The Gambling Symptom Assessment Scale (G-SAS): A reliability [610961]

The Gambling Symptom Assessment Scale (G-SAS): A reliability
and validity study
Suck Won Kima,⁎, Jon E. Granta, Marc N. Potenzab, Carlos Blancoc, Eric Hollanderd
aDepartment of Psychiatry, University of Minnesota Medical School, Minneapolis, MN, USA
bDepartment of Psychiatry, Yale University School of Medicine, New Haven, CT, USA
cNew York State Psychiatric Institute, College of Physicians & Surgeons, Columbia University School of Medicine, New York, NY, USA
dDepartment of Psychiatry, Mount Sinai School of Medicine, New York, NY, USA
Received 23 December 2006; received in revised form 15 May 2007; accepted 9 November 2007
Abstract
Two hundred seven patients with DSM-IV Pathological Gambling Disorder completed both the Gambling Symptom Assessment
Scale (G-SAS) and the Yale –Brown Obsessive –Compulsive Scale —modified for Pathological Gambling (PG-YBOCS) at baseline
visit and weekly or biweekly thereafter during the 12-week study period. The week 1 to week 2 visit data were used to assess test –
retest reliability. Weekly or biweekly data were used for the G-SAS validity. The PG-YBOCS reliability and validity data have been
published previously. We used the PG-YBOCS as the established scale and compared the G-SAS performance with the PG-YBOCS.Test–retest reliability was statistically significant. The correlations between the G-SAS and the PG-YBOCS and Clinical Global
Impression rating were excellent. Findings suggest that the G-SAS is reliable and valid in assessing changes in symptoms during a
drug treatment study.© 2007 Elsevier Ireland Ltd. All rights reserved.
Keywords: Gambling Symptom Assessment Scale (G-SAS); Yale –Brown Obsessive –Compulsive Scale-Pathological Gambling (PG-YBOCS);
Pathological Gambling Disorder; Rating scale; Reliability; Validity
1. Introduction
A need for gambling disorder treatment grew steadily
in recent years because of the expanding gamingbusiness in the U.S. and throughout the world, as wellas the increased recognition among clinicians and thegeneral public of the devastating consequences ofpathological gambling. Clinical investigators involved
in drug trials needed a validated rating scale to assess
changes in gambling symptoms. Lesieur and Blume
published the South Oaks Gambling Screen (SOGS) in1987 ( Lesieur and Blume, 1987 ). The SOGS is a
valuable instrument that is widely used in gamblingresearch; however, the instrument was designed primar-ily to screen for problem gamblers. Winters et al.reported a gambling symptom severity measurementscale for adolescents (Winters et al., 1993) .
Raylu and Oei (2004) published the Gambling Urge
Scale. This scale would serve well in assessing gamblingAvailable online at www.sciencedirect.com
Psychiatry Research 166 (2009) 76 –84
www.elsevier.com/locate/psychres
⁎Corresponding author. Department of Psychiatry, F282/2A West,
2450 Riverside Avenue, Minneapolis, MN 55454, USA. Tel.: +1 612273 9805; fax: +1 612 273 9779.
E-mail address: [anonimizat] (Suck Won Kim).
0165-1781/$ – see front matter © 2007 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.psychres.2007.11.008

urge symptoms. In 2005, de Castro et al. reported the
reliability and construct validity data on the GamblingFollow-up Scale (GFS), a physician-rated scale ( de Castro
et al., 2005 ). In the same year Petry (2005) presented the
reliability and validity data on the University of RhodeIsland Change Assessment (URICA) scale based on thedata from subjects with gambling disorder.
Recently, Hollander, Pallanti et al. modified the Yale –
Brown Obsessive –Compulsive Scale for use in gam-
bling disorder treatment studies (PG-YBOCS) (phy-sician-rated scale), and the psychometric data for thePG-YBOCS have been published ( Pallanti et al., 2005 ).
