The first analysis examined the endorsement rate of the various DSM-IV criteria. Table 1 shows the overall endorsement rate and also breaks it down by DSM-IV diagnostic category. As shown, criterion 8 (illegal acts) is one of the least often endorsed, even amongst offenders. Criteria 9 (jeopardizing a relationship, job, education, or career) and 10 (relying on others for money), however, were endorsed less often in total than illegal acts. The overall alpha for the criteria was 0.86; all criteria had substantial item total correlations ranging from 0.49 to 0.60. The illegal acts criterion did have the lowest item total correlation: 0.49. However, the alpha if deleted was lower for all ten criteria indicating that all of the criteria contributed to reliability of the scale.
The effect of the two rule changes on the percentage of offenders with a gambling disorder is shown in Table 2. The percentage of offenders identified as having a gambling disorder increased from 7.4 to 10.2 % using a cut score of four rather than five. Removing illegal acts, however, resulted in a reduction of the percentage from 7.4 to 6.7 % if the cut off was five or more, and from 10.2 to 9.3 % if the cut off was four or more.
Table 3 shows cross tabulations of the DSM-IV categorization with the various other possible DSM-5 models. Dropping illegal acts but retaining the cut off of five resulted in five offenders (10 %)—previously included as pathological gamblers—not being included in that same category. As expected, using a cut score of four increased the number of people who were classified as having a gambling disorder. As shown in Table 3, decreasing the cut score to four resulted in an increase from 50 to 63 (+26 %) in the number of offenders diagnosed with disordered gambling if illegal acts was removed, and an increase from 50 to 69 (+38 %) if illegal acts was retained. Table 4 shows the cross tabulation of the 4 of 9 model (excluding illegal acts) with the 4 of 10 model (with illegal acts). The removal of illegal acts produced a decrease in identified gambling disorders from 69 to 63.
As noted above, using a cut off of four with illegal acts, six more people were identified as having a gambling disorder than without illegal acts. If these additional cases represent actual cases, then the inclusion of illegal acts for case versus non-case categorization should yield higher convergent validity correlations than without illegal acts. Table 5 shows the correlation of the gambling frequency, SOGS, CPGI/PGSI, and a harmful consequences measure with the original DSM criteria, with either a cut score of four or five, and with or without illegal acts. It also shows the correlations of the various cut scores with the same variables. In terms of correlations, a cut off of four with or without illegal acts produced higher correlations than a cut off of five. However, the DSM scoring with illegal acts included had somewhat higher correlations with the CPGI/PGSI total score, gambling frequency, SOGS scores, and harmful consequences than without illegal acts. In addition, the kappa for the CPGI/PGSI (eight or more), and the SOGS (five or more) was highest for the 4 of 10 model that included illegal acts. By including illegal acts the correlation between the CPGI/PGSI and the DSM-5 increased from 0.671 to 0.696 and this difference accounts for 3.4 % of the variance. The change from the DSM-IV (5 of 10 criterion) to the revised DSM-5 (4 of 9 criterion) was an increase from 0.654 to 0.671 and this difference accounts for 2.3 % of the variance. Similarly, the highest convergent validity coefficients for the SOGS and for harmful consequences were found for a DSM-5 total that included the illegal acts criterion (4 of 10 criterion).
Next we examined the relationship between each of the DSM criteria and other measures of gambling problems to see if illegal acts played an important role in convergent validity. To do this we conducted a forced entry regression analysis of the DSM criteria with the CPGI/PGSI. The same analysis was also conducted with the SOGS, gambling frequency and harmful consequences. The question of interest was whether illegal acts contributed significantly to the relationship between DSM-IV criteria and these other measures of gambling problems. The results are shown in Table 6. Of particular note: the illegal acts criterion was significantly related to the total score for each of the four measures. Only two other criteria were significant in all four analyses; criterion 2 (gambling with increasing amounts of money) and criterion 6 (chasing).
To examine the importance of illegal acts in terms of severity of disordered gambling, we grouped people into mild (4–5), moderate (6–7) or severe (8–10) gambling disorders with and without the illegal acts criterion. As shown in Table 7, the illegal acts criterion affects the grouping of several individuals. The data indicate that by including illegal acts, more individuals would be identified as having moderate and severe gambling disorders. Interestingly, the illegal acts criteria even increased the number of sub-clinical problem gamblers that were identified.
As noted above, the illegal acts criterion is more often endorsed by disordered gamblers in the correctional system. One option for reducing the possibility of missing some disordered gamblers is if illegal acts are assessed under the “lying to other” criterion (criterion 7). This is a solution explicitly endorsed by DSM-5 (American Psychiatric Association 2013; Grant 2013). In particular they state “these instances of deceit may also include, but are not limited to, covering up illegal behaviors such as forgery, fraud, theft, or embezzlement to obtain money with which to gamble” (American Psychiatric Association 2013, p. 586). To investigate this as an option we conducted an analysis combining illegal acts (criterion 8) and lying to others (criterion 7) to determine if this combination would improve the scale. This entailed scoring the person a 1 if they had endorsed either criterion 7 or criterion 8. It should be noted that this is not identical to what is in DSM-5, where illegal acts are only assessed in terms of the deceit used and not the illegal act itself. This analysis suggested an improvement compared to the DSM-5 without illegal acts. As noted above in Table 2, 63 offenders (9.3 % of the total) were identified as GD without illegal acts and 69 (10.2 %) were identified as GD with illegal acts. By combining criteria 7 and 8, the number of people identified as GD was 67 (9.9 %).