Tourism marketing research

Empirical generalizations represent the second most frequently generated form of marketing knowledge in tourism; 38% of studies fall into this category. Satisfaction and loyalty (43 per cent), and motivation (16 per cent) are most frequently studied using empirical generalizations.Typical constructs investigated include consumer perceptions, attitudes, vacation experiences, perceptions of value, destination image, and information search. Findings have implications for branding, positioning, product design, and promotional messages.
Empirical generalizations are typically derived from the analysis of cross-sectional survey data, using structural equation models (49 per cent). To establish causality, structural equation models require experimental data or individual-level longitudinal data, where the cause occurs in the first and the effect in the second measurement. They cannot “discern causal relations in the absence of experimental or even quasi-experimental designs” (Kline, 2011, p. 9). Such data is rarely available in recent tourism marketing studies, and the same cross-sectional data is often used for deleting items based on unsatisfactory factor loadings, and testing and modifying the structural model. This practice has been criticized repeatedly in psychology and marketing (Baumgartner & Homburg, 1996; Breckler, 1990; Kline, 2012) and tourism (Mazanec, Ring, Stangl, & Teichman, 2010; Nunkoo, Ramkisson, & Gurosy, 2013), because it changes the intended confirmatory nature of structural equation modeling to being exploratory (Mazanec, 2007b), thus not permitting causal conclusions to be drawn. Consequently, studies using structural equation models relying on cross-sectional data do not go beyond empirical generalizations as they cannot establish causality.
A different approach is adopted by Bradley and Sparks (2012), who model antecedents and consequences of change in consumer value for the time-share market, using longitudinal data collected 12 months apart, thus adding knowledge in the area of making promises, which is, together with keeping promises, the most prominent area for empirical generalizations. Despite using a better than usual design, the authors state: “our non-experimental method prevented causal relations being inferred” (p. 203).