The G-SAS is a self-rated scale that encompassesgambling urges, thoughts, behaviors and interpersonalfunction. The scale description and its development aredescribed below.
1.1. The Gambling Symptom Assessment Scale (G-SAS)
1.1.1. Scale description
The G-SAS (see Appendix) is a 12-item self-rated scale
designed to assess gambling symptom severity and changeduring treatment. The G-SAS is not a diagnostic orscreening instrument. Each 12-item scale has a score
ranging from 0 to 4 (adjective anchors for 0 and 4 vary for
each item). All items ask for an average symptom based onthe past 7 days. Items 1 –4 ask for the average urge (only
gambling urges, excludes other urges) (analogous to PennAlcohol Craving Scale by Flannery et al., 1999 ) symptom
severity, frequency, duration and control; items 5 –7a s kf o r
the average frequency, duration and control of thoughtsassociated with gambling (excludes other thoughts); item 8
asks for the time spent on gambling or gambling-related
behavior, item 9 asks for the anticipatory tension and/orexcitement caused by an imminent gambling act; item 10asks for the excitement and pleasure associated withwinning, item 11 asks for emotional distress (mental pain,anguish, shame, guilt, or embarrassment) caused bygambling; and item 12 asks for personal trouble (relation-ship, financial, legal, job, medical or health) caused by
gambling. Every effort was given to enhance objectivity
and avoid an analog construct. Five items are semi-quantifiable and items 2, 5, and 8 are more objective thanthe rest. Items 1 –4 can be used to assess changes in urge
symptoms. The total score ranges from 0 to 48:extreme=41 –48, severe=31 –40, moderate=21 –30,
mild=8 –20.
1.1.2. Scale development
The G-SAS has incorporated the concepts used in the
Leyton Obsessional Inventory (LOI) ( Cooper, 1970 )a n d
the Yale –Brown Obsessive –Compulsive Scale (Y-BOCS)(Goodman et al., 1989a,b ). The LOI gathers information
through 46 inventory items encompassing obsessive orcompulsive symptoms and 23 trait items. The “yes”
answers given to each inventory item form the basis for anassessment of symptom severity in the form of resistanceand interference. The shortcomings of this scale are thatthe inventory items may not match the symptoms of agiven patient and as a result may lead to a low score even
though the patient may suffer from severe symptoms
(Kim et al., 1990 ). The Y-BOCS, on the other hand, asks
for an average of all obsessive –compulsive disorder
symptoms. Although this enhances the sensitivity of thescale, it does not allow investigators to analyze changes ina defined subgroup of symptoms such as washing orchecking over a treatment course.
The G-SAS items do not include specific individual
symptoms (such as thinking about stealing money to
gamble); rather, it broadly asks for symptoms thatpertain to the three domains (urges, thoughts, andgambling behavior). It also adopts an idea from the Y-BOCS and asks for average symptoms during the past7 days. The G-SAS does not assess the degree ofresistance against gambling-related thoughts and urges(see Section 4). In assessing the frequency of gambling,
the number of trips to a casino is counted instead of the
number of gambling activities in a given trip.
The G-SAS was developed not only to detect changes
in gambling symptom severity during a treatment butalso to provide the data on the pattern of changes for eachsubgroup of symptoms, allowing investigators toanalyze the variation of the treatment response patternunique to each treatment mode. For example, gambling
behavior may have stopped through a treatment method,
but the patient may still have severe gambling urges.
In addition to the rationale described above, the G-SAS,
a self-rated scale, was developed so that it could be used instudies in which physician administration of the scale(required for the PG-YBOCS) is not possible, such assurvey studies or large-scale studies.
