Content uploaded by Sherry H Stewart
Author content
All content in this area was uploaded by Sherry H Stewart
Content may be subject to copyright.
ORIGINAL PAPER
Effects of Gambling-Related Cues on the Activation
of Implicit and Explicit Gambling Outcome Expectancies
in Regular Gamblers
Melissa J. Stewart •Sunghwan Yi •Sherry H. Stewart
Springer Science+Business Media New York 2013
Abstract The current research examined whether the presentation of gambling-related
cues facilitates the activation of gambling outcome expectancies using both reaction time
(RT) and self-report modes of assessment. Gambling outcome expectancies were assessed
by having regular casino or online gamblers (N=58) complete an outcome expectancy
RT task, as well as a self-report measure of gambling outcome expectancies, both before
and after exposure to one of two randomly assigned cue conditions (i.e., casino or control
video). Consistent with hypotheses, participants exposed to gambling-related cues (i.e.,
casino cue video condition) responded faster to positive outcome expectancy words pre-
ceded by gambling prime relative to non-gambling prime pictures on the post-cue RT task.
Similarly, participants in the casino cue video condition self-reported significantly stronger
positive gambling outcome expectancies than those in the control cue video condition
following cue exposure. Activation of negative gambling outcome expectancies was not
observed on either the RT task or self-report measure. The results indicate that exposure to
gambling cues activates both implicit and explicit positive gambling outcome expectancies
among regular gamblers.
M. J. Stewart (&)S. H. Stewart
Department of Psychology, Dalhousie University, 1355 Oxford Street, Halifax, NS B3H 4J1,
Canada
e-mail: stewart.melissa@dal.ca
S. H. Stewart
e-mail: sherry.stewart@dal.ca
S. Yi
Department of Marketing & Consumer Studies, University of Guelph, Macdonald Institute Building,
50 Stone Road East, Guelph, ON N1G 2W1, Canada
e-mail: syi@uoguelph.ca
S. H. Stewart
Department of Psychiatry, QEII Health Sciences Centre, Dalhousie University, 5909 Veterans’
Memorial Lane, 8th floor, Abbie J. Lane Memorial Building, Halifax, NS B3H 2E2, Canada
S. H. Stewart
Department of Community Health & Epidemiology, Centre for Clinical Research, Dalhousie
University, 5790 University Avenue, Halifax, NS B3H 1V7, Canada
123
J Gambl Stud
DOI 10.1007/s10899-013-9383-8
Keywords Gambling outcome expectancies Gambling-related cues Implicit measures
Explicit measures Affective priming task
Introduction
Research on outcome expectancies has been extremely influential in the field of alcohol
addiction (see Goldman et al. 1999; Sayette 1999, for reviews). Researchers have dem-
onstrated that alcohol use behaviors are influenced by the outcomes that individuals expect
will occur from consuming alcohol (e.g., ‘‘If I drink, then…’’). Further, alcohol outcome
expectancies have been theorized as a key mediator of the relation between exposure to
alcohol-related cues (or drinking ‘triggers’) and alcohol use behavior (Goldman 2002;
Goldman and Rather 1993). Indeed, positive alcohol outcome expectancies have been
found to be strongly associated with more frequent and intense drinking (Goldman et al.
1999).
Given the theoretical significance of outcome expectancies in the alcohol field, as well
as the similarities between alcohol and gambling as addictions (e.g., Potenza 2006), out-
come expectancies also may play an important role in gambling. However, little research
has investigated the significance of outcome expectancies in relation to gambling. This is
troubling, as the few studies that have been conducted have shown that gambling outcome
expectancies are indeed associated with increased levels of gambling problems (Gillespie
et al. 2007b). Previous research on gambling outcome expectancies has primarily relied on
self-reports (Gillespie et al. 2007a,b; Shead and Hodgins 2009). For example, Gillespie
et al. (2007a) developed a self-report measure of gambling outcome expectancies, which
consists of three positive expectancy subscales (i.e., enjoyment/arousal, self-enhancement,
and money) and two negative expectancy subscales (i.e., over-involvement, and [negative]
emotional impact). Probable pathological gamblers scored higher than other gamblers on
their expectations of both the positive and negative outcomes of gambling (Gillespie et al.
2007b).
Although use of the self-report mode has been typical in assessing outcome expec-
tancies, its limitations have been increasingly recognized (e.g., Kramer and Goldman 2003;
Palfai and Ostafin 2003). Influenced by cognitive psychology, addiction researchers have
increasingly adopted the view that alcohol outcome expectancies are represented in the
associative memory network (e.g., Goldman et al. 1999; Stacy 1997). According to this
view, the strength of a given alcohol outcome expectancy is defined as the speed with
which the concept of drinking (or exposure to alcohol-related cues) facilitates the acti-
vation of the outcome expectancy in memory. For example, individuals who have a very
strong positive outcome expectancy of alcohol use should experience faster activation of
the positive outcome expectancy when exposed to beer or liquor bottles than those with
weak positive alcohol outcome expectancies.
In order to assess these individual differences in the strength of outcome expectancies,
addiction researchers have used implicit measures, such as reaction time (RT) tasks.
