How survey mode affects estimates of the prevalence of gambling harm: a multisurvey study

Recent general population surveys have produced highly variable estimates of the extent of problem gambling in Great Britain, ranging from as low as 0.4% to as high as 2.7% of adults. This level of uncertainty over the true level of problem gambling creates difficulties for policy makers and those planning treatment and support services for individuals and families affected by problem gambling. In this article, we assess the extent to which differences in approaches to sampling and measurement between surveys contribute to variability in estimates of problem gambling. We compare estimates of problem gambling using the Problem Gambling Severity Index across eight different surveys conducted at approximately the same time but which use different sampling and measurement strategies. Our findings show that surveys conducted online produce substantially higher estimates of problem gambling compared with in-person interview surveys. This is because online surveys, whether using probability or non-probability sampling, overrepresent people who are more likely to gamble online and to gamble frequently, relative to the proportions of these groups in the general population.