The test-retest reliability of the previous version of the
G-SAS showed a good correlation: N=58, r=0.704. For
the internal consistency, Cronbach's α=0.890 and one-
factor model G-SAS item loading ranged from 0.661 to0.811 ( Kim and Grant, 2001 ). The two-factor model
showed that urge symptom item 1 (severity measure)
coheres with the significant gambling symptoms suchas gambling-related subjective d istress, interpersonal dif-
ficulty, and gambling activities. The G-SAS showed a good
convergent validity when compared with the Pathological
Gambling-Clinical Global Impression (PG-CGI) (Improve-ment): N=48, r=0.043 (the first visit in which changes
in symptom are recorded), visits 4 –13 (visit 4 = 1 week77 Suck Won Kim et al. / Psychiatry Research 166 (2009) 76 –84

after changes in symptoms are recorded) “r”values range
from 0.677 to 0.823, Pb0.01 (2-tailed) in all visits.
Each scale item interval has been adjusted to five,
and two new items have been added to enhance sensi-tivity. Because of the changes we made, we examinedthe psychometric properties of the revised scale basedon the preliminary data available from an active treat-ment study. The revised G-SAS showed a superior
convergent validity when compared with the PG-CGI
(Improvement and Severity): the median correlation betweenthe G-SAS vs. CGI (Improvement) over an 8-week period(Nranges 16 –8, V1 –V4)=0.784. For the G-SAS vs. CGI
(Severity) r=0.812. The preliminary psychometric data
and the revised G-SAS have been published ( Kim et al.,
2001).
In the present report we used the recent multicenter
nalmefene treatment data to assess the test –retest
reliability and validity for the G-SAS.
2. Methods
Fifteen study sites randomized 207 subjects either to a
nalmefene or a placebo group. Subjects were recruitedthrough newspaper advertisements and referrals for
medication treatment. All subjects met primary DSM-IV
diagnosis of Pathological Gambling Disorder, clinician-administered Structured Clinical Interview for Pathologi-cal Gambling (SCI-PG) ( Grant et al., 2004 ), a minimum
score of 5 or more on the South Oaks Gambling Screen(SOGS) ( Lesieur and Blume, 1987 ), a score of 2 or more
on the urge component of the G-SAS (moderately severeurges or more). Gambling behavior within 2 weeks before
enrollment was required. Negative pregnancy test was
required for women subjects. Exclusion criteria were 1)current DSM-IV (SCID) Axis I diagnosis except fornicotine dependence, 2) life-time bipolar affectivedisorder confirmed through SCID, 3) SCID substanceuse or dependence within the past 3 months, 4) gamblingtreatment within the past 6 months (Gamblers Anon-ymous treatment was allowed), 5) baseline Hamilton
Depression and Anxiety Rating Scale scores b17, 6)
positive urine drug screen except for cannabis, 7)clinically significant medical disorder, 8) concomitantuse of psychotropic medication. At the end of the 16-weektreatment, the nalmefene-treated group improved signifi-cantly more than the placebo group ( n=102, F=5.46,
P=0.006). Consent form was obtained from all study
participants.
The reliabilities of both the PG-YBOCS and the G-SAS
were measured using two repeated test scores from eachsubject in the placebo group. The variability of thesescores from the subject's mean value was determined andexpressed as a pooled standard deviation. This gives an
estimate of the repeatability of the test results for a givensubject and is known as the test –retest or within-subject
standard deviation (S.D.) ( Snedocor and Cochran, 1980 ).
Due to the difference in means and ranges of the twoinstruments, the coefficient of variation (CV) was com-puted for each subject to adjust for the different mea-surement scales and to allow for a statistical comparison
of reliability between the G-SAS and the PG-YBOCS.
The Spearman correlation was calculated for each test asanother measure of reliability. Because the G-SAS andPG-YBOCS provide ordinal categorical measures,weighted kappa for each item was computed as anothermeasure of test –retest reliability.
For the validity study, a correlation between the total
G-SAS score and the total PG-YBOCS score was
obtained. The two tests were categorized as above or not
above the median value and cross-classified to reflectthe amount of agreement.
At the end of the 16-week nalmefene trial, clinical
global improvement scores were obtained from the patientand from the physician. Both scales ranged from 1 to 7,with 1 = very much improved, 2 = much improved, 3 =minimally improved, 4 = unchanged, 5 = minimally
worse, 6 = much worse, and 7 = very much worse.