Compared to self-report measures of outcome expectancies, RT measures are less sus-
ceptible to social desirability bias, more efficient, and more difficult for participants to
consciously control (De Houwer 2006; Wiers et al. 2002). One such implicit RT measure is
the affective priming task (Fazio et al. 1986), which is widely used for examining the
automatic activation of attitudes from memory. Specifically, this procedure assesses the
extent to which the presentation of a prime (e.g., a picture) activates an associated
J Gambl Stud
123
evaluation (i.e., positive or negative) from memory. On each trial, the presentation of a
prime is followed by the display of either a positive or negative evaluative adjective (i.e.,
target). The participant’s task is to indicate the connotation of the target word as quickly as
possible (e.g., is the word ‘positive’ or ‘negative’?). Participants’ RT latency to this
judgment represents the outcome measure (Fazio et al. 1995). Alcohol outcome expec-
tancies assessed using implicit measures, such as the affective priming task, have been
found to be positively associated with alcohol consumption (see Goldman et al. 2006).
Applying the affective priming paradigm to the gambling domain, differences in the
strength of gambling outcome expectancies theoretically can be assessed by comparing the
speed with which exposure to the concept of gambling facilitates the automatic activation
of gambling outcome expectancies in memory. Specifically, individuals who have a strong
positive expectancy of gambling outcomes should experience faster activation of the
positive outcome expectancy upon exposure to gambling-related cues than those with a
weak positive outcome expectancy of gambling.
Despite the advantages of implicit measures discussed above, self-report measures of
outcome expectancies are not necessarily inferior to RT measures. In their reflective-
impulsive model, Strack and Deutsch (2004) purport that behavior is controlled by two
interacting systems: the reflective system and the impulsive system. In the reflective system,
behavior is the result of a conscious decisional process whereas in the impulsive system,
behavior is evoked through unconscious associations. Thus, both self-report and RT mea-
sures can be considered complementary in assessing outcome expectancies in that self-
report measures assess deliberative determinants of behavior, while RT measures assess
automatic determinants (see Wiers and Stacy 2006). Given that implicit and explicit
measures appear to tap into different facets of outcome expectancies in the alcohol field
(e.g., de Jong Wiers et al. 2007; Kramer and Goldman 2003), it may be similarly important
to make use of both modes of assessment when examining gambling outcome expectancies.
The purpose of this study was to investigate factors that facilitate the activation of
gambling outcome expectancies using both RT and self-report modes of assessment.
Drawing upon the affective priming paradigm (Fazio et al. 1986), the current study
assessed whether the presentation of gambling-related concepts (i.e., primes) leads to the
automatic activation of gambling outcome expectancies stored in regular gamblers’
memory networks. Specifically, it has been previously postulated in the alcohol field
(Goldman and Rather 1993; Goldman 2002) that situational cues related to alcohol use that
are repeatedly paired with positive affective outcomes of drinking are stored together in
memory. When individuals are later exposed to situational alcohol cues, these cues are said
to activate positive outcome expectancies in memory. In fact, Palfai and Ostafin (2003)
found that the RT to positive alcohol outcome expectancy terms was significantly faster
when hazardous drinkers consumed a priming dose of alcohol than when they consumed a
non-alcoholic placebo beverage.
In the present study, it was proposed that exposure to gambling-related cues (i.e., a
video of casino scenes) immediately prior to the assessment of gambling outcome
expectancies would activate positive gambling outcome expectancies in memory among
regular gamblers. Thus, it was predicted that compared to those who viewed a video
unrelated to gambling (i.e., control cue video condition), gamblers who viewed a gam-
bling-related video (i.e., casino cue video condition) would subsequently be significantly
faster in responding to positive outcome expectancy targets when they were preceded by
gambling picture primes relative to non-gambling picture primes. In relation to the explicit
(self-report) measure of gambling outcome expectancies, it was predicted that participants
in the casino cue video condition would self-report significantly higher positive gambling
J Gambl Stud
123
outcome expectancies following cue exposure than those in the control cue video condi-
tion. We expected these effects to be observed only at the post-cue test phase (i.e., after
viewing the video). The pre-cue test phase was included and analyzed as a pre-manipu-
lation baseline. Drawing upon the reflective-impulsive model of behavior (Strack and
Deutsch 2004), because the casino cue video was of a relatively long duration, allowing
ample opportunity for participants to process the gambling stimuli, it was expected that the
video manipulation would have similar effects on both our implicit and explicit measures
of positive gambling outcome expectancies.
With respect to negative outcome expectancies, in the alcohol literature some studies
have found a negative association between negative outcome expectancies and drinking
while others have found a positive association. These findings suggest that negative alcohol
expectancies may be protective against heavy drinking or a consequence of heavy con-
sumption, respectively (Jones and McMahon 1996; Stacy et al. 1990). Since the direction
of the hypothesized relation between negative outcome expectancies and gambling remains
unclear, we included implicit and explicit measures of negative gambling outcome
expectancies to explore the effects of gambling cue exposure on these measures.
Method
Participants
Participants consisted of 58 adult gamblers (38 males and 20 females) who ranged in age
from 19 to 61 years (M=29.97, SD =12.04). Participants were recruited through
advertisements posted on university bulletin boards, as well as in local newspapers and
classified websites. Thirty-six participants were recruited from the Halifax Regional
Municipality in Nova Scotia, while the remaining 22 participants were recruited from the
greater Guelph area in Ontario. Upon leaving their contact information, potential partici-
pants were contacted by telephone and screened to determine eligibility.
In order to be eligible to participate, individuals had to have gambled at a casino or
online
1
at least three times over the past two months. As RT measures require extremely
rapid responses to English words, only individuals whose native language was English
were eligible to participate. Individuals were excluded if they were currently attempting to
quit gambling or receiving treatment for problem gambling given ethical concerns that
exposure to gambling-related cues could theoretically trigger a return to problem gambling.
Participants were compensated $30 for their participation in the study.
