Psychology
2011. Vol.2, No.4, 342-354
Copyright © 2011 SciRes. DOI:10.4236/psych.2011.24054
Problem Gambling, Gambling Correlates, and Help-Seeking
Attitudes in a Chinese Sample: An Empirical Evaluation
Jasmine M. Y. Loo, Tian Po Oei, Namrata Raylu
School of Psychology, The University of Queensland; Brisbane, Australia.
Email: jasminelmy@help.edu.my
Received March 17th, 2011; revised April 19th, 2011; accepted May 21st, 2011.
There is an increasing consensus that problem gambling (PG) is a serious social issue among the Chinese, but
little is known of the factors associated with PG among the Chinese using validated and improved PG measure-
ments. This study examined the patterns of PG and the PG predictive ability of variables such as gam-
bling-related cognitions, gambling urge, depression, anxiety, stress, and help-seeking attitudes among Chinese
individuals living in Taiwan. The participants consisted of 801 Taiwanese Chinese student and community indi-
viduals (Mean age = 25.36 years). The prevalence of PG (Problem Gambling Severity Index; PGSI) and patho-
logical gambling (South Oaks Gambling Screen; SOGS) are higher in this Taiwanese Chinese sample as com-
pared with past prevalence research. Significant differences were found between PGSI groups (i.e., non-PG,
low-risk, moderate-risk, and PG) in socio-demographic variables. Erroneous gambling-related cognitions and
overall negative psychological states significantly predicted PG. In addition, interaction effects of gender, me-
diation effects, and the predictive ability of help-seeking attitudes were discussed. The findings of this study
have important implications in the understanding of PG among the Chinese. Gambling-related cognitions and
negative psychological states are important factors that should be addressed in intervention programs.
Keywords: Chinese, Gambling, Problem Gambling, Help-Seeking, Cognitions, Psychological States
Introduction
The earliest documented accounts of gambling were recorded
in China where “keno” was first played 3000 years ago to fund
the building of the Great Wall (National Policy Toward Gam-
bling, 1974). Gambling was very popular in ancient China and
throughout Chinese history despite the fact that it was under
strict legislative controls and banned in some regions. Despite
being illegal in mainland China (except in Macau where casino
gambling is legalized) and Taiwan (except outlying Penghu
islands), gambling remains popular among the Chinese around
the world (i.e., Chinese Diaspora) due to the fact that it is an
acceptable form of social activity in the community (Hobson,
1995; Lai, 2006; Raylu & Oei, 2004b). In fact, social gambling
is expressed as a form of entertainment, often occurring during
festive seasons (e.g., Chinese New Year), birthday gatherings,
or wedding celebrations. This activity usually happens with
friends, family, or colleagues, and the gambling episode lasts
for a limited period of time without loss of control (Clarke, Tse
et al., 2006). Nevertheless, social gambling can escalate to se-
rious social gambling, problem gambling, and pathological
gambling.
To date, most researchers have concurred that the term prob-
lem gambling (PG) refers to gambling behaviour that is severe
enough to create negative outcomes for the problem gambler,
immediate family, and social networks (Brooker, Clara, & Cox,
2009; Raylu & Oei, 2002). Similarly, in this study, problem
gambling will be used in a broader sense to define the situation
where an individual is experiencing gambling problems that
causes disruption to the individual’s functioning that may ex-
tend to affect family members and social networks (Lesieur &
Blume, 1987). The term pathological gambling will be used to
define individuals who meet the diagnostic criteria in the Di-
agnostic and Statistical Manual, Fourth Edition, Text Revision
(DSM-IV-TR; American Psychiatric Association, 2000). For
the purpose of this study, gambling problems are conceptual-
ized on a latent continuum of duration and severity.
When gamblers win or lose, a range of readily perceptible
cognitions, emotions, and behaviours are evoked. In turn, they
drive a vicious cycle of excessive gambling with detrimental
consequences such as financial debt, work and health issues,
and strained relationships (Loo, Raylu, & Oei, 2008). These
detrimental effects of PG affects both problem gamblers and
their significant others (Raylu & Oei, 2002). Furthermore, the
availability and legalisation of gambling activities in various
countries (e.g., recent legalisation of gambling in Singapore)
perpetuates the frequency and severity of gambling-related
social issues among the Chinese and has raised concerns among
practitioners and governmental authorities (Tan, Yen, & Nayga,
2010). It is also not uncommon to encounter anecdotal media
coverage on PG among Chinese individuals with speculations
of prostitution and drug-dealing to repay debts, and parental
neglect of young children stemming from gambling addiction
(Blaszczynski, Huynh, Dumlao, & Farrell, 1998). Moreover,
empirical evidence of PG among the Chinese in Australia and
Hong Kong do suggest that gambling is a popular recreational
activity and prevalence rates are higher in this population in
comparison to Western populations (Blaszczynski et al., 1998;
Chen et al., 1993; Loo et al., 2008).
It has also been argued that there are differences in the de-
velopment of PG, perpetuation of gambling problems, and
help-seeking attitudes between individuals from the West and
Chinese individuals (Loo et al., 2008; Raylu & Oei, 2004a). To
J. M. Y. LOO ET AL. 343
date, much research has been conducted on Western samples
and results obtained were commonly used to guide research and
interventions among Chinese samples. As the Chinese ethnic
group is the largest ethnic group representing 22% of the
world’s population (Tseng, Lin, & Yeh, 1995) and as gambling
is popular among this group, it is important that we specifically
investigate the patterns of PG among the Chinese and highlight
the socio-demographic variables and correlates of PG such as
gambling-related cognitions, gambling urge, depression, anxi-
ety, stress, and help-seeking correlates in order to assist in iden-
tification of PG cases for early detection and intervention. In-
vestigating correlates of help-seeking attitudes is also an im-
portant step in future development of PG intervention that will
increase positive help-seeking propensity among Chinese prob-
lem gamblers.
Different conceptual understanding and theories of PG pro-
duces diverse forms of measurement tools; therefore, creating
varied empirical findings about the prevalence of PG (McMil-
len & Wenzel, 2006). Hence, utilising valid screening tools for
Chinese PG is essential in the advancement of our understand-
ing of PG among the Chinese. The predecessors of the Cana-
dian Problem Gambling Index (CPGI) such as South Oaks
Gambling Scale (SOGS) and Diagnostic and Statistical Manual
(DSM-IV) have been used widely in gambling prevalence re-
search among the Chinese residing in Hong Kong and Macao
(Fong & Ozorio, 2005; Wong & So, 2003), and also among
studies investigating PG correlates (e.g., Liao, 2008; Oei, Lin,
& Raylu, 2008; Oei & Raylu, 2010). These studies used cut
scores validated among Western samples, which often overes-
timate the rates of pathological gambling among the Chinese
(Tang, Wu, Tang, & Yan, 2010). Hence, this study will adopt
the Chinese SOGS cut scores proposed by Tang and colleagues
(2010) as an attempt to improve the accuracy of estimating
pathological gambling in this population.
A recent comparison of four gambling measures such as
SOGS, CPGI, Gamblers Anonymous-20 (GA-20), and DSM-IV
on a sample in Singapore have found CPGI to be the most reli-
able and valid in measurement construct (Arthur et al., 2008).
As a whole the CPGI consists of 31 items measuring gambling
involvement, PG assessment, and PG correlates (Ferris &
Wynne, 2001). However, only nine items are scored, which is
collectively named the Problem Gambling Severity Index
(PGSI) under the CPGI. Although many studies uses the term
CPGI synonymously with PGSI; for clarity reasons, PGSI will
be used in this study to identify the scored 9-items scale. To
date, there is a lack of research that investigates the patterns of
relationship between PG and other correlates utilizing the PGSI.
In this study, the Chinese validated version of PGSI will be
used for the above reasons to measure PG. For the purposes of
estimating the prevalence of PG and pathological gambling,
SOGS will be used concurrently with PGSI as a comparison.
Also, PGSI will be used to examine differences in socio-
demographic variables and to investigate the predictive ability
of gambling correlates and help-seeking attitudes on PG. Being
a relatively new scale (i.e., Chinese cut scores unavailable) and
as the PGSI was originally developed to remediate the problem
of overestimation in other scales, the original cut scores (Ferris
& Wynne, 2001) will be utilised in this study.
Considering measurement issues described here, there is a
gap in literature using newer PG screening tools on the patterns
of PG and important factors concerning PG such as psycho-
logical states, socio-demographic variables, gambling-related
cognitions, gambling urges, and help-seeking attitudes. Fur-
thermore, this study will extend our current knowledge of Chi-
nese help-seeking attitudes by exploring the impact of psycho-
logical openness, help-seeking propensity, and indifference to
stigma on PG among Chinese individuals. The subsequent sec-
tions discuss these factors in detail.