Changes in PG-YBOCS and G-SAS scores from baselineto week 16 were calculated as simple change scores and as% change from baseline. Spearman correlation coeffi-cients between these change scores and the global ratingsfrom the patient and physician were calculated andcompared using Fisher's r-to-ztransformation.
3. Results
3.1. Test –retest reliability and internal item consistency
The study did not include a placebo lead-in period, so
we used the week 1 to week 2 visits placebo group data forthe reliability study (in the current study, the G-SAS andPG-YBOCS data were obtained at screening visit but not
at baseline visit; thus, the week 1 and week 2 data
constitute the first week to week data) ( N=44). Table 1
presents the means and standard deviations of the G-SAStotal score and the PG-YBOCS total, thinking, and
Table 1
Week 1 and week 2 means and standard deviations
Placebo
group(N=44)G-SAS
Mean (S.D.)PG-YBOCS
Mean (S.D.)PG-YBOCS
thinkingMean (S.D.)PG-YBOCS
behaviorMean (S.D.)
Week 1 23.91 (8.19) 17.07 (7.92) 9.07 (3.34) 8.00 (5.03)
Week 2 22.64 (8.58) 15.20 (7.03) 7.82 (3.66) 7.39 (3.98)78 Suck Won Kim et al. / Psychiatry Research 166 (2009) 76 –84

behavior scores. The PG-YBOCS total score had a
significantly higher test –retest correlation than did the
G-SAS total score suggesting more reliability ( Table 2 ).
Table 3 shows excellent internal item consistency for
both the G-SAS and the PG-YBOCS. The G-SASshowed a higher internal item consistency (Cronbach'salpha) over the PG-YBOCS ( Table 3 ).Table 4 showsweighted kappa for G-SAS and PG-YBOCS items for
week 1 and week 2.
Test–retest simple correlations for the G-SAS were
0.56 ( Pb0.001) and 0.79 ( Pb0.001) for the PG-
YBOCS. To check if certain study subject data werecausing the G-SAS reliability to be low, we ran a
regression of visit 2 G-SAS total score on visit 1 G-SAS
total score and calculated the residual score for eachsubject. We then examined the distribution of the
residual scores using Tukey's Hinges ( Tukey, 1977 )t o
identify outliers. There were five outliers for the G-SAS
s c o r e s ,b u tt h e r ew e r en o n ef o rt h eP G – Y B O C S .Without the outliers, the G-SAS test –retest reliability
reached 0.781. The outliers may have occurred due in
part to the fact that the scale is rated by the patients.
Thus, the true correla tion score is still 0.56.
3.2. Validity
For the validity study comparing all subjects ( N=207)
at baseline, Spearman correlations between the PG-YBOCS total and the G-SAS total score were found
to be good (rho=0.51). At baseline, the PG-YBOCS and
G-SAS had 65% agreement when coded as above andbelow the median (see Table 5 ).
For the validity study, comparing changes during
treatment between the PG-YBOCS and G-SAS scores,Tables 6 and 7 present Spearman correlation coefficients
between the PG-YBOCS vs. G-SAS; the PG-YBOCS vs.physician-rated CGI; the G-SAS vs. physician-rated
CGI; the PG-YBOCS vs. patient-rated CGI; the G-SAS
vs. patient-rated CGI, respectively. Fisher r-to-ztrans-
formations were calculated to detect any significantdifferences between the correlations in the PG-YBOCSTable 2
Test–retest reliabilities of the G-SAS and the PG-YBOCS total scores
(N=44)
G-SAS total PG-YBOCS P-value of G-SAS
vs. PG-YBOCS
Mean score 23.27 16.14
Standard deviation 5.73 3.52CV 0.28 0.27 0.815
Rho 0.56 0.79 0.047
Standard deviation: standard deviation of within-patient variability;
CV: average coefficient of variability (paired t-test); rho = Spearman
correlation coefficient (Fisher's r-to-ztest). Mean score = mean of
week 1 and week 2 mean values for the placebo group. We used the
PG-YBOCS as the standard measure and compared the G-SASthrough S.D., CVand rho. All three values for each scale show that thePG-YBOCS was more reliable, but there was no statistical difference
between the two scales.