Using the Problem Gambling Severity Index (PGSI) from the Canadian Problem
Gambling Index (CPGI; Ferris and Wynne 2001), participants consisted of 3 non-problem
gamblers (i.e., total score of 0), 8 low-risk gamblers (i.e., total score ranging from 1 to 2),
32 moderate-risk gamblers (i.e., total score ranging from 3 to 7), and 15 high-risk/problem
gamblers (i.e., total score of 8 or above). Total scores on the PGSI ranged from 0 to 25
(M=6.26; SD =5.20). Participants engaged in a range of gambling activities during the
three months prior to taking part in the study, including casino gambling (e.g., slots,
blackjack, poker, roulette), video lottery terminal gambling, sports betting (e.g., Proline,
hockey pools), online gambling, card games with friends, and raffle and lottery tickets.
1
The inclusion criterion that individuals had to have gambled at a casino or online was made to ensure that
the gambling-related primes in the RT task would apply to all gambler participants equally.
J Gambl Stud
123
Materials
Problem Gambling Symptoms
The nine-item PGSI scale of the CPGI (Ferris and Wynne 2001) was used to assess the
presence and severity of gambling problems among participants. The PGSI contains five
items that assess problem gambling behavior (e.g., ‘‘Have you bet more than you could
really afford to lose?’’) and four items addressing the negative consequences of gambling
(e.g., ‘‘Has gambling caused you any health problems, including stress or anxiety?’’). For
each item, respondents indicated the frequency at which they have engaged in the behavior
or experienced the given consequence in the last 12 months using a four-point scale
ranging from 0 (never)to3(almost always). Previous research indicates that the PGSI has
good psychometric properties. Specifically, Ferris and Wynne (2001) found that the PGSI
demonstrated adequate reliability in terms of both internal consistency (a=.84) and test–
retest reliability (r=.78). The PGSI has also been found to demonstrate overall good
validity (i.e., construct, criterion, content validity) as a measure of problem gambling
(Ferris and Wynne 2001). Further, an independent study found that compared to other
gambling measures (e.g., SOGS), the PGSI demonstrated favourable psychometric prop-
erties in non-clinical populations in terms of construct validity, classification validity, and
item difficulty (McMillen and Wenzel 2006). In the present study, the PGSI demonstrated
good internal consistency (a=.89).
Self-Reported Gambling Outcome Expectancies
The 23-item Gambling Expectancy Questionnaire (GEQ; Gillespie et al. 2007a) was used
to assess self-reported gambling outcome expectancies at both pre and post-cue manipu-
lation. The GEQ consists of three positive expectancy subscales (enjoyment/arousal, self-
enhancement, and money) and two negative expectancy subscales (over-involvement and
emotional impact). Participants were asked to what extent they expected each item/out-
come (e.g., ‘‘I win money’’; ‘‘I feel excited’’; ‘‘I will feel guilty’’; ‘‘I will not be able to
stop’’) would occur when gambling on a seven-point scale ranging from 1 (no chance)to7
(certain to happen). In relation to its psychometric properties, Gillespie et al. (2007a)
reported that each of the subscales demonstrated adequate to good internal reliability. In
order to obtain an overall measure of participants’ positive gambling outcome expectan-
cies, the three positive subscales of the GEQ were combined in the present study. This
resulted in a 15-item scale assessing positive gambling outcome expectancies. The two
negative expectancy subscales of the GEQ were also combined to obtain an overall
measure of participants’ negative gambling outcome expectancies. This resulted in an
8-item scale assessing self-reported negative gambling outcome expectancies. The 15-item
positive gambling expectancy scale of the GEQ demonstrated adequate to good reliability
at both pre-cue (a=.78) and post-cue (a=.83) administration. Further, the 8-item
negative gambling expectancy scale demonstrated excellent reliability during both pre-cue
(a=.92) and post-cue administration (a=.95) in the present study.
Affective Outcome Expectancy RT Task
Adapted from the classic affective priming task (Fazio et al. 1995), this RT-based task was
used to assess the activation of affective gambling outcome expectancies. The task was
designed to measure how quickly individuals respond to positive and negative gambling
J Gambl Stud
123
outcome expectancy words (i.e., targets) immediately after being primed by gambling
versus control category (i.e., track and field) pictures.
2
The task was executed via
Empirisoft Inc.’s DirectRT experimental psychology software (Jarvis 2010). The target
word exemplars were selected based on a review of established self-report measures of
gambling outcome expectancies (e.g., GEQ; Gillespie et al. 2007a), as well as synonyms of
words from these measures. In total, there were 10 positive outcome expectancy words and
10 negative outcome expectancy words used as targets (see Table 1). In addition, 10
gambling-related and 10 non-gambling-related pictures were used as primes. The task
consisted of two phases: pre-cue test phase (i.e., baseline) and post-cue test phase. Each
phase began with one block of four practice trials, and two blocks with 20 test trials each
(total number of trials was 88 across the two phases). The stimuli for practice trials were
different than those presented during the test trials. Each phase was presented to partici-
pants as one continuous series. During each priming phase, each outcome expectancy
target word was presented four times: twice preceded by a gambling prime picture, and
twice preceded by a non-gambling prime picture. The order of primes and targets within
each block was counterbalanced across participants.
In the pre-cue test phase, each trial started with the presentation of either a gambling-
related or non-gambling-related (i.e., track and field) picture in the centre of the screen
which lasted for 200 ms. This was followed by a blank screen (100 ms), then by the
presentation of a target word (in the centre of the screen as well) that had either a
positive (e.g., excitement) or negative (e.g., tension) connotation. Participants were asked
to respond to words that had a negative connotation by clicking the ‘‘Z’’ key on the
keyboard, and to respond to words that have a positive connotation by clicking the ‘‘/’’
key. The length of the inter-trial interval was 1,000 ms. The post-cue test phase was
identical to the pre-cue test phase. Participants were told that they needed to pay
attention to the pictures presented on the screen as their memory for the pictures may be
tested later. Participants were also informed that the first four trials of each test phase
were practice trials.
