education (QE), and Employability Skills of Students (ES), as well as the mediating role of Quality Education (QE), the unit of analysis was at the individual level. This study analysed all path linkages in the research model for hypothesis test- ing. The approach utilised in this research is confirmatory factor analysis (CFA), as the con- structs incorporated in this model have been borrowed from previous investigations. PLS has the potential for two general applications, according to [46]: theory development or theory confirmation. PLS generates hypotheses in the latter scenario by investigating the interrela- tionships among variables. Furthermore, PLS methods are frequently preferable when the phenomenon being investi-

gated is relatively novel or evolving or when the theoretical model or measures are not well- defined [47]. Additionally [48], argue that PLS is more appropriate for developing theories rather than for verifying them. Thus, employing the CFA method in the present investigation is reasonable. Results Data analysis The statistical method used to analyze the measurement and structural model was partial least squares (PLS) modeling with SmartPLS 4.0.9.6 [49]. This choice was made because PLS model- ing does not necessitate the assumption of normality, which is often absent in survey research [50]. SPSS regression was deemed inadequate due to the complex model involving mediation, which couldn’t be tested simultaneously. Structural equation modeling was therefore implemented. Prior research [43, 51, 52] outlined the following benefits of utilizing PLS-SEM: • PLS-SEM can operate with a limited sample size. • PLS-SEM evaluation of mediation is superior to regression analysis. According to [48, 51], PLS-SEM is particularly well-suited for complex models where pre- diction is of greater importance than parameter estimation, and conditions such as sample size, independence, or normal distribution are not satisfied. Thus, it was appropriate to employ PLS-SEM in this investigation. The multilevel model was used for the analysis of data with a hierarchical or cluster structure and is popular in educational research [54]. This multilevel model is required because “Students” are generally considered a Level-2 analysis unit in educa- tional research [55].

In adherence to the recommendations [53, 54], the concern of Common Method Bias was mitigated by conducting a comprehensive collinearity test since the data were obtained from a single source. By employing this approach, each variable is regressed against a standard PLOS ONE Measuring the mediating role of quality education for ensuring employability skills PLOS ONE | https://doi.org/10.1371/journal.pone.0310815 October 7, 2024 8 / 18