Mastering unobserved heterogeneity in tourist behaviour research.

Abstract

This paper argues that tourist behaviour research should be based on explicit hypotheses or, even better, on a system of interconnected relationships (a 'model'). The models are subject to empirical testing to assess their explanatory and predictive capabilities. As in any other sub-field of behavioural science, a sample of tourists used for model testing is often heterogeneous in terms of variable values and association between variables. Therefore, the results gained for bivariate or multivariate relations may be artefacts due to spurious correlations. Heterogeneity is found to appear on different levels of tourist behaviour model building. It is easier to control if there are explicit assumptions about moderator variables. However, it becomes particularly troublesome if the hypothesized cause-effect relationships are subject to variation over an unknown structure of sub-groups. Examples are discussed to portray these cases.

Guidelines for treating unobserved heterogeneity in tourism research: A comment on Marques and Reis (2015)

“…Marques and Reis (2015) make a valuable contribution to the tourism research discipline by stressing the importance of considering unobserved heterogeneity in PLS-SEM. Using latent class techniques, tourism researchers can assess whether or not their results are distorted by unobserved heterogeneity (e.g., Mazanec, 2000). FIMIX-PLS takes an important role in this -10 -regard as it offers a means to treat heterogeneity nonparametrically in an SEM context.…”

“…1. INTRODUCTION Accounting for heterogeneity in tourism studies remains important to avoid parameter misspecifications in empirical models (e.g., Mazanec, 2000;Mazanec, Ring, Stangl, & Teichmann, 2010). Approaches applied in tourism research that allow the examination of observed heterogeneity (e.g., Dolničar, 2004) include, for example, multi-group comparisons based on a priori information when employing regressions analysis (e.g., Ye, Zhang, & Yuen, 2013), partial least squares structural equation modeling (PLS-SEM) (e.g., Song, van der Veen, Li, & Chen, 2012), or covariance structural equation modeling (CB-SEM) (e.g., Jurowski & Gursoy, 2004).…”

“…Approaches applied in tourism research that allow the examination of observed heterogeneity (e.g., Dolničar, 2004) include, for example, multi-group comparisons based on a priori information when employing regressions analysis (e.g., Ye, Zhang, & Yuen, 2013), partial least squares structural equation modeling (PLS-SEM) (e.g., Song, van der Veen, Li, & Chen, 2012), or covariance structural equation modeling (CB-SEM) (e.g., Jurowski & Gursoy, 2004). Differently, those approaches employed in tourism research to assess the potential of unobserved heterogeneity when certain grouping variables are not known a priori, for example, seek to identify latent classes (e.g., Mazanec, 2000;Mazanec, 2001). For instance, Assaf, Oh, and Tsionas (2015) employ Bayesian finite mixture modeling within CB-SEM, and Marques and Reis (2015) finite mixture modeling within PLS-SEM.…”

Why Do Tourists Persist in Visiting the Same Destination?

“…Hypothesis 6 (heterogeneity): the frequency of revisiting Portugal differs by destination region and by nationalities. Barros et al (2008) have outlined that tourism choice always entails heterogeneity which should be considered when deciding what statistical models to apply (see also Mazanec, 2000). Later, Correia and Kozak (2012) demonstrated that domestic tourists have different perceptions of the regions in which they stay while in the Algarve, in the south of Portugal.…”