Table 1 Word exemplars used
in the gambling outcome expec-
tancy RT task
For the practice trials, positive
outcome expectancy words
consisted of: ‘amusement’ and
‘happiness’, whereas negative
outcome expectancy words
consisted of ‘boredom’ and
‘sorrow’
Positive outcome
expectancy words
Negative outcome
expectancy words
Fun Guilt
Relaxation Shame
Excitement Tension
Enjoyment Confusion
Esteem Frustration
Acceptance Anxiety
Winning Worry
Stimulation Dissatisfaction
Pleasure Anger
Satisfaction Displeasure
2
We selected track and field as the control category because it is an activity that is similar to gambling in
both size and complexity. Specifically, both gambling and track and field are broad categories that
encompass a variety of different activities. Track and field is also an activity that could theoretically be
associated with both positive outcomes (excitement, winning) and negative outcomes (frustration, tension).
J Gambl Stud
123
Procedure
Upon arrival at the laboratory, participants provided informed consent and were randomly
assigned to the casino cue video condition or control cue video condition. Participants then
engaged in the first phase of the affective outcome expectancy RT task (pre-cue test
phase). Immediately after the pre-cue test phase, participants completed the GEQ
(Gillespie et al. 2007a) and a demographic questionnaire. Upon completing these ques-
tionnaires, participants were exposed to the cue manipulation to which they had been
randomly assigned (either the casino cue video or control cue video condition). In the
casino cue video condition (n=29), participants watched a 5-min video of typical casino
scenes with ambient noise reflecting the sounds that would be heard in a casino (e.g.,
sounds of slot machines paying out). In the control cue video condition (n=29), par-
ticipants watched a 5-min video of typical track and field scenes with ambient noise
reflecting the sounds that would be heard in a track and field audience (e.g., cheering and
clapping). After watching the video, participants engaged in the post-cue test phase of the
RT task. Participants then completed a second administration of the GEQ (Gillespie et al.
2007a) in order to determine whether any differences in outcome expectancies were
present upon being exposed to the gambling-related/control video. Participants were then
debriefed and compensated $30 for their time and effort.
Results
Preliminary Analyses
The data in both the pre-cue and post-cue manipulation phases contained a small pro-
portion of extremely slow and fast responses. Respectively, such responses typically
indicate momentary inattention and responses initiated prior to receiving the stimulus
(Greenwald et al. 1998). Not only are such responses considered problematic because they
lead to a distortion of means and inflation of variance, but also because they represent
phenomenon outside of interest. Following the recommended procedures to correct for
such responses (Greenwald et al. 1998), values below 300 ms were recoded to 300 ms and
those above 3,000 ms were recoded to 3,000 ms. Further, in order to reduce the charac-
teristic positive skewness of RT latencies and normalize the distribution, a log transfor-
mation was performed on the RT data prior to averaging mean RT scores (see Fazio 1990;
Greenwald et al. 1998). To aid in the interpretation of data, raw (untransformed) RTs are
displayed for descriptive purposes only. At each testing time (i.e., pre- and post-video
exposure), four composite RT scores were calculated for each participant. These repre-
sented the mean RTs for trials involving each of the four prime-target combinations (i.e.,
gambling prime-positive outcome expectancy target; gambling prime-negative outcome
expectancy target; non-gambling prime-positive outcome expectancy target; non-gambling
prime-negative outcome expectancy target).
In order to confirm random assignment and the equivalency of groups by experimental
condition, analyses were conducted to determine whether there were any systematic, pre-
existing differences between the casino cue video condition and the control cue video
condition on level of problem gambling severity, age, and gender. In relation to level of
problem gambling severity, an independent samples ttest revealed no significant difference
between the control cue video (M=6.07, SD =5.48) and casino cue video (M=6.44,
SD =5.00) conditions on level of problem gambling severity, t(56) =.28, p=.78.
J Gambl Stud
123
Further, an independent samples ttest revealed no significant age differences between
participants in the control cue video (M=30.66, SD =12.70) and casino cue video
(M=29.28, SD =11.52) conditions, t(56) =-.43, p=.67. Lastly, a Chi square test
revealed no significant gender differences between the control cue video (males: n=19;
females: n=10); and casino cue video (males: n=19; females: n=10) conditions, v
2
(1, N=58) =.00, p=1.00. As such, it was deemed that any differences observed
between groups may be attributed to the effect of the manipulation and not to pre-existing
or systematic differences between groups. Further, correlational analyses failed to reveal
any significant correlations (ps[.05) between the implicit and explicit measures of
gambling outcome expectancies in either the casino or control video cue conditions at pre-
and post-video cue manipulation (correlations ranged from -.08 to .33).