Table 3G-SAS and PG-YBOCS reliability analyses (internal item consistency)
Data
collectedG-SAS PG-YBOCS
Screen visit Screen visit
N 207 206
Scale mean 31.43 23.26
Scale S.D. 7.353 5.071Alpha 0.869 0.766
Alpha if item removed Alpha if item removed
Item #1 0.858 0.758
Item #2 0.862 0.728Item #3 0.852 0.751
Item #4 0.854 0.759
Item #5 0.862 0.750Item #6 0.853 0.737Item #7 0.853 0.728
Item #8 0.855 0.739
Item #9 0.858 0.775Item #10 0.873 0.741Item #11 0.861
Item #12 0.867
Scale mean = mean G-SAS and PG-YBOCS values from all study
subjects at baseline visit. The G-SAS shows a superior correlationalpha over the PG-YBOCS. Correlation alpha was also computed with
each item omitted to assess relative contribution of each item to the
overall alpha for each scale. For example, in a previous study, theresistance items within the Y-BOCS lowered the overall alpha. The
finding is extended in the current study (see the alpha without item 4 or
9 of the PG-YBOCS).Table 4G-SAS and PG-YBOCS reliability analyses: week 1 and week 2(weighted kappa)
G-SAS PG-YBOCS
Item #1 0.304 0.669
Item #2 0.528 0.509Item #3 0.393 0.573Item #4 0.339 0.731
Item #5 0.600 0.642
Item #6 0.485 0.699Item #7 0.311 0.568Item #8 0.526 0.509
Item #9 0.558 0.414
Item #10 0.523 0.566Item #11 0.394
Item #12 0.612
N=44; fair=0.21 –0.40; moderate=0.41 –0.60; substantial=0.61 –
0.80; almost perfect=0.81 –1.00 ( Bartko, 1991; Landis and Koch,
1977 ).79 Suck Won Kim et al. / Psychiatry Research 166 (2009) 76 –84

and G-SAS scores vs. the CGI ratings of change. For
subject-rated CGI, there is no difference between G-SASand PG-YBOCS in agreement with the subject's rating.
For the physician ratings, the PG-YBOCS total score
showed a better agreement than the G-SAS total score.
4. Discussion
The G-SAS data we present herein suggest that the
G-SAS is reliable and valid in assessing symptomseverity and changes in gambling symptoms during a
treatment. The PG-YBOCS test –retest reliability, how-
ever, was greater than that of the G-SAS. Cronbach'salpha for the baseline visit, on the other hand, shows thatthe internal consistency of the G-SAS is greater than thatof the PG-YBOCS.
The validity study showed that the G-SAS is valid in
measuring symptom severity at baseline and also inassessing changes in symptom severity during treatment.
Since the PG-YBOCS has been validated previously(Pallanti et al., 2005 ), we compared the G-SAS symptom
change scores with the % symptom change scores tothose of the PG-YBOCS. The correlation values for thesymptom change scores and % symptom change scoreswere good (0.81 and 0.85 each). When the physician-
rated CGI and the patient-rated CGI were used as
established standards, both the PG-YBOCS and G-SASscores showed good correlations except in the case of thephysician-rated CGI, where the PG-YBOCS scoresshowed a higher correlation value.
It is interesting to note that the correlation value
between the G-SAS (self-rated) total scores and the self-rated CGI scores is better (although a difference of 0.15
is well within random variation) than the correlation
value between the G-SAS total scores and the physician-rated CGI scores. The physician-rated PG-YBOCS totalscores are also more tightly correlated with thephysician-rated CGI scores than the patient-rated CGIscores. The variations in correlation values observedabove might be, to a small degree, linked to a tendencyby clinicians and patients to rate symptom severity
similarly for the two different scales.
