Exploring the Grocery Store Satisfaction of England's Older Population: An Evaluation of Antecedents and Consequences Using Structural Equation Modelling

Abstract

The number of people aged 60 years and above is increasing in the UK. In total, this age group represents 22% of the population with estimates indicating a rise to 29% by 2050 (United Nations, 2009). One market sector that is extremely important to the health and wellbeing of older people is grocery retail (Khan, 1981). However, little previous research has addressed how older people rate the service delivered by their grocery provider, particularly in regard to satisfaction. To reconcile this gap in theoretical understanding, a sequential transformative mixed-method research design was specified using 36 qualitative interviews and quantitative questionnaires with 524 subjects. A model including both drivers and consequences of satisfaction was formulated using past research. As such, an antecedent scale for grocery store image was developed via procedures suggested in the extant literature (e.g. DeVellis, 2003). Pre-existing scales (i.e. commitment and loyalty) representing exemplary reliability and validity were borrowed and specified as consequences. The scales were modified and integrated into a ‘structural equation model’. Older people were found to place a high level of importance in aspects of merchandise, store environment, personnel and services. Price/promotions and clientele were found to be insignificant in driving satisfaction. Differences in factor mean scores and structural parameters were then analysed using ‘finite mixture structural equation modelling’ to identify segments of similar respondents (Jedidi et al, 1997). Using posterior probabilities, the emerging segments were subjected to profiling using personal and behavioural variables (Hahn et al, 2002). Market Segmentation showed three groups of similar respondents in the sample population, differing in factor mean scores and psychological operationalisation of satisfaction. Nonetheless, only several differences in personal and behavioural characteristics were found between the segments. Whilst, the results show that segmenting this group is necessary when measuring satisfaction, basing this purely on a priori descriptive variables might be erroneous given the inherent levels of unobserved heterogeneity. The model developed and tested in this study is considered the most up-to-date available in the literature.