Estimation of Reliability of Zimbardo Time Perspective Inventory using Bayesian Approach for the Promotion of Quality Education and Life Long Learning in Engineering Discipline

Abstract

Commonly reported frequentist measures of reliability like Cronbach’s alpha and McDonald’s omega, along with the less reported Greatest lower bound coefficient suffer from either underestimation or overestimation of reliability coefficients. Moreover, there is inherent uncertainty associated with the obtained estimates. An alternative approach to resolve this issue can be estimation of reliability of psychological instruments using the Bayesian approach, involving the reporting of posterior distribution of the obtained estimate. In this context, the Zimbardo Time Perspective Inventory Short Form by [17], is chosen for estimation of the reliability of its dimensions, partly due to the lack of consensus in reporting of stable estimates of this estimand in multiple contexts. The data was collected from 187 computer science engineering students of Lovely Professional University, Phagwara, Punjab, during their regular classroom session. R Package ‘Bayesrel’ was used to find the Bayesian and Frequentist frameworks-based reliability coefficients of the five dimensions of the scale using RStudio Ver. 2022.12.0+353. Except for the dimensions Present Hedonistic and Future time perspective, the reliability of rest the other three dimensions of time perspective were found to fall short of the acceptable benchmark of 0.6 [31]. The graphical posterior predictive check figures showed good fit between eigen values based posterior unidimensional model and data implied covariance matrices. The psychometric implications of the study in the context of Indian engineering students are discussed.