Past Prevalence Estimates, Types of Gambling, and
Gender Differences
To date, the estimates of reported gambling participation
varied from 26.6% (Lai, 2006; Sin, 1997) to 92% (Clarke et al.,
2006) in Chinese samples from Canada and New Zealand re-
spectively. However, estimates of PG and pathological gam-
bling closely resembled each other. In past research, the esti-
mates of PG using SOGS ranged from 2.5% (Fong & Ozorio,
2005; Sin, 1997) to 4.0% (Wong & So, 2003). Meanwhile,
using the DSM-IV Gambling-Behaviour Index and SOGS, the
estimated percentages of pathological gambling in the pool of
empirical studies ranged from 1.78% in a sample of Macao
residents (Fong & Ozorio, 2005), 1.8% among Hong Kong
residents (Wong & So, 2003), and Canadian residents (Sin,
1997) to 2.9% in an Australian Chinese speaking sample
(Blaszczynski et al.1998). Due to the possibility of false posi-
tive cases in prevalence studies using either DSM or SOGS, it
has been suggested that more research needs to be conducted to
clarify prevalence results (Blaszczynski et al., 1998). Hence,
both SOGS using Chinese cut scores and PGSI will be meas-
ured in this study in an attempt to address the issue of
over-estimation. It is important to note also that the underre-
porting of problems among Chinese PGs are common issues
particularly among males in past research (Blaszczynski et al.,
1998) and hence will be investigated in this study by comparing
self-rated score and actual scores from screening tools.
Studies have reported that males participated more in gam-
bling than females and were at higher risk of gambling prob-
lems (Blaszczynski et al., 1998; Chen et al., 1993); however,
there is a lack of research on the preferred types of gambling
according to gender. Hence, the patterns of PG and frequency
of participation in various forms of gambling will be further
investigated in this study. The moderating effects of gender
among gambling-related correlates in predicting PG will also be
investigated. In a research among Australian Chinese individu-
als, more entertaining forms of gambling such as Lotto, Keno,
Powerball, and poker machines at the casino (in that order)
were found to be the most common gambling activities (VCGA,
2000). As the country in which a problem gambler resides in
influences legislations and the forms of gambling activities
available, it is important to examine the popular forms of gam-
bling among the Chinese and also between genders in their
respective countries. Also, there is a lack of research reporting
on the participation frequency according to different types of
gambling and the average amount of money spent in each gam-
bling type by Chinese individuals in Taiwan, which will be thus
investigated in this study. Such information on the popularity of
certain gambling activities will prove valuable in making in-
formed decisions about policy regulations, channelling of re-
sources, and intervention strategies targeting specific at-risk
type gamblers.
J. M. Y. LOO ET AL.
344
Socio-Demographic Factors, Psychological States,
Gambling-Related Cognitions, and Gambling Urge
Research evidences suggest that gambling-related cognitions
such as erroneous beliefs, expectancies, illusion of control and
perpetuating gambling thoughts play a crucial role in the de-
velopment and maintenance of gambling behaviour (Griffiths,
1994; Loo et al., 2008; Oei et al., 2008). The influences of
gambling urge have also been considered as an important factor
in the development of PG (Raylu & Oei, 2002, 2004; Sharpe,
2002). Although previous studies have discussed the role of
urges in the development of cognitions (Niaura, 2000; Skinner
& Aubin, 2010; Tiffany, 1999; Tiffany & Conklin, 2000), these
mediating effects of gambling-related cognitions have not been
empirically investigate in the literature. Hence, the mediating
process of gambling-related cognitions by which gambling
urges impacts on PG severity (i.e., gambling urges impacts on
PG via gambling-related cognitions) will be examined in this
study among Chinese individuals living in a traditionally Chi-
nese country (i.e., Taiwan).
Furthermore, in Western samples, an individual’s gambling
severity typically differ with socio-demographic factors such as
age, gender, marital status, income, education, and employment
status (Ocean & Smith, 1993). However, results are varied
across different ethnic groups and countries, and consistent
evidences are lacking among Chinese samples as compared
with Western samples. This study will further investigate the
patterns of socio-demographic factors according to the severity
of PG.
In the gambling literature, negative psychological states (e.g.,
depression, anxiety, and stress) have been found to play an
important role in the development and maintenance of PG (Loo
et al., 2008; Moodie & Finnigan, 2006a; Oei et al., 2008; Raylu
& Oei, 2002). Affective disorders often comorbid with PG
(Coman, Burrows, & Evans, 1997; El-Guebaly et al., 2006);
however, the direction of effects is often unclear, especially
among the Chinese as limited research has been conducted in
this sample. In Western samples, problem gamblers have con-
sistently reported higher rates of affective disorders, particu-
larly for depression (Becona, Lorenzo, & Fuentes, 1996;
Blaszczynski & McConaghy, 1989; Moodie & Finnigan,
2006a). Similarly, research has postulated the contribution of
anxiety and stress in the impairment of decision-making (Miu,
Heilman, & Houser, 2008), and the development, maintenance,
and relapse of problem gambling (Blaszczynski, McConaghy,
& Frankova, 1991; Raylu & Oei, 2004b). This study will ex-
amine the predictive ability of psychological states in PG with
modifications made in screening tools (described above) in a
sample of Chinese residents in Taiwan.
Help-Seeking Attitudes
It is well-documented that Chinese gamblers have difficulty
admitting that PG is an issue that has to be dealt with, and
viewing gambling as an avenue of gaining financial wealth
despite having financial difficulties (Loo et al., 2008; Papineau,
2001; Scull & Woolcock, 2005). Difficulty in admitting the
problem and seeking help are common characteristics among
the Chinese, which affects their propensity for help-seeking
behaviour and in turn increase their susceptibility to mental
health issues (Pagura, Fotti, Katz, & Sareen, 2009). Such be-
haviours are seen as a sign of weakness and vulnerability,
which produces the feared reality of losing respect in the com-
munity (Basu, 1991; GAMECS Project, 1999). Hence, this
study will empirically investigate the predictive ability of three
aspects of attitudes toward seeking mental health services—
psychological openness, help-seeking propensity, and indiffer-
ence to stigma, which will contribute to our understanding of
the antecedents of help-seeking among Chinese problem gam-
blers. Psychological openness explores an individual’s ability
in acknowledging psychological problems and willingness to
seek professional help for it if deemed necessary, while
help-seeking propensity explores the extent of willingness and
ability to seek psychological help (Mackenzie, Knox, Gekoski,
& Macaulay, 2004). Finally, indifference to stigma represents
the degree of concern for the opinions of others when they are
found to be experiencing psychological problems.
Rationale for This Study
One of the main objectives of this study is to provide a de-
tailed evaluation of similarities and differences in patterns of
PG, pathological gambling, and socio-demographic correlates.
This study will also examine the PG predictive ability of gam-
bling-related cognitions, gambling urge, negative psychological
states, and help-seeking attitudes among Chinese individuals
residing in Taiwan. As an attempt to remediate the issue of PG
over-estimation in past Chinese studies, modifications in PG
screening tools will be made (i.e., Chinese SOGS cut scores
and PGSI scale). Patterns of pathological gambling, PG, fre-
quency of gambling participation according to types of gam-
bling activities, amount spent on gambling, and gender differ-
ences will also be detailed in this study. It is hypothesized that
gambling-related cognitions, gambling urges, depression, anxi-
ety, and stress will be able to predict PG (i.e., higher scores on
these correlated will predict higher PGSI score). Furthermore, it
is predicted that gambling-related cognitions will mediate the
relationship between gambling urges and PG. Based on past
research that have consistently found gender differences in PG,
it is hypothesized that there will be a moderating effect of gen-
der in these gambling correlates, which will be investigated
using interaction effects. In help-seeking attitudes, it is hy-
pothesized that more positive help-seeking attitudes (i.e., high
psychological openness, high help-seeking propensity, and low
indifference to stigma) will predict lower PG severity. An ac-
curate understanding of the influences of these factors is im-
portant in making informed decisions among policy makers and
devising early intervention strategies that target high-risk indi-
viduals that minimises the detrimental effects of PG. Further-
more, a solid understanding of patterns of gambling among the
Chinese and the effect of gambling correlates on PG is inter-
esting in its own right but, just as important, it is fundamental
for future development of effective treatment programs among
the Chinese.
Method
Participants
The participants consisted of 801 Chinese participants from
Taiwan (i.e., Taiwanese Chinese; 52.38% were males and
47.62% were females). All 801 participants can read and write
J. M. Y. LOO ET AL. 345
in Chinese language (i.e., Mandarin). The mean age was 25.36
years (SD = 10.25) with an age range of 18 to 74 years. In rela-
tion to employment status, 69.50% of participants were stu-
dents; while 20.10% were employed full-time and 5.10% were
employed part-time. The remainder participants were either job
hunting (1.70%), under disability pension (2.50%), or retired
(1.10%). Most participants have never married (82.7%), while
15.5% are currently married, 1.1% was separated or divorced,
0.4% was widowed, and 0.3% was in a domestic partnership.
In relation to education, 57.2% of participants have had some
college education, 28% completed a Bachelor’s degree, 12.3%
had up to high school education, and 2.5% completed a Post-
graduate degree. Most participants were ancestor worshippers
(30.2%), while 23.5% were Buddhists, 22.3% had no religion,
18.4% were Taoists, 4.6% were Catholics or Christians, and 1%
believed in other religions. The majority of participants earned
less than Taiwan Dollar (TWD) 100,000 (73.8%), while 8.1%
earned between TWD 100,000 - TWD 300,000; 11.5% earned
between TWD 300,000 - TWD 700,000; 4.5% earned between
TWD 500,000 - TWD 700,000; and 2.1% earned more than
TWD 1,000,000.
