‘Commentary’ articles are new to the Journal of Gambling Studies, and therefore articles that reply to Commentaries are equally novel. Nevertheless, we felt that we should address some of the issues raised in Griffiths (2014), concerning our study ‘Rates of problematic gambling in a British homeless sample: A preliminary study’ (Sharman et al. 2014).

The Commentary provides some contextual literature on mental health problems in the homeless, and draws attention to the need for extra care in the interpretation of our study, given the potential for data to be used by those with a specific legal agenda. Naturally, the same care is required in the interpretation of the Commentary.

One of Griffiths’ key claims is that “One of the most notable findings in the Sharman et al. (2014) study—and on which there was no comment—was the fact that 80 % of the 456 homeless people (n = 363) has not gambled in the year prior to the study.” This claim is incorrect, and appears to arise from a misunderstanding of our description of our procedure. As we felt would be clear from a careful reading of our manuscript, the 363 individuals referred to are those who did not agree with the screening item “In the last 12 months, have you bet more than you could afford to lose?”, which was used as a precursor to the full Problem Gambling Severity Index, and 363 of 456 participants (80 %) responded negatively to this item. Clearly, one cannot conclude that these individuals have not gambled at all in the past year.

We are particularly disappointed with the final section of the Commentary, not least due to it being titled ‘Factual Errors’ (a term that is also used in the abstract of the Commentary). In our view, this section does not refer to any genuine ‘factual errors’, only to typographic errors and differences of opinion. We will address each of the five points raised in turn.

‘Finally, it should also be pointed out that the Sharman et al. (2014) study made a number of factual errors in their paper. Firstly, the authors claimed in the ‘Discussion’ section that changes in the rate of problem gambling in Great Britain as reported in the BGPS have been “minimal” over time. However, the rate of problem gambling increased 50 % between the 2007 study (Wardle et al. 2007) and the 2011 study as measured by the DSM-IV (Wardle et al. 2011). This was a statistically significant increase in the rate of problem gambling’.

In the 2010 BGPS, the rate of problem gambling as measured using DSM criteria did increase 50 % from 0.6 to 0.9 % (p = .049). Using the PGSI (the more relevant measure, in that it was the scale used in our survey) the change was non-significant, from 0.5 to 0.7 % (p = .23). Even if we had wished to describe the former change, an increase of 0.3 % of the population can still, we feel, be justifiably interpreted as minimal, even if that minimal increase is statistically significant due to the large sample size.

Our interpretation is actually very much in line with the BGPS report itself, which recommends that ‘caution is exercised when interpreting (the DSM) result’ as it is driven in real terms by an increase of just 17 individuals. Furthermore, in the subsequent 2012 Health Survey for England (Wardle and Seabury 2013), PGSI problem gambling was evident in 0.6 % of males (and 0.1 % of females), and DSM problem gambling was evident in 0.8 % of males (and 0.2 % of females). Therefore, a general conclusion that rates are stable seems entirely reasonable.

‘Secondly, the authors claimed the screening tool they used (the Problem Gambling Severity Index) is “clinically recognized diagnostic tool” (p. 6). However, this is not true. The PGSI was not developed for clinical use at all but was specifically designed for epidemiological use’.

The PGSI was developed for epidemiological use, but it has now widely succeeded the South Oaks Gambling Screen as a continuous measure of gambling severity, in both community and clinical research. While we acknowledge some mild hyperbole in our description, the validation of PGSI cutoff scores for indicating problem gambling (e.g. Currie et al. 2013) supports our claim that it is a ‘clinically recognized diagnostic tool’.

‘Thirdly, the authors claim that the prevalence rate of “disordered gambling” (p. 2) in Great Britain using the PGSI is 0.7 %. However, the BGPS has never used the term ‘disordered gambling’ or assessed ‘disordered gambling’ in any of the three published studies to date. Those in the gambling studies field should not use such terms interchangeably without defining them first’.

Disordered gambling is a useful umbrella term given the semantic difficulties in applying the terms ‘problem gambling’ and ‘pathological gambling’ across studies that use different measures and methodologies. These inconsistencies frustrate many in the field, and contributed to the name change to ‘gambling disorder’ in the DSM-5 in 2013.

‘On a minor note, the seminal researcher Sheila Blume (co-developer of the South Oaks Gambling Screen [Lesieur and Blume 1987]—one of the most widely used screens in the gambling studies field—was cited as ‘Bloom’ not ‘Blume’). Additionally, the authors kept citing the most recent BGPS as being published in 2010 when in fact it was 2011 (Wardle et al. 2011)’.

It is unfortunate that Dr Blume’s surname was misspelt once (although correctly in the bibliography) (Lesieur and Blume 1987), and that the 2010 British Gambling Prevalence Survey was not cited as Wardle et al. (2011). These typographic errors are embarrassing, but they do not deserve to be described as ‘factual errors’: individual words within citations do not convey facts. For comparison, we note that Prof. Griffiths makes a factual error in the Commentary when referring to PGSI cutoffs: a score of 7 would be classified ‘moderate risk’ not ‘problem gambling’. Should the Journal of Gambling Studies decide to include future Commentaries as a publication format, we appeal to future authors to aim for more insightful critique than the cataloging of minor typos.