In the original G-SAS we included ‘weekly amount of
money loss ’as a measure to assess gambling symptom
change, but treatment outcome study data (factor analysis)showed that the item that measures money loss showed thelowest factor loading (0.56) in comparison to the rest of thescale items (0.66 to 0.81). When we went back to themedical records and interviewed the patients who showed
a rather dramatic improvement in terms of money loss, we
found that the reason for not losing money was notnecessarily because of gambling symptom improvement,rather it was because they did not have money left toTable 5
Comparison of the G-SAS score and the PG-YBOCS total score
G-SAS at or
below medianG-SAS above
medianTotal
PG-YBOCS total
At or below median 68 36 104Above median 38 65 10364.3% agreement
Chi-square=16.813106 101 207
PG-YBOCS thinking
At or below median 74 40 114Above median 32 61 93
65.3% agreement
Chi-square=19.074106 101 207
PG-YBOCS behavior
At or below median 73 40 113
Above median 33 61 9464.7% agreementChi-square=17.867106 101 207
Screen visit data —all subjects ( n=207).
Excellent agreement (Chi-square) found between the G-SAS scoresand PG-YBOCS scores at baseline visit.
Table 6Drug group only
Correlated with Rho
Change score Change score
G-SAS PG-YBOCS total 0.812
PG-YBOCS thinking 0.777PG-YBOCS behavior 0.725
% Change score % Change score
G-SAS PG-YBOCS total 0.851
PG-YBOCS thinking 0.826PG-YBOCS behavior 0.764
N=125 (LOCF) correlations of G-SAS and PG-YBOCS change
scores.Table 7Drug group only
Change
scoreRho Rho Fisher
r-to-zP-value
G-SAS vs.
P-CGI0.690 PG-YBOCS total vs.
P-CGI0.794 1.825 0.068
PG-YBOCS thinking
vs. P-CGI0.711 0.321 0.748
PG-YBOCS behavior
vs. P-CGI0.765 1.249 0.212
G-SAS vs.
S-CGI0.705 PG-YBOCS total vs.
S-CGI0.729 0.385 0.700
PG-YBOCS thinking
vs. S-CGI0.647 −0.834 0.404
PG-YBOCS behavior
vs. S-CGI0.714 0.141 0.888
N=124 (LOCF) correlations with physician- and subject-rated CGI.
P-CGI = physician-rated CGI, S-CGI = subject-rated CGI.80 Suck Won Kim et al. / Psychiatry Research 166 (2009) 76 –84

gamble. So, it gave a false impression that they were
improving. Therefore, we removed the money loss itemfrom the G-SAS.
Likewise, we were careful not to include resistance
items in the G-SAS. This was based on our previousstudy, which showed that resistance items within the Y-BOCS adversely affected scale sensitivity ( Kim et al.,
1994 ). Ideally, each scale item should move in tandem
with the rest of the scale items in the course of symptom
change. In the case of the Y-BOCS, resistancesymptoms seem to move independent of the rest of thescale items. When symptoms are mild, patients often areable to resist and suppress the symptoms. As thesymptoms escalate, resistance symptoms also seem toincrease along with the rest of the gambling symptoms.However, when symptoms become severe, resistance
symptoms seem to decrease in intensity while the rest of
the gambling symptoms seem to increase in intensity.That is, resistance symptoms seem to move in anopposite direction from the rest of the scale items.Therefore, resistance symptoms are viewed as a healthypart (not a disease process) of the self trying to cope withthe pathological side (PGD) of the disease process.
Although not the primary objectives of this article,
the two- and three-factor, factor-analytic study (Varimax
rotated) of the current version of the PG-YBOCSshowed that resistant items within the gamblingobsessions (item 4) and gambling compulsions sections(item 9) segregate and affect the scale performanceadversely (data not shown). This finding essentiallyconfirms our previous findings ( Kim et al., 1994 ) that
resistance items do not change in tandem with the rest of
the gambling symptoms. The overall psychometric datawe have on the PG-YBOCS suggest that the scale
performs very well during clinical trials; we believe,however, that with minor adjustment, the scale wouldperform even better.