Measures
As part of a larger project, the materials included a demo-
graphics form and a set of self-report questionnaires. All meas-
ures without existing validated Chinese versions were trans-
lated into Chinese from English and back-translated again (i.e.,
reverse translation) to check for consistency and face validity.
A bilingual psychologist and graduate student who underwent
both Western and Chinese education completed the translations.
The scales were also checked by two bilingual clinical psy-
chology PhD candidates who were blind to the study to ensure
accuracy of translation. Pilot tests were conducted on 10 uni-
versity students to verify the semantic integrity of each item
and to ensure ease of understanding. The preceding steps were
then repeated for highlighted items after pilot testing. Discrep-
ancies between the versions were thoroughly discussed and
resolved until a translated version was found to have semantic
equivalence with the original English version.
The Problem Gambling Severity Index (PGSI; Ferris &
Wynne, 2001) The PGSI is a 9-item measure of PG, derived
from the 31-item CPGI. Five items of PGSI originated from
SOGS, 2 items from DSM-IV, and 2 items were developed for
the PGSI. It uses a 4-point rating scale ranging from “0—Nev-
er” to “3—Almost always.” Items are totalled and the total of 0
identifies a non-gambler, 1-2 identifies a low-risk gambler, 3-7
identifies a moderate-risk gambler, and 8 or more identifies a
problem gambler. The Cronbach’s alpha was good, at 0.84,
with a test-retest reliability of 0.78 (Ferris & Wynne, 2001).
The PGSI has good criterion-related validity because it matches
up fairly well with the DSM-IV and the SOGS, correlating at
0.83 with both measures (Ferris & Wynne, 2001). The Cron-
bach’s alpha in a Chinese sample was reported to be 0.77 with
good concurrent, predictive, and discriminant validities (Loo,
Oei, & Raylu, under review).
South Oaks Gambling Screen (SOGS; Lesieur & Blume,
1987) is the most commonly used 20-item self-administered
instrument for assessing “pathological gambling,” which was
developed based on DSM-III criteria. Cronbach’s alpha was
0.97 and the test-retest reliability was 0.71 (Lesieur & Blume,
1987). This 20-item scale uses yes/no responses and scores
ranged from 0 to 20. A score of 0 indicates no pathological
gambling, 1-4 indicates at-risk gambling behaviour or possible
pathological gambling, and a score of 5 or more indicates
pathological gambling (Stinchfield, 2002). In Chinese samples,
the Cronbach’s alpha was 0.75 with good construct validity
(Blaszczynski et al., 1998).
Gambling Frequency, Amount Spent, and Self-Rated PG be-
haviour. The gambling frequency questions consisted of four
items measuring frequency of gambling and amount of money
(TWD) spent per day in (1) gaming machines, (2) table games,
(3) animals such as horse racing, and (4) other forms of gam-
bling such as bingo, lottery, and sports betting. Frequency re-
sponses were measured on a 5-point scale ranging from
“1—Never,” “2—monthly or less,” “3—2 to 4 times a month,”
“4—2 to 3 times per week,” to “5—4 or more times per week.”
Amount of money spent required open-ended responses. The
final item asked participants to indicate on a scale whether they
consider themselves to be a non-problem gambler or a problem
gambler.
The Gambling Related Cognitions Scale-Chinese Version
(GRCS-C; Oei, Lin, & Raylu, 2007a; Raylu & Oei, 2004b).
The GRCS-C is a 23-item scale measuring erroneous gambling
cognitions and included items such as “I have some control
over predicting my gambling wins.” There are five subscales in
the GRCS-C: (1) GE—Gambling expectancies, (2) IC—Illu-
sion of control, (3) PC—Predictive control, (4) IS—Inability to
stop gambling, and (5) IB—Interpretative bias. The responses
were measured on a 7-point Likert scale (1 = strongly disagree
to 7 = strongly agree) with higher scores indicating more cog-
nitive distortions held by the individual. The GRCS-C reported
a Cronbach’s alpha 0.95 and ranged from 0.83 to 0.89 for the
five factors (Oei et al., 2007a). The GRCS-C also reported good
concurrent, predictive, and discriminant validities.
The Gambling Urge Scale-Chinese Version (GUS-C; Oei,
Lin, & Raylu, 2007b; Raylu & Oei, 2004). The 6-item ques-
tionnaire measured gambling urges, which has been found to be
important in the maintenance of PG. The GUS-C included
items such as “All I want to do now is to gamble” and “Nothing
would be better than having a gamble right now.” Participants
responded on a 7-point Likert scale (1 = strongly disagree to 7
= strongly agree) with higher scores indicating a stronger urge
to gamble. The Cronbach’s alpha for the Chinese version was
reported to be 0.87 and has adequate concurrent, predictive, and
criterion validities (Oei et al., 2007b).
Depression, Anxiety, and Stress Scale (DASS; Lovibond &
Lovibond, 1995). This 21-item scale (rated on a 4-point scale
ranging from 0 “did not apply to me at all” to 3 “applied to me
very much or most of the time”) assesses symptoms of depres-
sion (DASS-D), anxiety (DASS-A), and stress (DASS-S). The
scale reported good internal consistency (α = 0.94, 0.87 and
0.91 for the subscales of depression, anxiety and stress respec-
tively) (Antony, Bieling, Cox, Enns, & Swinson, 1998). In a
Chinese sample, the Cronbach’s alpha for the three subscales
were also high (α depression = .85; α anxiety = .87; α stress
= .82) and the Chinese version showed good criterion and pre-
dictive validities (Oei et al., 2008).
Inventory of Attitudes toward Seeking Mental Health Ser-
vices (IASMHS; Mackenzie et al., 2004). The IASMHS is a
24-item scale measuring mental health help-seeking behaviour
J. M. Y. LOO ET AL.
346
and included items such as “If I were to experience psycho-
logical problems, I could get professional help if I wanted to”.
The responses were measured on a 5-point Likert scale (0 =
disagree to 4 = agree) with higher total scores indicating more
positive attitudes toward mental health help-seeking. There are
three subscales in the IASMHS: 1) psychological openness
(higher scores represent higher psychological openness), 2)
help-seeking propensity (higher scores represent stronger
help-seeking propensity), and 3) indifference to stigma (lower
scores represent indifference; higher scores represent fear of
stigma). Cronbach’s alpha for IASMHS was .87 (α = 0.82, 0.76
and 0.79 for the subscales of psychological openness,
help-seeking propensity, and indifference to stigma respec-
tively). The three factors are significantly positive correlated.
The IASMHS reported good discriminant, predictive, and con-
current validity. The Cronbach’s alphas in a Chinese sample
were reported to be 0.74 for the total scale (Atkinson, 2007).
The scale reported good criterion, predictive, and discriminant
validities.
Procedure
As a part of a larger study, university participants from Tai-
wan were recruited from universities in Southern Taiwan and
Northern Taiwan. Ethical clearances were provided by the re-
spective organisational ethics committee and all procedures
were carried out accordingly. The university participants were
recruited from these departments: 1) Nursing, 2) General Edu-
cation, 3) Mechanical Engineering, 4) Electrical Engineering, 5)
Recreation Administration, 6) Business Administration, and 7)
Medicine. The community participants were recruited from
Southern and Northern Taiwan by word-of mouth and commu-
nity contacts with companies. Voluntary participation was fol-
lowed by an introduction to the research study, explanation on
informed consent, and freedom to withdraw participation. No
personal identification information was requested and privacy
was assured. Paper and pencil questionnaires were administered
individually and participants were thanked and debriefed upon
completion. All participants were reimbursed with TWD
100.00 (i.e., approximately AUD 4.00) and average time taken
to complete the questionnaire was 30 minutes.
Results
Preliminary Data Analysis
All data cleaning and descriptive analyses were conducted
using SPSS version 17 (SPSS Inc., 1988). Data cleaning in-
cluded checking accuracy of data entry, missing values, and
assumptions of multivariate analysis. All outliers were checked
for accurate data entry and were retained as each case is from
the intended sample and is a true reflection of the data collected
from participants (Tabachnick & Fidell, 2007). There were 396
males and 360 females (45 missing data). Missing gender data
were not imputed for the same reasons. Non-systematic and
minor missing data (less than 5% missing) for all variables
were replaced using mean substitution (Tabachnick & Fidell,
2007). Four participants’ entries were removed due to more
than 40% of missing items in each entry. Visual screening of
the histogram and statistical tests indicated that there was some
univariate kurtosis and skewness. Data was positively skewed
and hence, logarithm transformation was performed (Tabach-
nick & Fidell, 2007). All analyses were conducted with both
non-transformed and transformed data—as no substantive dif-
ferences were found only the non-transformed results are re-
ported. As shown in Table 1, all scales used in this study re-
ported good Cronbach’s alpha with α ranging from 0.66 to 0.98,
reflecting good internal consistency within each measurement
scale.
