RESEARCH QUESTIONS
Are all the questions in the Problem Gambling Severity Index (PGSI) measuring the same factor? Does the factor structure change if only non-problem gamblers are assessed using the PGSI? How are PGSI scores related to measures of gambling frequency?
PURPOSE
As gambling opportunities are increasingly available, assessing and understanding problem gambling is a research priority. Brief self-report measures that work well in both general and clinical populations are essential tools both for researchers and as screening instruments for problem gambling. This research was performed to examine the validity of one such tool, the Problem Gambling Severity Index (PGSI).
HYPOTHESES
None stated.
PARTICIPANTS
Participants were 12,299 adults (49% male) aged 18 years and older. All participants had reported gambling in the past year.
PROCEDURE
Data were obtained from the Ontario Problem Gambling Research Centre and were collected through prior telephone surveys conducted in several different Canadian provinces (i.e., Ontario, Alberta, British Columbia, Manitoba and Newfoundland) between 2001 and 2005. Participants completed measures of problem gambling, gambling frequency, and gambling cognitions.
MAIN OUTCOME MEASURE
All participants completed the Canadian Problem Gambling Index (CPGI). The PGSI is a 9-item self-report measure of problem gambling contained within the CGPI. Four items assess problem gambling behaviours (e.g., “How often have you bet more than you can afford to lose?”) and five items assess adverse consequences of gambling (e.g., “How often has your gambling caused you any health problems, including stress or anxiety?”). Following convention, participants were classified into gambling subtypes based on their PGSI scores as follows: 0 = non-problem gambler; 1–2 = low risk gambler, 3–7 = moderate risk gambler, 8 and over = problem gambler. Frequency of participation in several gambling activities (i.e., lottery, horse race betting, internet gambling, bingo, raffles, sports betting, slots, and bookie) was obtained from questions in the CPGI. Faulty cognitions were assessed via two self-report questions, one asking about belief in the gambler’s fallacy and the other assessing belief in gambling systems. For a subset of 3283 cases, problem gambling was also assessed using the 20 item self-report SOGS.
KEY RESULTS
Analysis of the entire sample confirmed that the PGSI measured just one factor; all items appear to be measuring problem gambling. However, when examined by gambling subtype, more factors emerged. For low risk gamblers, there were six factors identified indicating the lack of any clear factor structure. For the moderate risk group there were four factors identified. For the problem gambling group three factors were identified: awareness of the existence of a gambling problem; dysfunctional behaviours (e.g., chasing losses, borrowing to get gambling money); and guilt/betting more than can afford. Supporting the validity (i.e., accuracy) of the PGSI, gambling frequency was different between gambling subtypes. Belief in the gambler’s fallacy and usefulness of gambling systems increased as problem gambling status increased. Compared to the PGSI, the SOGS had a less clear factor structure, less internal consistency, and weaker relationships with behavioural measures of gambling.
LIMITATIONS
The PGSI and SOGS analyses are not directly comparable due to the much smaller sample size for the SOGS analysis. The telephone survey may not represent households without a landline.
CONCLUSIONS
The PGSI appears to be a viable alternative to the SOGS for assessing degrees of problem gambling severity in a non-clinical context.