Limitations: In developing the G-SAS we paid close
attention to the urge symptom subscales. Most gamblershave reasonably strong gambling urge symptoms andsome of them have very strong urges but not all. Some
gamblers, especially chronic gamblers, often present
mild or no urges to gamble or excitement after winning.To them gambling is a form of habit, very much likewhat clinicians see from chronic drug users. We expectthat the G-SAS would not be as sensitive for thesepatients. Also, for those patients who gamble primari-ly to escape from depression, loneliness or stress, theG-SAS may not be as sensitive a scale as it was intended
to be. Investigators using the G-SAS need to pay close
attention during the first and second visits to ensurethat each study subject understands the questionnairescorrectly and also to minimize carelessness that mightarise due to the fact that it is a self-rated scale. Also,a self-rated scale is usually inferior to a clinician-ratedscale.
Acknowledgment
The multisite study, from which the data reported in
the current study were derived, was sponsored by BioTieTherapies, Helsinki, Finland. The work was, in part,supported by grants from the National Institute of MentalHealth (K23 MH069754-01A1) to Dr. Grant and theNational Institute of Drug Abuse (R01 DA013039; R01
DA019039) to Dr. Potenza.
Appendix A. Gambling Symptom Assessment Scale (G-SAS)
The following questions are aimed at evaluating gambling symptoms. Please
read the questions carefully before you
answer.
1) If you had unwanted urges to gamble during the past WEEK, on average, how strong were your urges? Please
circle the most appropriate number:
2) During the past WEEK, how many times did you experience urges to gamble? Please circle the most appropriate
number.
0) None
1) Once
2) Two to three times3) Several to many times4) Constant or near constant81 Suck Won Kim et al. / Psychiatry Research 166 (2009) 76 –84

3) During the past WEEK, how many hours (add up hours) were you preoccupied with your urges to gamble? Please
circle the most appropriate number.
4) During the past WEEK, how much were you able to control your urges? Please circle the most appropriate number.
5) During the past WEEK, how often did thoughts about gambling and placing bets come up? Please circle the most
appropriate number.
0) None
1) Once2) Two to four times3) Several to many times4) Constantly or nearly constantly
6) During the past WEEK, approximately how many hours (add up hours) did you spend thinking about gambling
and thinking about placing bets? Please circle the most appropriate number.
7) During the past WEEK, how much were you able to control your thoughts of gambling? Please circle the most
appropriate number.
8) During the past WEEK, approximately how much total time did you spend gambling or on gambling related
activities? Please circle the most appropriate number.
9) During the past WEEK, on average, how much anticipatory tension and/or excitement did you have shortly
before you engaged in gambling? If you did not actually gamble, please estimate how much tension and/or excitementyou believe you would have experienced if you had gambled. Please circle the most appropriate number.
82 Suck Won Kim et al. / Psychiatry Research 166 (2009) 76 –84

10) During the past WEEK, on average, how much excitement and pleasure did you feel when you won on your bet.
If you did not actually win at gambling, please estimate how much excitement and pleasure you would haveexperienced if you had won. Please circle the most appropriate number.
11) During the past WEEK how much emotional distress (mental pain or anguish, shame, guilt, embarrassment) has
your gambling caused you? Please circle the most appropriate number.
12) During the past WEEK how much personal trouble (relationship, financial, legal, job, medical or health) has
your gambling caused you? Please circle the most appropriate number.
Maximum score=48
Extreme=over 40
Severe=31 –40
Moderate=21 –30
Mild=8 –20
References
Bartko, J.J., 1991. Measurement and reliability: statistical thinking
considerations. Schizophrenia Bulletin 17, 483 –489.
de Castro, V., Fuentes, D., Tavares, H., 2005. The gambling follow-up
scale: development and reliability testing of a scale for pathological
gamblers under treatment. Canadian Journal of Psychiatry 50,
81–86.