A Comparison of Patt erns of Patho logical Gambling
(SOGS) and Problem Gam bling (PGSI)
As shown in Table 2 and according to the Chinese SOGS
cut-offs (Tang et al., 2010), 89.3% of participants are non-
pathological gamblers, 6.5% are at-risk pathological gamblers,
and 4.2% are pathological gamblers. Using the PGSI cut-offs
(Ferris & Wynne, 2001), 42.1% of participants are classified as
non-problem gamblers, 21.6% are low-risk gamblers, 27.5%
are moderate-risk gamblers, and 8.9% are problem gamblers
(see Table 2). The Goodness-of-Fit Chi-Square tests were used
to test whether the observed pattern of events described below
differs significantly. There is a significant difference between
the lower percentage of estimated pathological gamblers ac-
cording to SOGS (4.2%) as compared to the percentage of es-
timated problem gamblers in PGSI (8.9%), Δ χ2 (3) = 35, p
< .001. In the non-problem gambler (PGSI) group, there are
significantly more females (47.8%) than males (36.9%), Δ χ2 (1)
= 26, p < .001. The opposite was true for the non-pathological
gambler (SOGS) group as there are significantly less females (n
Table 1.
Reliability analyses for each scale and subscale.
Chinese Scale α
Problem Gambling Severity Index PGSI 0.77
South Oaks Gambling Screen (SOGS) 0.83
Gambling frequency 0.66
Gambling Cognitions (GRCS-C) 0.98
GRCS-GE 0.91
GRCS-IC 0.87
GRCS-PC 0.91
GRCS-IS 0.94
GRCS-IB 0.88
Gambling Urge (GUS-C) 0.94
DASS total 0.95
Depression (DASS-d) 0.88
Anxiety (DASS-a) 0.86
Stress (DASS-s) 0.86
Inventory of Attitudes toward Seeking
Mental Health Services (IASMHS) total 0.72
IASMHS-Psychological openness 0.71
IASMHS-Help-seeking propensity 0.82
IASMHS-Indifference to stigma 0.78
J. M. Y. LOO ET AL. 347
Table 2.
Frequency and percentage of participants in each category of SOGS and PGSI.
Scale Males Females Total
SOGS
Non-pathological gambler (SOGS = 0 to 4) 349 (88.1%) 326 (90.6%) 675 (89.3%)
At-risk pathological gambler (SOGS = 5 to 7) 29 (7.3%) 20 (5.6%) 49 (6.5%)
Pathological gambler (SOGS = 8 or more) 18 (4.5%) 14 (3.9%) 32 (4.2%)
PGSI
Non-problem gambler (PGSI=0) 146 (36.9%) 172 (47.8%) 318 (42.1%)
Low-risk problem gambler (PGSI = 1 to 2) 89 (22.5%) 74 (20.6%) 163 (21.6%)
Moderate-risk problem gambler (PGSI = 3 to 7) 114 (28.8%) 94 (26.1%) 208 (27.5%)
Problem gambler (PGSI = 8 or more) 47 (11.9%) 20 (5.6%) 67 (8.9%)
= 326) than males (n = 346), Δ χ2 (1) = 20, p < .001. There is a
general trend in the PGSI groups to have significantly more
males in all three low-risk (Δ χ2 (1) = 15, p < .001), moder-
ate-risk (Δ χ2 (1) = 20, p < .001), and problem gambling (Δ χ2
(1) = 27, p < .001) groups. In the SOGS groups, there are sig-
nificant differences between the frequency of males and fe-
males in the at-risk pathological gambler group (Δ χ2 (1) = 9, p
= .005) and the pathological gambler group (Δ χ2 (1) = 4, p
= .05). It is interesting to note that among the males, the fre-
quency of non-pathological gamblers (SOGS; n = 349) equals
the summed frequency of non-PG, low-risk, and moderate-risk
PG (PGSI; total n = 349). However, the frequencies among
females are significantly different (Δ χ2 (3) = 14, p < .005).
Also, among the males, the summed frequency of at-risk
pathological gamblers and pathological gamblers (SOGS; n =
49) equals the frequency of problem gambler (PGSI; total n =
49). However, the frequencies among females are significantly
different (Δ χ2 (3) = 14, p < .005).
In the sample, 330 males (83.3% of males) and 304 females
(84.4%) rated themselves to be non-problem gamblers. These
frequencies are significantly different when compared with the
actual PGSI scores (males Δ χ2 (3) = 184, p < .001; females Δ
χ2 (3) = 132, p < .001) and SOGS scores (males Δ χ2 (2) = 19, p
< .001; females Δ χ2 (2) = 14, p < .001). Self-rating scores also
showed that 14 males (3.5%) and 14 females (3.9%) rated
themselves to be problem gamblers. These frequencies are not
significantly different when compared with the actual SOGS
scores for males (Δ χ2 (2) = 4, p > .05, ns), while the frequen-
cies are exactly the same for females. The self-rated problem
gambler scores were significantly different from the PGSI
scores for males (Δ χ2 (3) = 33, p < .001), but not significantly
different for females (Δ χ2 (3) = 6, p > .05, ns). In sum,
self-rated PG was similar to actual PGSI and SOGS scores
respectively with the exception of PGSI scores for males (i.e.,
according to PGSI scores, males tended to significantly under-
estimate their problem gambling). Meanwhile, non-problem
gambling was significantly overestimated by the participants
when compared with both SOGS and PGSI scores.
The frequency, percentages, and average amount spent on
each type of gambling are illustrated in Table 3. The rate of
gambling varies according to gender and type of gambling.
Table games are the most popular gambling activity, followed
by other forms of gambling such as lottery and sports betting,
then gaming machines, and the least popular was animals gam-
bling. However, the highest amount of money was spent on
gaming machines (Δ χ2 (3) = 1667.52, p < .001) although it was
not the most popular type of gambling activity, as followed by
other forms of gambling such as lottery (Δ χ2 (3) = 126922.80,
p < .001), table games (Δ χ2 (3) = 424.65, p < .001). The least
amount of money was spent on animals gambling. Males gam-
bled significantly more frequently than females on table games
(Δ χ2 (1) = 4.00, p < .05) with the exception of significantly
more females reported having gambled monthly or less on table
games (Δ χ2 (1) = 15.00, p < .001) and other forms of gambling
such as lottery (Δ χ2 (1) = 7.00, p < .01).
Socio-Demographic Variables and Problem Gambling
as Measured by PG SI
Results were analysed to examine differences between PGSI
groups (i.e., non-problem gambler, low-risk PG, moderate-risk
PG, and PG) in socio-demographic variables such as age, gen-
der, marital status, education, employment status, annual in-
come, and religion. A One-way ANOVA revealed that there
was a statistically significant difference in age according to
PGSI cut-off groups, F(3,793) = 7.55, p < .0001, partial ή2 =
0.028. Employing the Bonferroni post-hoc test, significant dif-
ferences in age were found between non-problem gamblers (M
age = 26.75, SD = 11.27) and moderate-risk gamblers (M =
23.10, SD = 7.15, p < .0001); and between moderate-risk gam-
blers (M = 23.10, SD = 7.15) and problem gamblers (M = 27.75,
SD = 12.82, p = .005).
The Multi-Dimensional Chi-Square test was used to examine
characteristics of problem gamblers in various socio-demo-
graphic variables described below. There is a significant rela-
tionship between PG behaviour and gender, χ²(3, N = 756) =
14.63, p = .002. Males (N = 47, 70.1%) are more likely to be
problem gamblers as compared to females (N = 20, 29.9%). It
is a small association (Phi, φ = 0.139, p = .002) and thus gender
accounted for 1.9% of the variance in PGSI score. Never mar-
ried (N = 55, 76.4%) individuals are more likely to be problem
amblers as compared to married (N = 13, 18.1%) individuals, g
J. M. Y. LOO ET AL.
348
Table 3.
Frequency and percentage of participants according to gambling types.
Gambling type and frequency Males Females Total
Gaming machines (e.g., pokies)a
Never 308 (78.2%) 280 (78.4%) 588 (78.3%)
Monthly or less 77 (19.5%) 71 (19.9%) 148 (19.7%)
2 - 4 times a month 5 (1.3%) 2 (0.6%) 7 (0.9%)
2 - 3 times per week 0 (0%) 2 (0.6%) 2 (0.3%)
4 or more times per week 4 (1.0%) 2 (0.6%) 6 (0.8%)
Table games (e.g., cards)b
Never 196 (49.9%) 156 (43.6%) 352 (46.9%)
Monthly or less 157 (39.9%) 172 (48.0%) 329 (43.8%)
2 - 4 times a month 27 (6.9%) 22 (6.1%) 49 (6.5%)
2 - 3 times per week 5 (1.3%) 4 (1.1%) 9 (1.2%)
4 or more times per week 8 (2.0%) 4 (1.1%) 12 (1.6%)
Animals (e.g., horse bet)c
Never 378 (95.9%) 346 (96.9%) 724 (96.4%)
Monthly or less 13 (3.3%) 8 (2.2%) 21 (2.8%)
2 - 4 times a month 1 (0.3%) 0 (0%) 1 (0.1%)
2 - 3 times per week 0 (0%) 3 (0.8%) 3 (0.4%)
4 or more times per week 2 (0.5%) 0 (0%) 2 (0.3%)
Other forms (e.g., lottery)d
Never 231 (58.6%) 207 (57.8%) 438 (58.2%)
Monthly or less 121 (30.7%) 128 (35.8%) 249 (33.1%)
2-4 times a month 28 (7.1%) 15 (4.2%) 43 (5.7%)
2-3 times per week 9 (2.3%) 5 (1.4%) 14 (1.9%)
4 or more times per week 5 (1.3%) 3 (0.8%) 8 (1.1%)
aAverage money spent on gaming machines = TWD 143, 171.05; bAverage money spent on table games = TWD 14, 580.73; cAverage money spent on animals = TWD 14,
156.08; dAverage money spent on other forms of gambling = TWD 141, 503.53.