Affective Outcome Expectancy RT Task Performance
Prior to testing our hypotheses, we examined whether there were any cue manipulation con-
dition differences in RTs to positive or negative outcome expectancy targets when they were
preceded by gambling versus non-gambling primes during the pre-cue manipulation phase (i.e.,
baseline) of the RT task. To do so, two separate 2 92 mixed factorial ANOVAs were con-
ducted—one for the RT data for the positive outcome expectancy targets, and the other for the
RT data for the negative outcome expectancy targets. In both ANOVAs, the between subjects
factor was cue manipulation condition (casino cue video vs. control cue video) and the within
subjects factor was type of prime (gambling vs. non-gambling). The analyses confirmed that
during the pre-cue manipulation phase, there were no significant condition, prime, or interac-
tion effects on RTs to positive or negative outcome expectancy targets (i.e., all ps[.05). These
findings indicate that there was no tendency to associate gambling primes with positive (or
negative) outcomes in either of the two randomly assigned cue conditions (casino or control
video cue), prior to video cue exposure. Descriptive statistics for RTs to negative and positive
outcome expectancy targets when preceded by gambling and non-gambling primes for the
casino cue video and control cue video conditions during the pre-cue manipulation phase are
displayed in Table 2. This lack of effects at the pre-cue exposure phase meant that baseline RTs
did not need to be controlled in hypothesis testing.
We employed a priori planned comparisons in our hypotheses testing. Specifically,
following the analytic strategy of Birch et al. (2008), we analyzed RT data on the post-cue
manipulation affective priming task with relation to the initial hypotheses by decomposing
Table 2 Means and standard deviations of pre-cue RTs (in milliseconds) to positive and negative outcome
expectancy words upon presentation of gambling and non-gambling primes for the control cue video and
casino cue video conditions
Gambling primes Non-gambling primes
MSDM SD
Control cue video condition
Positive target words 807.82 237.81 795.86 194.79
Negative target words 840.56 264.90 866.45 301.84
Casino cue video condition
Positive target words 837.39 250.08 818.97 218.77
Negative target words 833.65 218.11 863.00 247.41
J Gambl Stud
123
the full 2 (cue manipulation: casino cue video vs. control cue video) 92 (primes: gam-
bling vs. non-gambling) 92 (targets: positive vs. negative outcome expectancy words)
table of means into a series of directional paired samples ttests. As recommended by
Tabachnick and Fidell (2007), conventional alpha levels were used to analyse the com-
parisons of primary interest first (Tabachnick and Fidell 2007). Specifically, these direc-
tional paired-samples ttests were used to compare RTs to categorize outcome expectancy
targets after exposure to gambling versus non-gambling primes at the post-cue manipu-
lation testing time for each cue manipulation condition and target type separately.
Descriptive statistics for RTs to negative and positive outcome expectancy targets when
preceded by gambling and non-gambling primes for the casino cue and control cue video
conditions during the post-cue manipulation phase are displayed in Table 3.
Consistent with hypotheses, participants in the casino cue video condition responded faster
to positive outcome expectancy words when they were preceded by gambling primes relative to
non-gambling primes, t(28) =-1.69, p=.05, representing a marginally significant differ-
ence in RTs (see Table 3). While participants in control cue video condition tended to respond
faster to positive outcome expectancy words whenthey were preceded by non-gambling primes
relative to gambling primes, a significant facilitation of positive gambling outcome expec-
tancies by non-gambling primes was not observed in the control cue condition, t(28) =1.04,
p=.16. This lack of significance can be attributed to greater variability (i.e., larger standard
deviation) in RT among participants in the control cue video condition relative to those in the
gambling cue video condition when responding to positive target words after being exposed to
both gambling primes and non-gambling primes (see Table 3).
When examining RTs to negative outcome expectancy targets among participants in the
casino cue video condition, no significant differences were found between participants’
RTs to targets preceded by gambling primes versus non-gambling primes, t(28) =-.46,
p=.33 (see Table 3). Similarly, there were no significant differences in the RTs to
negative outcome expectancy targets among participants in the control cue video condition
when they were exposed to gambling primes versus non-gambling primes, t(28) =-.33,
p=.37 (see Table 3).
Cue Condition Differences in Self-Reported Gambling Outcome Expectancies
Prior to assessing whether exposure to gambling-related cues led to an increase in self-
reported gambling outcome expectancies, we examined whether any group differences
Table 3 Means and standard deviations of post-cue RTs (in milliseconds) to positive and negative outcome
expectancy words upon presentation of gambling and non-gambling primes for the control cue video and
casino cue video conditions
Gambling primes Non-gambling primes
MSDM SD
Control cue video condition
Positive target words 750.28 262.45 712.74 177.81
Negative target words 788.64 258.26 801.77 281.38
Casino cue video condition
Positive target words 721.19* 160.79 739.34* 166.19
Negative target words 740.34 157.61 753.23 190.37
*Indicates a marginally significant difference between means (p=.05)
J Gambl Stud
123
existed in self-reported positive and negative outcome expectancies [as measured by the
GEQ (Gillespie et al. 2007a)] before viewing the cue manipulation videos (i.e., at baseline).
In order to test this, two separate independent-samples ttests were conducted. In both ttests,
the independent variable was cue manipulation condition (casino cue video vs. control cue
video). In the first ttest, the dependent variable was self-reported positive gambling out-
come expectancies on the GEQ, whereas the dependent variable in the second ttest was self-
reported negative gambling outcome expectancies on the GEQ. The analyses revealed that
there were no significant differences in either self-reported positive (casino cue video:
M=4.46, SD =.61; control cue video: M=4.27, SD =.68) or negative (casino cue
video: M=2.96, SD =1.35; control cue video: M=2.72, SD =1.05) gambling outcome
expectancies prior to the cue manipulation (i.e., both ps[.05). These findings indicate
random assignment to cue manipulation conditions was effective in equating the two
conditions on their baseline (pre-cue exposure) positive and negative gambling outcome
expectancies. This lack of condition effects at the pre-cue exposure (baseline) phase meant
that baseline GEQ scores did not need to be controlled in hypothesis testing.