Cooper, J., 1970. The Leyton Obsessional Inventory. Psychological
Medicine 1, 48 –64.
Flannery, B.A., Volpicelli, J.R., Pettinati, H.M., 1999. Psychometric
properties of the Penn Alcohol Craving Scale. Alcoholism,Clinical & Experimental Research 23, 1289 –1295.
G o o d m a n ,W . K . ,P r i c e ,L . H . ,R a s m u s s e n ,S . A . ,M a z u r e ,C . ,
Fleischmann, R.L., Hill, C.L., Heninger, G.R., Charney, D.S.,1989a. The Yale –Brown Obsessive Compulsive Scale. I. De-
velopment, use, and reliability. Archives of General Psychiatry 46,
1006 –1011.
Goodman, W.K., Price, L.H., Rasmussen, S.A., Mazure, C., Delgado,
P., Heninger, G.R., Charney, D.S., 1989b. The Yale –Brown
Obsessive Compulsive Scale. II. Validity. Archives of General
Psychiatry 46, 1012 –1016.
Grant, J.E., Steinberg, M., Kim, S.W., Rounsaville, B., Potenza, M.N.,
2004. Preliminary validity and reliability testing of a StructuredClinical Interview for Pathological Gambling (SCI-PG). Psychia-
try Research 128, 79 –88.Kim, S.W., Grant, J.E., 2001. An open naltrexone treatment study in
pathological gambling disorder. International Clinical Psychophar-
macology 16, 285 –289.
Kim, S.W., Dysken, M.W., Kuskowski, M., 1990. The Yale –Brown
Obsessive –Compulsive Scale: a reliability and validity study.
Psychiatry Research 34, 99 –106.
Kim, S.W., Dysken, M.W., Pheley, A.M., Hoover, K.M., 1994.
The Yale –Brown Obsessive –Compulsive Scale (Y-BOCS):
measures of internal consistency. Psychiatry Research 51,
203–211.
Kim, S.W., Grant, J.E., Adson, D.E., Shin, Y.C., 2001. Double-
blind naltrexone and placebo comparison study in the treat-
ment of pathological gambling. Biological Psychiatry 49,
914–921.
Landis, J.R., Koch, G.G., 1977. The measurement of observer
agreement for categorical data. Biometrics 33, 159 –174.
Lesieur, H.R., Blume, S.B., 1987. The South Oaks Gambling
Screen (SOGS): a new instrument for the identification ofpathological gamblers. American Journal of Psychiatry 144,1184 –1188.
Petry, N.M., 2005. Stages of change in treatment-seeking patho-
logical gamblers. Journal of Consul ting and Clinical Psychology 73,
312–322.
Pallanti, S., Decaria, C.M., Grant, J.E., Urpe, M., Hollander, E., 2005.
Reliability and validity of the Pathological Gambling adaptation of
the Yale –Brown Obsessive –Compulsive Scale (PG-YBOCS).
Journal of Gambling Studies 21, 431 –443.83 Suck Won Kim et al. / Psychiatry Research 166 (2009) 76 –84

Raylu, N., Oei, T.P., 2004. The gambling urge scale: development,
confirmatory factor validation, and psychometric properties.Psychology of Addictive Behaviors 18, 100 –105.
Snedocor, G.W., Cochran, W.G., 1980. Statistical Methods. Iowa State
University Press, Ames, IA, pp. 238 –246.Tukey, J.W., 1977. Exploratory Data Analysis. Addison-Wesley,
Reading, Mass.
Winters, K.C., Stinchfield, R.D., Fulkerson, J., 1993. Toward the
development of an adolescent gambling problem severity scale.
Journal of Gambling Studies 9, 63 –84.84 Suck Won Kim et al. / Psychiatry Research 166 (2009) 76 –84