χ²(15, N = 796) = 28.22, p = .02. It is a small association (Phi, φ
= 0.188, p = .02) and thus marital status accounted for 3.5% of
the variance in PGSI score. There is no significant relationship
between PG behaviour and education, χ²(18, N = 793) = 23.98,
p = .156, ns. Analysis revealed that there is a significant asso-
ciation between PG behaviour and employment status, χ²(18, N
= 792) = 33.17, p = .016. Students (N = 47, 66.2%) are more
likely to be classified as problem gamblers as compared to in-
dividuals in full-time employment (N = 15, 21.1%). It is a small
association (Phi, φ = 0.205, p = .016) and thus employment
status accounted for 4.2% of the variance in PGSI score. There
is no significant association between PG behaviour and annual
income, χ²(18, N = 794) = 27.54, p = .069, ns. Similarly, no
significant relationship was reported between PG and religion,
χ²(24, N = 728) = 31.00, p = .154, ns. In sum, there are signifi-
cant associations between PG and age, gender, marital status,
and employment status respectively.
Predictor Variables of Problem Gambling Behaviour
Examination of the linear relationships between socio-de-
mographic variables (i.e., age, gender, marital status, employ-
ment) and PG showed significant correlations between PG and
gender; hence, gender will be controlled for in the analysis. All
variables except the help-seeking attitudes subscales showed
significant positive correlations with PGSI. However, IASMHS
total and Indifference to stigma showed significant negative
correlation with PGSI (i.e., positive help-seeking attitude pre-
dicts lower PGSI scores and stronger indifference to stigma
J. M. Y. LOO ET AL. 349
predicts higher PGSI scores, respectively).
Preliminary steps were taken in all analyses to check for ad-
herence to assumptions. To reduce problems associated with
multicollinearity, all independent variables and moderator var-
iables were centred (i.e., standardised) (Frazier, Tix, & Barron,
2004; Jaccard, Wan, & Turrisi, 1990). Centred variables were
created by subtracting the mean value from the variable while
the interaction variable was created by multiplying the two
mean centred independent and moderator variables together
(Jaccard et al., 1990). A moderation effect was considered evi-
dent only when the interaction term (e.g., gender x GRCS total)
in the regression was significant. With the use of a p < .001
criterion (i.e., values larger than 24.322, df = 7) for Mahalano-
bis distance (Tabachnick & Fidell, 2007), eight outliers were
identified and deleted from dataset. Using the variation infla-
tion factor (VIF), multi-collinearity was checked and all vari-
ables reported values below 10 (Field, 2000), which indicate
that the data did not violate the assumption of multi-collinea-
rity.
First, Hierarchical Multiple Regression (HMR) was used to
assess the extent to which these variables could predict PG.
PGSI total score was used as the dependent variable (DV). The
independent variables (IV) and interaction variables were:
(Step 1) Gender, (Step 2) GRCS-total, GUS-total, DASS-total,
and IASMHS-total, and (Step 3) Two-way interactions between
gender and total scores. Table 5 displays the unstandardised
regression coefficients (B), standard error of B, the standardised
regression coefficients (β), R2 change, R, R2, and Adjusted R2
after entry of all independent variables. R was significantly
different from zero at each step. HMR results showed that the
model was significant F (9, 738) = 19.34, p < .001. The R2
value of 0.19 indicates that 19% of the variability in PGSI
scores is accounted for by the predictors. Only these variables
significantly predicted and accounted for variance in PGSI
scores: 1) gender (accounted for 1.8% of variance); 2) GRCS
total (17%); and 3) DASS total (3.3%).
Predictor Variables of Problem Gambling Behaviour
Examination of the linear relationships between socio-de-
mographic variables (i.e., age, gender, marital status, employ-
ment) and PG showed significant correlations between PG and
gender; hence, gender will be controlled for in the analysis.
Table 4 displays the correlations between the variables. All
variables except the help-seeking attitudes subscales showed
significant positive correlations with PGSI. However, IASMHS
total and Indifference to stigma showed significant negative
correlation with PGSI (i.e., positive help-seeking attitude pre-
dicts lower PGSI scores and stronger indifference to stigma
predicts higher PGSI scores, respectively).
Preliminary steps were taken in all analyses to check for ad-
herence to assumptions. To reduce problems associated with
multicollinearity, all independent variables and moderator var-
iables were centred (i.e., standardised) (Frazier, Tix, & Barron,
2004; Jaccard, Wan, & Turrisi, 1990). Centred variables were
created by subtracting the mean value from the variable while
the interaction variable was created by multiplying the two
mean centred independent and moderator variables together
(Jaccard et al., 1990). A moderation effect was considered evi-
dent only when the interaction term (e.g., gender x GRCS total)
in the regression was significant. With the use of a p < .001
criterion (i.e., values larger than 24.322, df = 7) for Mahalano-
bis distance (Tabachnick & Fidell, 2007), eight outliers were
identified and deleted from dataset. Using the variation infla-
tion factor (VIF), multi-collinearity was checked and all vari-
ables reported values below 10 (Field, 2000), which indicate
that the data did not violate the assumption of multi-collinea-
rity.
Table 4.
Results of the hierarchical multiple regression (HMR) assessing gambling correlates, help-seeking attitudes, and interactions with gender (Total
scores).
DV IV R² B SE B Beta
PGSI total
Step1 Gender .019** .706 .247 .096*
Step2 GRCS-total .171** .050 .007 .350**
GUS-total -- .028 .028 .051
DASS-total -- .012 .005 .085*
IASMHS-total -- .006 .013 .016
Step 3 Gender × GRCS-total .001 .000 .007 .002
Gender × GUS-total -- .003 .028 .005
Gender × DASS-total -- .004 .005 .026
Gender × IASMHS-total -- .009 .013 .024
F (9, 738) = 19.34** R = 0.44**
Adjusted R2 = 0.18** R2= 0.19**
*
p < 0.01. **p < 0.001.
J. M. Y. LOO ET AL.
350
First, Hierarchical Multiple Regression (HMR) was used to
assess the extent to which these variables could predict PG.
PGSI total score was used as the dependent variable (DV). The
independent variables (IV) and interaction variables were:
(Step 1) Gender, (Step 2) GRCS-total, GUS-total, DASS-total,
and IASMHS-total, and (Step 3) Two-way interactions between
gender and total scores. Table 4 displays the unstandardised
regression coefficients (B), standard error of B, the standardised
regression coefficients (β), R2 change, R, R2, and Adjusted R2
after entry of all independent variables. R was significantly
different from zero at each step. HMR results showed that the
model was significant F (9, 738) = 19.34, p < .001. The R2
value of 0.19 indicates that 19% of the variability in PGSI
scores is accounted for by the predictors. Only these variables
significantly predicted and accounted for variance in PGSI
scores: 1) gender (accounted for 1.8% of variance); 2) GRCS
total (17%); and 3) DASS total (3.3%).
Second, another HMR analysis was performed to assess the
extent to which the subscales could predict PG. PGSI total
score was used as the dependent variable (DV). The independ-
ent variables (IV) and interaction variables were: (Step 1)
Gender, (Step 2) GRCS-GE, GRCS-IC, GRCS-PC, GRCS-IS,
GRCS-IB, GUS, DASS-D, DASS-A, DASS-S, Psychological
openness, and Indifference to stigma, and (Step 3) Two-way
interactions between gender and subscale scores. Table 5 dis-
plays the unstandardised regression coefficients (B), standard
error of B, the standardised regression coefficients (β), R2
change, R, R2, and Adjusted R2 after entry of all independent
Table 5.
Results of the hierarchical multiple regression (HMR) assessing gambling correlates, help-seeking attitudes, and interactions with gender (Subscales
scores).
DV IV R² B SE B Beta
PGSI total
Step1 Gender .019** .412 .127 .112**
GRCS-GE .196** .076 .064 .100
GRCS-IC -- .062 .051 .084
GRCS-PC -- .011 .045 .022
GRCS-IS -- .084 .049 .128
GRCS-IB -- .222 .069 .309**
DASS-D -- .077 .026 .194**
DASS-A -- .095 .029 .227**
DASS-S -- .015 .025 .039
Psychological openness -- .002 .025 .003
Step2
Indifference to stigma -- .016 .025 -.024
Gender × GRCS-GE .007 .014 .033 .101
Gender × GRCS-IC -- .018 .063 .025
Gender × GRCS-PC -- .035 .058 .070
Gender × GRCS-IS -- .096 .059 .145
Gender × GRCS-IB -- .056 .075 .078
Gender × DASS-D -- .010 .026 .026
Gender × DASS-A -- .022 .029 .053
Gender × DASS-S -- .026 .025 .067
Gender × Psychological openness -- .012 .025 .017
Step 3
Gender × Indifference to stigma -- .008 .025 .013
F (21, 726) = 9.83** R = 0.47**
Adjusted R2 = 0.20** R2 = 0.22**
*
p < 0.01. **p < 0.001.
