B3 Article in conference proceedings
A Quantile Regression Approach in Predicting Patronage in the Shopping Centre Context

List of Authors: Meri Malmari, Heli Marjanen, Janne Engblom, Anna-Maija Kohijkok
Publication year: 2016
Book title *: 23rd Recent advances in retailing & services science conference: Conference proceedings
ISBN: 978-90-386-4048-8


Consumer store choice behaviour has always been in the interests of scholars and retailers. To explain and predict store choice behaviour, both demographic and behavioural variables have been used. There are two aspects to store choice behaviour; the store choice decision and the frequency of visits. The current study combines these two into a single dependent called relative shopping frequency. The study focuses on determinants that have been found powerful in explaining store choice and patronage in numerous earlier studies. The aim is to investigate the predictive power of these determinants in modelling the relative shopping frequencies for an out-of-town shopping centre. The revealed shopping frequencies in 2006 were used to predict the relative shopping frequencies in 2011. To do this, quantile regression analysis was chosen due to highly skewed distribution of relative shopping frequencies. To model the estimates for quartiles (Q1, Md, Q3) of  relative shopping frequencies, two separate models, the demographic and the behavioural were constructed using household survey data collected in 2006 (n=2082). Data collected in 2011 (n=1566) was used to evaluate the predicting power of the models. The model estimates were adjusted for the temporal change, i.e. changes in the retail provision in the area, by using the ratio method. Next, interaction terms were added to the models to investigate the possible changes in the performance of the selected variables in explaining the relative shopping frequencies. The particular strength in quantile regression is that it offers the means to formally test the differences in each independents performance in explaining the relative market share at two points of time. The results suggest changes in the predicting power of the determinants. Significant differences in how variables explain the store choice were found for independents distance separating respondents and the shopping centre, household size, parking, and dining. More specifically, the relative importance of distance in explaining the share of visits has increased, whereas the importance of parking has decreased.    

Last updated on 2019-29-01 at 14:38