Developing an integrated predictive- explanatory model of behavioural intentions for mobile cellular services

Abstract

Creating a critical mass of loyal customers and their retention has been recognised as the cornerstone for the future survival of an organisation. However, customer retention strategies depend on the accurate prediction of behavioural intentions (BIs), which is an immediate predictor of actual behaviour. To improve the prediction of BIs, organisations need to understand all the possible factors that can affect a buyer‟s relationship with a brand. Perceived justice (PJSR), service recovery satisfaction (RSat), service quality (SQ), overall satisfaction (OCS), and switching barriers (SBs), have been considered as determinants of BIs in simple bivariate explanatory models. However, we live in a complex multivariate world such that, studying the determinants of BIs in isolation would seem artificial and inconsequential. Furthermore, the previous BIs models have been largely explanatory, which provide little practical relevance without assessing their predictive capacity. Consequently, there has been a call for the development of comprehensive models of BIs that incorporate more variables in a single model, which is both explanatory and predictive. Despite this call, an integrated explanatory-predictive framework that explains how BI antecedents‟ nomological causal relationships collectively lead to the formation of BIs in a consumer's mind is yet to emerge. To address this gap in the research, the researcher synthesised literature from different research streams to construct an integrated explanatory-predictive framework of BIs in situations where service failure and service recovery are involved. Cross-sectional survey data from 405 mobile phone subscribers, collected using a self-administered questionnaire in different districts of Lesotho was analysed using SmartPLS 3.2.9. The model had a high explanatory power (R2 = 0.75), high insample predictive relevance (Q2 value for BI = 0.547) and high out-of-sample predictive capacity [(low positive values of the linear regression model (LM) - the root mean square error (RMSE)]. The mediation tests show that OCS is the central construct through which all the other variables influence the formation of BI. The moderation test of SBs was only significant on the SQ->BI relationship, specifically revealing that when SBs are low, the influence of SQ on BIs is strong, but high SBs tend to obscure the effects of SQ on BI. The importance-performance matrix analysis (IPMA) reveal that OCS was the most important construct, followed by perceived justice (PJSR), RSat and SQ, in descending order. Overall, the results reveal that the performance of the cellular industry on these constructs is above average (50%), but more effort is required to improve their performance. The IPMA results of the indicator items reveal that the reliability dimension of SQ was considered the most important item in determining BI, but in general, the performance of all the indicator items was slightly above the average (50 per cent) mark. Theoretically, the study expands the current knowledge and understanding of the formation of behavioural intentions. Conceptually, the study is unique in that to the researcher‟s knowledge, it is the first of its kind to construct a framework that combines an analysis of the explanatory power, predictive relevance, the importance and performance of the determinants of BI in a single framework. In that regard, the study is a respond to the call for the development of consumer behaviour models that have more practical relevance but being grounded in strong theoretical explanations. Besides its relevance in predicting the future intentions of subscribers in the mobile phone industry, the model also offers specific, actionable recommendations for that guide management when developing customer retention strategies. A detailed explanation of the contribution of this study is discussed in Chapter 8 of this thesis.