J. M. Y. LOO ET AL. 351
variables. R was significantly different from zero at each step.
HMR results showed that the model was significant F (21, 726)
= 9.83, p < .001. The R2 value of 0.22 indicates that 22% of the
variability in PGSI scores is accounted for by the predictors.
Only these variables significantly predicted and accounted for
variance in PGSI scores: 1) gender (accounted for 1.8% of
variance); 2) GRCS-IB (Interpretative bias; 17.47%); 3)
DASS-D (Depression; 1.96%); and 4) DASS-A (Anxiety;
4.28%).
Examining W hether GRCS Mediates the Impact of
GUS on PG Severity
To test for mediation, it was established that GUS was asso-
ciated with the mediator (GRCS), r = 0.73, p < 0.01. First, a
HMR was conducted with GUS (IV) and gender (control vari-
able) as the predictor variables and GRCS total as the DV. The
IV (i.e., GUS) and gender accounted for significant variance in
the mediator (GRCS), R2= 0.52, F (2, 745) = 399.91, p < 0.001,
and the coefficients for GUS was significant, Beta = 0.72, p <
0.001. Second, another HMR was conducted with gender and
GUS (IV) in Step 1 and GRCS (mediator) in Step 2, and PGSI
scores as the DV. Table 6 reports the output from this HMR
analysis. GUS and the control variable (gender) accounted for
significant variance in PGSI, R2= 0.12, F (2, 745) = 52.44, p <
0.001, and the coefficient for GUS was significant, Beta = 0.33,
p < 0.001. In Step 2, the mediator (GRCS) added significantly
to the variance accounted for in PGSI scores, R2 change = 0.06,
F (1, 744) = 2.25, p < 0.001. The coefficient for the mediator
was significant, Beta = 0.35, p < 0.001. When the mediator was
entered in Step 2, the coefficient for the IV (GUS) decreased to
a non-significant Beta = 0.07, p = 0.134, ns; which indicates
that GRCS fully mediates the GUS-PGSI relationship. As illus-
trated in Figure 1, there was a significant indirect effect of GUS
(IV) via GRCS (mediator) on PGSI scores (DV), Sobel’s z =
7.05, p < 0.001.
Discussion
The current study examined the patterns of PG, socio-de-
mographic correlates, PG correlates, and help-seeking attitudes
among Chinese individuals residing in Taiwan with modifica-
tions in PG screening tools in an attempt to remediate the issue
of PG over-estimation in past Chinese studies (i.e., Chinese
SOGS cut scores and PGSI scale). The overall rates of both PG
(measured with PGSI) and pathological gambling (measured
with SOGS using Chinese cut scores) are higher in this Tai-
wanese Chinese sample as compared to participation rates re-
ported in previous prevalence research among Macao residents
(Fong & Ozorio, 2005), Hong Kong residents (Wong & So,
2003), Canadian residents (Sin, 1997), and Australian Chinese
speaking sample (Blaszczynski et al.1998; Oei et al., 2008; Oei
& Raylu, 2010). One plausible explanation for higher PG and
pathological gambling rates (despite using Chinese SOGS cut
scores) is the greater social acceptability and newly increased
accessibility of gambling venues in Taiwan’s outlying Penghu
islands since its recent legalization in early 2009.
Partially supporting past research that have found un-
der-reporting of PG to be a common issue among the Chinese
(Blaszczynski et al., 1998), the results showed that self-rated
PG was similar to actual PGSI and SOGS scores respectively
with the exception of PGSI scores for males (i.e., according to
PGSI scores, males tended to significantly underestimate their
problem gambling). However, self-rated non-problem gambling
was significantly overestimated by the participants when com-
pared with both SOGS and PGSI scores. The lower percentage
of self-reported male problem gamblers may reflect reluctance
to admit personal failure and to “save face,” which is a phe-
nomenon highly common among the Chinese particularly
among men (Loo et al., 2008). Slightly varied from past re-
βa = 0.719*
βc = 0.072, ns
βd = 0.326*
β
b
= 0.354**
Figure 1.
Examining the full mediation of GRCS total on the relationship between
GUS total and PGSI scores. **p < 0.001. βa = beta coefficient of the IV
predicting the mediator (with all controls in the equation). βb = beta for
the mediator predicting DV with IV and controls in the equation. βc =
beta for the IV when the mediator and controls are in the equation. βd =
beta for the IV when the controls are in the equation but the mediator
has not been entered.
Table 6.
Results of the HMR using gender, GUS, and GRCS to predict PGSI.
DV IV R2 B SE Β Beta
PGSI total
Step1 Gender .123** .346 .123 .094*
GUS-total -- .040 .027 .072
Step 2 GRCS-total .060** .051 .007 .354**
F (3, 744) = 55.88** R = 0.43**
Adjusted R2 = 0.18** R2 = 0.18**
*p < 0.005. **p < 0.001.
J. M. Y. LOO ET AL.
352
search on the choice of gambling activity (VCGA, 2000), table
games are the most popular gambling activity among partici-
pants in this study, followed by other forms of gambling such
as lottery and sports betting, then gaming machines, and the
least popular is animals gambling. However, the largest sum of
money was spent on gaming machines, as followed by other
forms of gambling such as lottery although it was not the most
popular type of gambling activity. The least money was spent
on animals gambling and table games. Males gambled more
frequently than females on all forms of gambling.
When comparing participants across PGSI groups (i.e.,
non-PG, low-risk, moderate-risk, and PG), significant differ-
ences were found between PG groups in age, gender, marital
status, and employment status respectively. Non-PG and PGs
are significantly older than moderate-risk gamblers. These ef-
fects, however, are diminished when explored using HMR as
the age difference in PG is not a linear association. There are
significant gender differences with males reporting significantly
higher PGSI score than females, and more females than males
are categorised as non-problem gamblers. This finding is con-
sistent with previous studies on Chinese communities where
males reported higher PG rates as compared to females
(Blaszczynski et al., 1998; Chen et al., 1993; Oei et al., 2008;
Oei & Raylu, 2010). Never married individuals are more likely
to be problem gamblers as compared to married individuals.
Students are more likely to be classified as problem gamblers as
compared to individuals in full-time employment. However,
these findings on marital and employment status should be
interpreted with caution as there were generally more never
married than married participants, and more students than em-
ployed participants in this study.
The results showed significant positive relationships between
PG and factors such as gambling-related cognitions, gambling
urge, depression, anxiety, and stress—all of which provided
good support for past research in these areas (Oei et al., 2007a,
2007b; Petry, 2005; Raylu & Oei, 2002; Sharpe, 2002). In other
words, problem gamblers exhibited significantly higher levels
of erroneous gambling-related cognitions, gambling urge, de-
pression, anxiety, and stress respectively, as compared to
non-problem gamblers. It was hypothesized that gambling-
related cognitions, gambling urges, depression, anxiety, and
stress will be able to predict PG (i.e., higher scores on these
correlated will predict higher PGSI score). The results indicated
that GRCS and DASS total scores significantly predicted PG
severity (i.e., PGSI scores) in this Taiwanese Chinese sample.
These findings provide support for past research that have
found that erroneous gambling-related cognitions (Griffiths,
1994; Moodie & Finnigan, 2006b; Oei et al., 2008) and nega-
tive psychological states such as depression, anxiety, and stress
(Loo et al., 2008; Moodie & Finnigan, 2006a; Oei et al., 2008)
play an important role in the development of PG.
The HMR analyses on the subscales revealed that gender
(control variable; i.e., being male), interpretative bias (GRCS-
IB), depression, and anxiety were significant predictors of PG.
The predictive ability of GRCS-IB partially support findings
obtained in past research (Oei et al., 2007a; Raylu & Oei,
2004b). As suggested by the scale developers (Oei et al., 2007a;
Raylu & Oei, 2004b), it is recommended that the total GRCS
score be used instead of the subscales to predict PG. Analyses
of the DASS subscales showed that only depression and anxiety
significantly predicted PG. This may be due to the high in-
ter-correlation between DASS subscales and the overlaps in
predictive variance on PG. Also, the participants in this study
may not exhibit symptoms of depression, anxiety, and stress
levels that are high enough to detect a significant predictive
relationship with PG, as compared to participants from a clini-
cal population (Oei et al., 2008).
Contrary to prediction, non-significant gender interaction ef-
fects for all predictor variables entered in the HMR model re-
vealed that gender does not moderate the effects of these vari-
ables on PG severity. These results partially support past re-
search (Blaszczynski et al., 1998; Chen et al., 1993; Ocean &
Smith, 1993), as gender acts as a significant predictor of PG but
not a significant moderator of the relationships between other
predictor variables and PG. In other words, being male or fe-
male does not change the relationship between these predictor
variables and PG. One possible explanation is that both males
and females experience these gambling correlates in similar
processes and in turn exhibit PG outcomes independent of their
gender.