We then examined our a priori planned comparison regarding cue condition differences
in self-reported positive gambling outcome expectancies following exposure to the cue
manipulation by performing directional independent samples ttests. Consistent with our
hypothesis, participants in the casino cue video condition (M=4.26, SD =.60) reported
significantly higher scores on the self-report measure of positive gambling outcome
expectancies than those in the control cue video condition (M=3.87, SD =.85),
t(56) =1.98, p=.03. However, this effect was not found for negative gambling outcome
expectancies. Specifically, an independent samples ttest revealed that participants in the
casino cue video (M=2.66, SD =1.47) and control cue video condition (M=2.45,
SD =1.29) did not significantly differ in their self-reported negative gambling outcome
expectancies after exposure to the cue manipulation, t(56) =.58, p=.56.
Discussion
Although outcome expectancies have been found to play an important role in addictive
behaviors (e.g., Goldman et al. 1999; Sayette 1999), a paucity of research has been con-
ducted on the relation between outcome expectancies and gambling. To address this gap in
the literature, the current research investigated the possibility that exposure to gambling-
related cues facilitates the activation of gambling outcome expectancies using implicit (i.e.,
RT), as well as explicit (i.e., self-report) modes of assessment. It was hypothesized that
compared to those who viewed a video unrelated to gambling (i.e., control cue video
condition), gamblers who viewed a gambling-related video (i.e., casino cue video condi-
tion) would subsequently be significantly faster in responding to positive outcome
expectancy targets when they are preceded by gambling picture primes relative to non-
gambling picture primes. In relation to the explicit (self-report) measure of gambling
outcome expectancies, it was predicted that participants in the casino cue video condition
would self-report significantly higher positive gambling outcome expectancies following
cue exposure than those in the control cue video condition.
Cue Condition Differences in the RT Measure of Outcome Expectancies
Consistent with our predictions, participants who were exposed to a video of typical
gambling-related scenes (i.e., casino cue video condition) responded faster to positive
J Gambl Stud
123
outcome expectancy words when they were preceded by gambling primes relative to non-
gambling primes (p=.05). This facilitation of positive outcome expectancies by gambling
primes was not observed in the control cue video condition, as participants who viewed a
video of typical track and field scenes did not significantly differ in RTs to positive
outcome expectancy words when they were preceded by gambling primes relative to non-
gambling primes. While participants in this latter condition, who were exposed to a track-
and-field control video, tended to respond faster to positive outcome expectancy words
when they were preceded by track-and-field control primes relative to gambling primes,
this difference was not statistically significant (p[.05). This pattern of findings rules out
the possibility that participants just responded faster to positive outcome expectancy words
following primes that were ‘‘similar’’ to the previously-viewed videos (i.e., following
gambling primes in the gambling video condition; following track-and-field primes in the
track-and-field video condition) than following primes that were ‘‘different’’ than the
previously-viewed videos (i.e., following track-and-field primes in the gambling video
condition; following gambling primes in the track-and-field video condition).
Instead, these results suggest that exposure to gambling-related cues immediately prior
to the assessment of gambling outcome expectancies activates positive outcome expec-
tancies in memory among regular gamblers. Further, these findings provide additional
support to previous research (Goldman 2002; Goldman and Rather 1993) on the role of
outcome expectancies in alcohol use, which proposes that situational cues related to
alcohol use that are repeatedly paired with positive affective outcomes are stored together
with these outcomes in memory. When later exposed to alcohol cues, these cues sub-
stantially facilitate the degree to which the concept of alcohol activates positive outcome
expectancies. In addition to providing further empirical support for this proposal, the
current study extends this line of reasoning by showing evidence of its applicability to
gambling. It is also important to highlight that engaging in gambling activities was not
necessary to obtain these findings. Specifically, the activation of positive gambling out-
come expectancies was found to occur when participants were simply exposed to typical
gambling scenes, an experience that appears quite similar to watching others gamble or
viewing gambling-related advertisements.
In addition, these findings partially coincide with Palfai and Ostafin’s (2003) research
assessing the activation of alcohol outcome expectancies using a similar RT task. Spe-
cifically, Palfai and Ostafin (2003) found that compared to administration of a non-alco-
holic beverage, administration of a low dose of alcohol was associated with a faster RT to
positive alcohol outcome expectancies among hazardous drinkers. Unexpectedly, this did
not differ depending upon whether participants in their study received alcohol or non-
alcohol related primes prior to the presentation of outcome expectancy targets. In contrast,
the current research found that participants in the casino cue video condition exhibited
evidence of faster responses to positive outcome expectancy targets when primed by
gambling pictures than when primed by non-gambling pictures.
One possible explanation for this discrepancy in findings may relate to the stimuli used
as primes. Specifically, primes in the current research consisted of gambling and non-
gambling pictures, whereas in Palfai and Ostafin’s (2003) research, alcohol and non-
alcohol related words were used as primes. Given that pictures have been found to be
remembered better than words (e.g., Grady et al. 1998; Seifert 1997) and the primes in both
studies were presented only briefly, it is possible that participants in the current study were
better able to store the gambling and non-gambling primes in memory. As a result of
a potential increased memory of primes in the current research, the presentation of gam-
bling primes relative to non-gambling primes among those previously exposed to
J Gambl Stud
123
gambling-related cues on the video led to a faster activation of positive outcome expec-
tancy concepts in memory. Applying this to the alcohol domain, results of the current
research suggest that the activation of positive outcome expectancies among regular
drinkers may only require exposure to drinking scenes rather than actual alcohol use.
A further potential explanation for the discrepancy in findings may relate to the different
analytic strategies employed in the present study and Palfai and Ostafin’s (2003) research.
