High quality gambling participation and prevalence studies are important to monitor and manage risks associated with an ever changing gambling environment. Such studies are expensive and choices must be made to balance precision and cost constraints. Common approaches include not screening infrequent gamblers and/or gamblers with low gambling expenditure, on the assumption that these gamblers are unlikely to be problem gamblers. Some studies combine moderate risk and problem gamblers to obtain robust groups for analysis. The resultant methodological variations compromise comparability and interpretation of changes over time and across jurisdictions. The Victorian Gambling Study 2008-2012 (VGS) and the Swedish Longitudinal Gambling Study 2008-2014 (Swelogs) are large jurisdiction-wide surveys with sample sizes of 15,000 and 8,165 respectively. Both studies screened all past-year gamblers. This paper reports on the modelling of some common methodological variants. Understanding the impact of these different variations enables better design of future studies and facilitates comparisons between existing prevalence studies within and across jurisdictions.