Gambling urge has been considered in past research to in-
fluence PG (Raylu & Oei, 2002; Sharpe, 2002). Although there
was a significant moderate correlation between GUS total score
and PGSI scores in this study, gambling urge did not signifi-
cantly predict PG severity. It was a possibility due to cognitive
models of addiction (Skinner & Aubin, 2010; Tiffany, 1999;
Tiffany & Conklin, 2000) and the strong correlation (more than
r = 0.30; Baron & Kenny, 1986) between GRCS and GUS that
gambling-related cognitions acts as a mediator of the relation-
ship between gambling urge and PG (i.e., GUS predicts PG
because GUS predicts GRCS, which predicts PG). Mediation
analyses revealed that gambling-related cognitions significantly
predicted gambling urge and the path from GUS to PGSI is
reduced in absolute size when the mediator is controlled for in
the analysis. There is a significant full mediation of gam-
bling-related cognitions on the relationship between gambling
urge and PG severity. These results suggest that GUS predicts
PG because GUS predicts GRCS, which then predicts PG.
Significant negative relationships were found between PG
and overall help-seeking attitudes and the two subscales—
psychological openness and indifference to stigma; however, no
significant relationships were found between PG and help-
seeking propensity. Results on overall help-seeking attitudes
suggest that stronger engagement in help-seeking behaviour is
related to lower PG severity as reflected in lower PGSI score.
On the flipside, stronger indifference to stigma (i.e., lower
score on the indifference to stigma subscale; see Methods sec-
tion for details) relates to stronger PG severity (i.e., higher
PGSI score). In other words, higher PG severity relates to a
stronger indifference to stigma and less positive overall help-
seeking attitudes, which links in with the notion that Chinese
gamblers find it difficult to seek help for PG issues (Loo et al.,
2008; Papineau, 2001; Scull & Woolcock, 2005). Stronger
indifference to stigma is related to the Chinese problem gam-
blers’ lack of help-seeking propensity. Over time, PG may no
longer be viewed as a problem or stigma and considered as a
way of life. Hence, with this knowledge, it is important to in-
crease awareness among at-risk gamblers of the detrimental PG
outcomes on the gambler and significant others. Consequently,
as awareness of stigma increases, it is likely that help-seeking
J. M. Y. LOO ET AL. 353
behaviour will also increase. All these correlational inferences
must be viewed in light of the results obtained from the predic-
tive analyses of HMR. It was hypothesized that more positive
help-seeking attitudes (i.e., high psychological openness, high
help-seeking propensity, and low indifference to stigma) will
predict lower PG issues. Findings from the analyses suggested
that help-seeking attitudes did not predict PG severity. Al-
though help-seeking does not predict PG, it is a possibility that
given a clinical sample, help-seeking attitudes may differ be-
tween non-problem gamblers and problem gamblers.
All findings in this study should be interpreted in light of the
limitations of this study. The participants were recruited using
convenient sampling method as opposed to random sampling
(e.g., using census data) where every member of the population
has an equal opportunity of being selected. For obvious reasons,
such research will require national collaborative effort and sig-
nificant funding. Hence, the current study provided a good
descriptive and inferential analysis of patterns of PG among the
Chinese despite the detail that the participants reported here
may not be an accurate representation of the respective general
population. As with all survey research, we relied on self-re-
ported PG involvement, which is dependent on demand charac-
teristics and recall bias. Cross-sectional data was used in this
study instead of longitudinal data and therefore did not assess
the validity of results as time progresses. It will be interesting
to examine third-party estimates of PG and simulating an ex-
perimental gambling test that will accurately measure actual
gambling behaviour while manipulating variables such as erro-
neous cognitions and indifference to stigma.
The findings of this study have important implications in the
understanding and treatment of PG among the Chinese. Gam-
bling-related cognitions and negative psychological states re-
ported above are important factors that should be addressed by
mental health professionals in preventative programs among
Chinese individuals. Results reported here provided support for
findings reported in past research and strengthens the cogni-
tive-behavioural perspectives of PG. Countries considering
gambling legalisation should provide sufficient preventative
and treatment support as governmental agencies have to be
prepared for the increase in PG and social issues that will in-
evitably follow increases in availability of gambling venues in
Asia and beyond.
References
American Psychiatric Association. (2000). Diagnostic and statistical
manual of mental disorders: DSM-IV-TR (4th, ed.). Washington, DC:
American Psychiatric Association.
Antony, M. M., Bieling, P. J., Cox, B. J., Enns, M. W., & Swinson, R.
P. (1998). Psychometric properties of the 42-item and 21-item ver-
sions of the Depression Anxiety Stress Scales in clinical groups and a
community sample. Psychological Assessment, 10, 176-181.
doi:10.1037/1040-3590.10.2.176
Arthur, D., Tong, W. L., Chen, C. P., Hing, A. Y., Sagara-Rosemeyer,
M., Kua, E. H., & Ignacio, J. (2008). The validity and reliability of
four measures of gambling behaviour in a sample of Singapore uni-
versity students. Journal of Gambling Studies, 24, 451-462.
doi:10.1007/s10899-008-9103-y
Atkinson, N. W. (2007). Chinese and North American college students’
attitudes toward seeking professional psychological help: Gender
and ethnic comparisons. Master’s Thesis, Arcata, California: Hum-
boldt State University,.
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable
distinction in social psychological research: Conceptual, strategic,
and statistical considerations. Journal of Personality and Social Psy-
chology, 51, 1173-1182. doi:10.1037/0022-3514.51.6.1173
Basu, E. O. (1991). Profit, loss, and fate: The entrepreneurial ethic and
the practice of gambling in an overseas Chinese community. Modern
China, 17, 227. doi:10.1177/009770049101700203
Becona, E., Lorenzo, M. D. C., & Fuentes, M. J. (1996). Pathological
gambling and depression. Psychological Reports, 78, 635-640.
Blaszczynski, A., Huynh, S., Dumlao, V. J., & Farrell, E. (1998). Prob-
lem gambling within a Chinese speaking community. Journal of
Gambling Studies, 14, 359-380. doi:10.1023/A:1023073026236
Blaszczynski, A., & McConaghy, N. (1989). Anxiety and/or depression
in the pathogenesis of addictive gambling. International Journal of
the Addictions, 24, 337-350.
Blaszczynski, A., McConaghy, N., & Frankova, A. (1991). A compari-
son of relapsed and non-relapsed abstinent pathological gamblers fol-
lowing behavioural treatment. British Journal of Addiction, 86,
1485-1489. doi:10.1111/j.1360-0443.1991.tb01734.x
Brooker, I. S., Clara, I. P., & Cox, B. J. (2009). The Canadian Problem
Gambling Index: Factor structure and associations with psychopa-
thology in a nationally representative sample. Canadian Journal of
Behavioural Science, 41, 109-114. doi:10.1037/a0014841
Chen, C. N., Wong, J., Lee, N., Chan-Ho, M. W., Lau, J. T. F., & Fung,
M. (1993). The Shatin community mental health survey in Hong
Kong: II. Major findings. Archives of General Psychiatry, 50,
125-133.
Clarke, D., Abbott, M., Tse, S., Townsend, S., Kingi, P., & Manaia, W.
(2006). Gender, age, ethnic and occupational associations with
pathological gambling in a New Zealand urban sample. New Zealand
Journal of Psychology, 35, 84-91.
Clarke, D., Tse, S., Abbott, M., Townsend, S., Kingi, P., & Manaia, W.
(2006). Key indicators of the transition from social to problem gam-
bling. International Journal of Mental Health and Addiction, 4,
247-264. doi:10.1007/s11469-006-9024-x
Coman, G. J., Burrows, G. D., & Evans, B. J. (1997). Stress and anxiety
as factors in the onset of problem gambling: Implications for treat-
ment. Stress Medicine, 13, 235-244.
doi:10.1002/(SICI)1099-1700(199710)13:4<235::AID-SMI748>3.0.
CO;2-4
El-Guebaly, N., Patten, S. B., Currie, S., Williams, J. V. A., Beck, C. A.,
Maxwell, C. J., & Li Wang, J. (2006). Epidemiological associations
between gambling behavior, substance use & mood and anxiety dis-
orders. Journal of Gambling Studies, 22, 275-287.
doi:10.1007/s10899-006-9016-6
Ferris, J., & Wynne, H. (2001). The Canadian Problem Gambling
Index: Final report. Ottawa: Canadian Centre on Substance Abuse.
Field, A. (2000). Discovering statistics using SPSS (3rd ed.). London:
Sage.
Fong, D. K., & Ozorio, B. (2005). Gambling participation and preva-
lence estimates of pathological gambling in a far-east city: Macao.
UNLV Gaming Research & Review Journal, 9, 15-28.
Frazier, P. A., Tix, A. P., & Barron, K. E. (2004). Testing moderator
and mediator effects in counseling psychology research. Journal of
Counseling Psychology, 51, 115-134.
doi:10.1037/0022-0167.51.1.115
GAMECS Project. (1999). Gambling among members of ethnic com-
munities in Sydney: Report on “Problem gambling and ethnic com-
munities” (Part 3). Sydney: Ethnic Communities’ Council of NSW.