Specifically, the current research analyzed the RT data on the post-cue manipulation
affective priming task with relation to our initial hypotheses by conducting specific
planned comparisons, whereas Palfai and Ostafin’s (2003) did not decompose the full
tables of means (i.e., video cue manipulation, primes, and targets) when analyzing the RT
task performance in their research and instead used an omnibus ANOVA.
Although exposure to gambling-related cues appeared to activate positive gambling
outcome expectancies in memory among regular gamblers in our study, this was not the
case for negative gambling outcome expectancies. Specifically, when examining RTs to
negative outcome expectancy targets among participants in the casino cue video condition,
no significant differences were found between participants’ RTs to targets that were pre-
ceded by gambling primes versus non-gambling primes. Similarly, there were no signifi-
cant differences in the RTs to negative outcome expectancy targets among participants in
the control cue video condition after they were exposed to gambling primes versus non-
gambling primes. These findings are consistent with previous research in the alcohol field,
which suggests that negative outcome expectancies reflect outcomes less proximal to
alcohol use than positive outcome expectancies (Jones et al. 2001).
Cue Condition Differences in Self-Reported Gambling Outcome Expectancies
When examining participants’ self-reported positive gambling outcome expectancies fol-
lowing exposure to the cue manipulation, it was found that participants in the casino cue
video condition scored significantly higher on the self-report measure of positive gambling
outcome expectancies than those in the control cue video condition. However, this was not
the case for negative gambling outcome expectancies, as participants in the casino cue and
control cue video conditions did not significantly differ in their self-reported negative
gambling outcome expectancies after exposure to the video cue manipulation.
In line with our predictions, these findings suggest that the presentation of gambling-
related cues leads to an increase in the expected positive outcomes that gamblers report
will occur from gambling. These findings are consistent with the results of the RT task and
as such, provide converging evidence of the impact of gambling-related cues on the
facilitation of positive gambling outcome expectancies. That is, using both implicit and
explicit modes of assessment, the current research found that the presentation of gambling-
related cues leads to an activation of positive gambling outcome expectancies in memory.
Further, these results build upon previous research (e.g., Gillespie et al. 2007b) using this
self-report measure to assess gambling outcome expectancies by suggesting that exposure
to gambling-related cues is associated with an increase in self-reported positive gambling
outcome expectancies. In addition, consistent with some findings from the alcohol outcome
expectancy literature (e.g., de Jong et al. 2007; Jajodia and Earleywine 2003; Kramer and
Goldman 2003), the implicit and explicit measures of positive gambling outcome expec-
tancies were not significantly correlated, suggesting that these two modes of assessment
may be assessing distinct aspects of gambling cognitions. In order to determine whether
this is the case, it is important that future research investigate whether implicit measures of
gambling outcome expectancies assess a unique facet of the gambling cognition domain
J Gambl Stud
123
that cannot be accessed through explicit measurement. Further, it is important that future
research also examine situational variables that may enhance the predictive validity of
implicit measures of gambling outcome expectancies on gambling behavior.
It is also important to note the potential impact of the length of the gambling-cue
exposure on the activation of implicit and explicit positive gambling outcome expectan-
cies. Although we found that the 5-min video of gambling scenes activated positive out-
come expectancies, as measured by both self-report and the RT task, different results may
have been found had the video been of a shorter duration. Specifically, a relatively brief
presentation of gambling-related cues may not allow individuals the time to engage in the
conscious, deliberative processing of gambling outcome expectancies that is captured by
self-report modes of assessment. As such, had the gambling cue been shorter (e.g., 30 s),
an activation of implicit but not explicit positive outcome expectancies may have been
observed among participants in the current study. Further research is necessary to inves-
tigate whether shorter exposure to gambling-related cues (e.g., gambling advertisements)
activates implicit but not explicit gambling outcome expectancies.
Limitations
Some limitations of the present study should be addressed. First, the hypothesized effect on
the implicit task (i.e., faster RTs to positive expectancy words following gambling vs.
control primes in the gambling video condition) reached marginal significance (p=.05)
rather than the conventional alpha level criterion of p\.05. Thus, it is important that this
effect be replicated in future research to determine its reliability. Second, although findings
from the current research provide support to the prediction that, relative to non-gambling-
related cues, gambling-related cues would lead to an increased activation of positive
gambling outcome expectancies among gamblers, it did not include a control group of
individuals who do not gamble. As such, it is not known whether gamblers respond to this
RT task differently than non-gamblers. In order to address this limitation, future research
should use the affective priming paradigm (Fazio et al. 1986) to determine whether
gamblers display a stronger activation of positive gambling outcome expectancies fol-
lowing exposure to gambling-related cues than non-gamblers.
A further caveat of the current research involves the failure to assess differences in the
activation of gambling outcome expectancies based on level of problem gambling severity.
Specifically, given the current sample size, we were unable to assess whether the presentation
of gambling-related cues leads to a greater activation of positive gambling outcome expec-
tancies among problem gamblers relative to at-risk or low-risk gamblers. As such, it would be
important for future research to examine whether the findings of the current research differ
depending upon level of problem gambling severity. In addition, while the current research
found that exposure to gambling-related cues led to the activation of both implicit and explicit
positive gambling outcome expectancies, we did not determine whether such modes of
assessing gambling outcome expectancies are capable of predicting actual gambling
behavior. Further, we did not examine whether implicit and explicit measures of gambling
outcome expectancies contribute unique, as well as shared variance in the prediction of
different forms of gambling behavior, such as the amount of time and money spent gambling.
As such, it is important that future research make use of both modes of assessment to
determine whether they are independent predictors of gambling behavior, as has been pre-
viously shown in the alcohol outcome expectancy literature (e.g., Wiers et al. 2002).
