Development and validation of the measurement model of group-level positive psychological capital: Composing group-level variables from individual-level data using data aggregation models

Abstract

     Data aggregation models or composition models were used to guide the aggregation of individual-level data to represent group-level variables. All studies of group-level, positive, psychological capital, or simply group-level PsyCap, composed this group-level construct from individual-level data. The purpose of this study was to develop and validate the measurement models of group-level PsyCap using different data aggregation models, namely, the direct consensus model or self-referent model and the referent-shift consensus model or group-referent model. The sample consisted of 303 student groups, selected by multi-stage random sampling, from 10 higher education institutions in Thailand in the 2013 academic year. The instrument was the PsyCap scale, consisting of self-referent items and group-referent items. The study was conducted at the group-level of analysis. Second-order confirmatory factor analysis was undertaken to validate the measurement model at the group-level PsyCap. The results showed that the measurement models of group-level PsyCap using the direct consensus model (χ² = 172.593, df = 146, p = .066, AGFI = .914, RMSEA = .025) and the referent-shift consensus model (χ² = 155.609, df = 154, p = .449, AGFI = .926, RMSEA = .006) were both valid and fitted the empirical data.