Griffiths, M. D. (1994). The role of cognitive bias and skill in fruit
machine gambling. British Journal of Psychology, 85, 351-369.
doi:10.1111/j.2044-8295.1994.tb02529.x
Hobson, J. S. P. (1995). Macao: Gambling on its future. Tourism Man-
agement, 16, 237-243. doi:10.1016/0261-5177(95)91465-T
Jaccard, J., Wan, C. K., & Turrisi, R. (1990). The detection and inter-
pretation of interaction effects between continuous variables in mul-
tiple regression. Multivariate Behavioral Research, 25, 467-478.
doi:10.1207/s15327906mbr2504_4
Lai, D. W. L. (2006). Gambling and the older Chinese in Canada.
J. M. Y. LOO ET AL.
354
Journal of Gambling Studies, 22, 121-141.
doi:10.1007/s10899-005-9006-0
Lesieur, H. R., & Blume, S. B. (1987). The South Oaks Gambling
Screen (SOGS): A new instrument for the identification of patho-
logical gamblers. American Journal of Psychiatry, 144, 1184-1188.
Liao, M. S. (2008). Intimate partner violence within the Chinese com-
munity in San Francisco: Problem gambling as a risk factor. Journal
of Family Violence, 23, 671-678.
doi:10.1007/s10896-008-9190-7
Loo, J. M. Y., Oei, T. P. S., & Raylu, N. (under review). Psychometric
evaluation of the Problem Gambling Severity Index (PGSI-C). Jour-
nal of Gambling Studies.
Loo, J. M. Y., Raylu, N., & Oei, T. P. S. (2008). Gambling among the
Chinese: A comprehensive review. Clinical Psychology Review, 28,
1152-1166. doi:10.1016/j.cpr.2008.04.001
Lovibond, P. F., & Lovibond, S. H. (1995). The structure of negative
emotional states: Comparison of the Depression Anxiety Stress
Scales (DASS) with the beck depression and anxiety inventories.
Behaviour Research and Therapy, 33, 335-343.
doi:10.1016/0005-7967(94)00075-U
Mackenzie, C. S., Knox, V. J., Gekoski, W. L., & Macaulay, H. L.
(2004). An adaptation and extension of the attitudes toward seeking
professional psychological help scale. Journal of Applied Social Psy-
chology, 34, 2410-2435. doi:10.1111/j.1559-1816.2004.tb01984.x
McMillen, J., & Wenzel, M. (2006). Measuring problem gambling:
Assessment of three prevalence screens. International Gambling
Studies, 6, 147-174. doi:10.1080/14459790600927845
Miu, A. C., Heilman, R. M., & Houser, D. (2008). Anxiety impairs
decision-making: Psychophysiological evidence from an Iowa Gam-
bling Task. Biological Psychology, 77, 353-358.
doi:10.1016/j.biopsycho.2007.11.010
Moodie, C., & Finnigan, F. (2006a). Association of pathological gam-
bling with depression in Scotland. Psychological Reports, 99,
407-417. doi:10.2466/pr0.99.2.407-417
Moodie, C., & Finnigan, F. (2006b). Prevalence and correlates of youth
gambling in Scotland. Addiction Research & Theory, 14, 365-385.
doi:10.1080/16066350500498015
National Policy Toward Gambling. (1974). Gambling in perspective: A
review of the written history of gambling and an assessment of its ef-
fect on modern American society. Washington, DC: National Tech-
nical Information Service.
Niaura, R. (2000). Cognitive social learning and related perspectives on
drug craving. Addiction, 95, 155-163.
doi:10.1046/j.1360-0443.95.8s2.4.x
Ocean, G., & Smith, G. J. (1993). Social reward, conflict, and com-
mitment: A theoretical model of gambling behavior. Journal of
Gambling Studies, 9, 321-339. doi:10.1007/BF01014625
Oei, T. P. S., Lin, J., & Raylu, N. (2007a). Validation of the Chinese
version of the Gambling Related Cognitions Scale (GRCS-C). Jour-
nal of Gambling Studies, 23, 309-322.
doi:10.1007/s10899-006-9040-6
Oei, T. P. S., Lin, J., & Raylu, N. (2007b). Validation of the Chinese
version of the Gambling Urges Scale (GUS-C). International Gam-
bling Studies, 7, 101-111. doi:10.1080/14459790601157970
Oei, T. P. S., Lin, J., & Raylu, N. (2008). The relationship between
gambling cognitions, psychological states, and gambling: A
cross-cultural study of Chinese and Caucasians in Australia. Journal
of Cross-Cultural Psychology, 39, 147-161.
doi:10.1177/0022022107312587
Oei, T. P. S., & Raylu, N. (2010). Gambling behaviours and motiva-
tions: A cross-cultural study of Chinese and Caucasians in Australia.
International Journal of Social Psychiatry, 56, 23-34.
doi:10.1177/0020764008095692
Pagura, J., Fotti, S., Katz, L. Y., & Sareen, J. (2009). Help seeking and
perceived need for mental health care among individuals in Canada
with suicidal behaviors. Psychiatric Services, 60, 943-949.
doi:10.1176/appi.ps.60.7.943
Papineau, E. (2001). Pathological gambling in the Chinese community,
an anthropological viewpoint. Loisir & Societe-Society and Leisure,
24, 557-582.
Petry, N. M. (2005). Pathological gambling: Etiology, comorbidity,
and treatment. Washington, DC: American Psychological Associa-
tion. doi:10.1037/10894-000
Raylu, N., & Oei, T. P. S. (2002). Pathological gambling: A compre-
hensive review. Clinical Psychology Review, 22, 1009-1061.
doi:10.1016/S0272-7358(02)00101-0
Raylu, N., & Oei, T. P. S. (2004). The Gambling Urge Scale: Devel-
opment, confirmatory factor validation, and psychometric properties.
Psychology of Addictive Behaviors, 18, 100-105.
doi:10.1037/0893-164X.18.2.100
Raylu, N., & Oei, T. P. S. (2004a). Role of culture in gambling and
problem gambling. Clinical Psychology Review, 23, 1087-1114.
doi:10.1016/j.cpr.2003.09.005
Raylu, N., & Oei, T. P. S. (2004b). The Gambling Related Cognitions
Scale (GRCS): Development, confirmatory factor validation and psy-
chometric properties. Addiction, 99, 757-769.
doi:10.1111/j.1360-0443.2004.00753.x
Scull, S., & Woolcock, G. (2005). Problem gambling in non-English
speaking background communities in Queensland, Australia: A
qualitative exploration. International Gambling Studies, 5, 29-44.
doi:10.1080/14459790500097939
Sharpe, L. (2002). A reformulated cognitive-behavioral model of prob-
lem gambling: A biopsychosocial perspective. Clinical Psychology
Review, 22, 1-25. doi:10.1016/S0272-7358(00)00087-8
Sin, R. (1997). Gambling and problem gambling among the Chinese in
Quebec: An exploratory study. Quebec: Chinese Family Service of
Greater Montreal.
Skinner, M. D., & Aubin, H.-J. (2010). Craving’s place in addiction
theory: Contributions of the major models. Neuroscience and Biobe-
havioral Reviews, 34, 606-623. doi:10.1016/j.neubiorev.2009.11.024
Stinchfield, R. (2002). Reliability, validity, and classification accuracy
of the South Oaks Gambling Screen (SOGS). Addictive Behaviors,
27, 1-19. doi:10.1016/S0306-4603(00)00158-1
Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics
(5th ed.). Boston, MA: Pearson.
Tan, A. K. G., Yen, S. T., & Nayga, R. M. Jr. (2010). Socio-demo-
graphic determinants of gambling participation and expenditures:
Evidence from Malaysia. International Journal of Consumer Studies,
34, 316-325. doi:10.1111/j.1470-6431.2009.00856.x
Tang, C. S., Wu, A. M. S., Tang, J. Y. C., & Yan, E. C. W. (2010).
Reliability, validity, and cut scores of the South Oaks Gambling
Screen (SOGS) for Chinese. Journal of Gambling Studies, 26,
145-158. doi:10.1007/s10899-009-9147-7
Tiffany, S. T. (1999). Cognitive concepts of craving. Alcohol Research
& Health, 23, 215-224.
Tiffany, S. T., & Conklin, C. A. (2000). A cognitive processing model
of alcohol craving and compulsive alcohol use. Addiction, 95,
145-153. doi:10.1046/j.1360-0443.95.8s2.3.x
Tseng, W. S., Lin, T. Y., & Yeh, E. K. (1995). Chinese societies and
mental health. Hong Kong: Oxford University Press.
Victorian Casino and Gaming Authority. (2000). The impact of gaming
on specific cultural groups report. Melbourne, Victoria.
Wong, I. L. K., & So, E. M. T. (2003). Prevalence estimates of problem
and pathological gambling in Hong Kong. American Journal of Psy-
chiatry, 160, 1353-1354. doi:10.1176/appi.ajp.160.7.1353