Introduction

In their systematic review of the literature, Ladouceur et al. (2017) evaluated the extent to which peer-reviewed empirical evidence supported the effectiveness of current responsible gambling strategies. Noting that investigators conduct literature reviews and report all available studies including those containing meaningful methodological flaws, these authors adopted the following four inclusion criteria on an a priori basis: (1) studies conducted within real gambling environments with ‘real’ gamblers and which included one of the following elements: (2) a matched control or comparison group; (3) repeated measures; and (4) one or more measurement scales. From an initial pool of 2548 articles identified in the literature search, only 29 articles met at least one of these inclusion criteria. Ladouceur et al. (2017) concluded that there was limited evidence supporting the overall effectiveness of extant responsible gambling strategies and argued for more rigorous scientific studies to be undertaken. Given vested interests and suggestions that funding sources influence the conduct of research, we conducted the present study to empirically investigate whether or not there were differences in the design and methodology of studies funded from industry or other sources. To our knowledge, this research is the first of its kind focusing on gambling studies and represents an initial step in investigating the extent to which funding influences research methodologies and, ultimately the interpretation of data.

Critics of gambling industry funded research claim that the funding sources influence what research questions will be investigated, the design/methodologies of studies, and biases in the interpretation of data (Adams et al. 2010; Andréasson and McCambridge 2016; Livingstone and Adams 2016). This might be due to researchers fearing future funding applications being compromised through subtle influences on the conduct of research and biases in the interpretation of data, that is, neglecting or overlooking negative findings that are critical or not in the interests of the funders. For example, Cassidy (2014) encouraged “…others to continue to reflect on their own activities and to expose gambling research to critical attention.” Like Cassidy others have speculated that gambling or alcohol industry funding has biased research and compromised investigative integrity (Livingstone and Adams 2016; Adams et al. 2010; Andréasson and McCambridge 2016). Although the focus of current criticisms is on industry funding, others have argued that government funding is open to similar criticisms given the central roles governments playing legislating, regulating and benefit through taxation revenue (Livingstone and Adams 2016).

Some stakeholders have identified the parameters of conditions under which scientists and industry can coexist ethically (Wilsnack et al. 2016) while others advance the principle that no research should be funded by proceeds of gambling (Livingstone and Adams 2016). This prohibitionist funding principle is an ideal but not a realistic stance given that all gambling funding emanates from gambling proceeds whether it originates directly from the gambling industry or government taxation revenue. The primary issue is not so much the source or origin of funding, but rather guaranteeing processes inclusive of independent peer-review, non-interference in design and methodology, freedom to publish without constraints, open and transparent disclosure of funds, and sufficient methodological procedures to allow independent replication of results. Biases and the conduct of research emanate not only from the vested interests of industry or government but also from academics representing advocacy groups masked by the face of public health. Irrespective of their discipline, scientists are governed by ethical principles and codes of conduct such that any claims of undue influence, whatever the source, must be empirically justified. Innuendos and simple assertions have no place in the scientific process.

A number of academics have expressed considerable concern about gambling research and the influence of funding from the gambling industry (Orford 2002; Adams 2007; 2011; Hancock and Smith 2017). We do not dispute that open and transparent disclosure of funding sources is a necessary and important step toward informing the reader about the degree of independence associated with a research publication. However, the fact that industry or governments have funded a research project does not, per se, indicate that the conduct of research and its findings are flawed, biased or misleading. Critics, as in the case of the tobacco and pharmaceutical industries, need to rely on empirical evidence challenging reported findings through the process of independent replication, or by identifying and reporting any identified methodological flaws and inadequacies in the research design protocol, measurements and/or interpretation of data.

As a first step, we have examined a segment of gambling research to determine empirically whether, compared to non-industry, industry funded research differs with respect to key design and methodological variables. This study represents a critical first step to empirically evaluate the potential for funding bias. The body of RG scientific literature provides a foundation that guides evidence-based RG strategies around the world.

Method

From a systematic search of the primary academic databases of peer-reviewed publications in the area of responsible gambling published from 1962 to October, 2015, (see Ladouceur et al. 2017), we identified 29 studies that satisfied inclusion criteria. We reexamined these 29 studies to determine whether there was an association between funding sources and study characteristics. The funding sources included five categories: (1) Government (Federal/State/or Local Government); (2) non-profit (e.g. NCRG)/or independent research center (e.g. Gambling Research Exchange Ontario (GREO), (3) Industry funded, (4) Public Health agency (e.g. WHO), and (5) missing. To measure study characteristics, we used the following eight variables: (1) type of RG initiative (e.g., self-exclusion, pop-up messages, pre-commitment, etc.); (2) presence of a comparison or matched control group; (3) use of a measurement scale (i.e. screener); (4) inclusion of repeated measures; (5) publication source (e.g., scientific journal that published the study); (6) Number of criteria met in our previous meta-analysis; (7) secondary funding source (i.e., studies reporting more than one source of funding); and (8) year of publication (i.e., to examine the chronological pattern of progression for funding sources).

Results and Discussion

Fisher’s exact tests were conducted to examine potential associations between eight characteristics of the studies and the sources of funding. To manage the likelihood of an inflated Type 1 error for multiple comparisons, we conducted Fisher’s exact tests with a Bonferroni adjustment (i.e., p < 0.0025). After adjusting the alpha according to this procedure, none of the associations were statistically significant. Table 1 summarizes these statistical tests and includes all study covariates.

Table 1 Tests for association: funding source and eight covariates
Full size table

This preliminary study represents the first attempt to determine whether funding source influences the topic and methodological rigor of RG activities. The results reveal that there is no evidence of difference in methodologies employed as a function of funding source. These results highlight the importance and need for further investigations into the evidence to justify the negative criticisms toward industry funding research, such as the those that Cassidy (2014) and Livingstone and Adams (2016) formulated. Their negative comments regarding research biases and, more importantly, their suggestion that no researchers should accept funding from the industry remain to be supported by empirical evidence. At the moment, these positions of influence and prohibition rest more on ideological and personal opinions than on evidence-based arguments.

Nonetheless, readers must consider these findings with caution. There are many potential reasons for null findings. We do not want to interpret that absence of statistical differences to suggest that there are no factors that can bias responsible gambling research or the interpretation of data. The latter is established through independent replication of results. It is possible that the relatively small sample size or small effect size limited the power to identify statstically significant differences. Other features of this study might have introduced statstical noise making it difficult to identify sources of influence that can bias responsible gambling differences.

Finally, an important secondary finding reveals that more than 30% of the studies included in this analysis did not reveal their sources of funding. We encourage scientific and scholarly publications to require and request this information before publishing responsible gambling research articles.

In conclusion, as the Reno model suggests, the field of gambling studies needs to develop an empirical approach toward responsible gambling (11). This strategy includes carefully examining the potential influence of funding sources on responsible gambling related studies. Relying on opinions and morality risks bias of a personal kind; this circumstance will not advance the science of gambling studies.