Prelims

Bad Science Practices in Tourism Science

The use of symmetric testing (e.g., F-tests, correlations, MRA and SEM) and NHST (p < 0.05) is pervasive currently and dominates tourism science even though their critics are unanimous that constructing and testing symmetric hypotheses is bad science practice. Hubbard (2005) and Ziliak and McCloskey (2008) provide extensive reviews of this literature. Symmetric testing in tourism focuses on rejecting the null hypothesis that the relationship between X and Y is not significant statistically different from zero. The majority of symmetric-based theory and testing studies seek to show that X relates to Y positively or that the relationship is negative: cases with high X have high Y scores and cases with low X have low Y scores – or the reverse. The relationship proposals are symmetric, that is, one simple statement of directionality holds for the variance values for X and Y. Using MRA and SEM these studies usually report the ‘net effects’ typically of two-to-twenty or so X independent variables with a Y, dependent, variable. Such variable-relationship studies miss-match case-outcome theory and variable analysis (Fiss, 2007). Scientists construct theories of processes and events resulting in specific outcomes. Such theories are case-based, not variable-based. Usually, the miss-match occurs implicitly, the researcher is aware of the requirements of symmetric testing, and thus, constructs directional hypotheses (e.g., increases in X associates with increase in Y) without recognizing the theory-analysis mismatch. Constructing such directional hypotheses represents a low, shallow bar of accomplishment. ‘Low’ because earlier all relationships rejecting the null are supportable by statistical significance testing with sufficiently large samples. ‘Shallow’ for two reasons: in research, cases exhibiting contrarian relationships to the hypothesis occur even when the effect size of the predicted relationship is very large (e.g., r ² ≥ 0.6) and usually, no independent variable X associates with Y sufficiently to accurately predict that cases high in X are high in Y. The reliance on symmetric testing and NHST is so pervasive while being bad science practices, the editor of at least on behavior science journal (Basic and Applied Social Psychology) now bans the reporting of NHST findings in articles accepted for publication (Trafimow & Marks, 2015).

What recipes (configurations) of antecedent X conditions indicate the occurrence of Y (e.g., top or bottom quintile scores for Y) with consistent accuracy? Given that data include cases with low X scores having high Y scores and cases of high X scores having low Y scores, even then the main effect indicates a highly statistically significant, positive, XY relationship, what are complex configurations of antecedent conditions for these two types of contrarian cases? Good science practice recognizes that the causes of what happens relate asymmetrically with Y, not symmetrically. An antecedent, X, relates to an outcome, Y, in four fundamental ways. Data include cases whereby cases high in X have high Y values; cases low in X have high values; cases high in X have low Y values; and cases low X in case have low Y values. The presence of specific combinations of X conditions differ in content and direction among the recipes resulting in high Y versus low Y. Consequently, separate theories and testing for high Y versus low Y outcomes are necessary. Multiple configurations of complex antecedent conditions indicate the same high Y outcome – and others indicate the same low Y outcome (Woodside, Hsu, & Woodside, 2011; Wu et al., 2014). Researchers adopting good science practices not only recognize these complexity tenets, but construct their theories on foundations of these tenets (Woodside, 2014).

Good Science Practices in Tourism Science

Reading Hubbard (2015) is a necessary but insufficient step for understanding the new and ongoing paradigm shift from using NHST to SPOT and using consistency indexes indicating high accuracy with odds greater than 4-to-1 that a specific outcome occurs for cases fulfilling all requirements in an asymmetric model – a ‘somewhat precise outcome test’ (SPOT). Such SPOT testing is used widely in practice (e.g., use of stockbrokerage screens in selecting stocks and consumers creating used car screens at carguru.com and other websites to create a small consideration set of alternatives). Ragin (2008) is a primer in how to use ‘fuzzy-set qualitative comparative analysis’. Reading in two websites, COMPASSS.com and fsQCA.com, are useful steps to deepen understanding of, and for learning the latest developments in, asymmetric modelling (i.e., theory and analytics for using SPOT). Because of the appearance of several simple conditions in a configuration, a substantial share of management science literature in general and the marketing literature in particular will shift from bad-to-good likely in the third decade of the 21st century. Hopefully, this conclusion is an accurate asymmetric prediction that becomes reality.

The following studies are transformation examples of the shifts from asymmetric to symmetric theory and analysis in tourism and hospitality. Rather than reporting the net effects of culture values on attempting to study cultures as complex wholes (Hofstede, 2001), Woodside, Hsu and Marshall (2011) report how different cultures as complex wholes influence specific behaviors by tourists. Given that cultures are complex wholes, the point here is that in reporting the net effects of each cultural value (e.g., collectivism, masculinity, uncertainty avoidance and power distance), Hofstede (2001) never actually examines cultures empirically. Woodside, Prentice and Larsen (2015) use complexity theory and report alternative asymmetric models of customer assessments of their experiences with hospitality services in casinos. Using the same data set described by Prentice and Woodside (2013), Woodside, Prentice and Larsen (2015) match an asymmetric theory with asymmetric analytic tools to overcome the mismatch of offering an asymmetric theory with symmetric testing as done previously (Prentice & Woodside, 2013).

Describing the asymmetric realities among the antecedents to climate change, tourism sustainability and destination competitiveness is one of the major contributions of the chapters in this volume. The tenets of complexity theory (Urry, 2012) provide a foundational perspective for usefully understanding these realties. Constructing asymmetric models and testing these models with algorithmic tools (e.g., fsQCA) provide a way forward towards more useful theory and reducing the use of bad science practices.