Lastly, a limitation inherent to the RT task used in the present study should be
acknowledged. Specifically, the current research assessed associations between gambling
J Gambl Stud
123
and outcome expectancies relative to associations with another control activity (i.e., track
and field). As such, we cannot discern whether similar results would persist if an activity
other than track and field had been used as a control. In order to address this limitation, it is
important that future research examine the associations between gambling-related cues and
the facilitation of outcome expectancies relative to associations with different activities.
Implications
Despite the limitations noted, findings from the current research have a number of important
practical as well as clinical implications. Firstly, our results appear to provide an argument
against the legalization of casino advertisements, both online and offline. Specifically, as
results revealed that exposure to gambling-related cues led to an increased activation of
positive gambling outcome expectancies, it may be the case that gambling advertisements
facilitate the activation of positive gambling outcome expectancies among gamblers. Fur-
thermore, chronic activation of positive gambling outcome expectancies may pose a risk for
problematic gambling behavior among members of communities located in close proximity
to gambling venues, as well as employees of such establishments. In terms of clinical
implications, results of the current research point to the potential utility of focusing on
altering not only explicit cognitions but also implicit cognitions as interventions for problem
gambling. For example, expectancy challenges, interventions that which aim to reduce
individuals’ expectancies about the rewarding properties of a substance, have been used to
successfully reduce positive explicit outcome expectancies in the alcohol area (Darkes and
Goldman 1993; Darkes et al. 1998). More recently, cognitive retraining methods have been
developed that alter implicit associations with alcohol from positive to negative (Houben
et al. 2010). Importantly, both methods have been associated with reduced drinking
(Houben et al. 2010; Wiers et al. 2005). Given their encouraging results in the alcohol field,
such interventions may prove effective in reducing both implicit and explicit positive
outcome expectancies among gamblers. Moreover, the present findings suggest that these
interventions might be most optimally employed following gambling cue exposure.
Conclusion
Despite the importance of outcome expectancies in addictive behaviors (e.g., Goldman
et al. 1999; Sayette 1999), research examining outcome expectancies in the gambling field
is in its nascent stage. In order to address this gap in the literature as well as facilitate
further research on gambling outcome expectancies, the present study assessed whether
exposure to gambling-related cues activates gambling outcome expectancies using both
implicit (i.e., RT) and explicit (i.e., self-report) assessment modes. As predicted, findings
from the current research indicate that exposure to gambling-related cues selectively
activates both implicit and explicit positive, but not negative, gambling outcome expec-
tancies among regular gamblers. Findings stemming from this preliminary investigation of
the role of gambling-related cues on the implicit and explicit activation of positive gam-
bling outcome expectancies highlight the relevance of outcome expectancies in the field of
gambling research as well as point to the need for further research that examines the role of
gambling outcome expectancies on gambling behavior. Further, results of the current
research indicate that this novel RT task is a useful instrument, in addition to self-report
measures, in terms of its ability to measure the activation of gambling outcome expec-
tancies among regular gamblers.
J Gambl Stud
123
Acknowledgments We would like to acknowledge and thank Pamela Collins and Scott Connors for their
research assistance with this study. This research was supported by a Research Grant from the Ontario
Problem Gambling Research Centre (#44449) to Sunghwan Yi and Sherry H. Stewart.
References
Birch, C. D., Stewart, S. H., Wiers, R. W., Klein, R. M., MacLean, A. D., & Berish, M. J. (2008). The mood-
induced activation of implicit alcohol cognition in enhancement and coping motivated drinkers.
Addictive Behaviors, 33, 565–581.
Darkes, J., & Goldman, M. S. (1993). Expectancy challenge and drinking reduction: Experimental evidence
for a mediational process. Journal of Consulting and Clinical Psychology, 61, 344–353.
Darkes, J., Greenbaum, P. E., & Goldman, M. S. (1998). Sensation seeking-disinhibition and alcohol use:
Exploring issues of criterion contamination. Psychological Assessment, 10, 71–76.
De Houwer, J. (2006). What are implicit measures and why are we using them? In R. W. Wiers &
A. W. Stacy (Eds.), Handbook of implicit cognition and addiction (pp. 11–28). Thousand Oaks, CA:
Sage Publications.
de Jong, P. J., Wiers, R. W., van de Braak, M., & Huijding, J. (2007). Using the extrinsic affective Simon
test as a measure of implicit attitudes towards alcohol: Relationship with drinking behavior and alcohol
problems. Addictive Behaviors, 32, 881–887.
Fazio, R. H. (1990). Multiple processes by which attitudes guide behavior: The MODE model as an
integrative framework. In M. Zanna (Ed.), Advances in experimental social psychology (Vol. 23,
pp. 75–109). San Diego, CA: Academic Press.
Fazio, R. H., Jackson, J. R., Dunton, B. C., & Williams, C. J. (1995). Variability in automatic activation as
unobtrusive measures of racial attitudes: A bona fide pipeline? Journal of Personality and Social
Psychology, 69, 1013–1027.
Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C., & Kardes, F. R. (1986). On the automatic activation of
attitudes. Journal of Personality and Social Psychology, 50, 229–238.
Ferris, J., & Wynne, H. (2001). The Canadian problem gambling index: User’s manual. Ottawa, Canada:
Canadian Centre on Substance Abuse.
Gillespie, M. A. M., Derevensky, J., & Gupta, R. (2007a). Adolescent problem gambling: Developing a
gambling expectancy instrument. Journal of Gambling Issues, 19, 51–68.
Gillespie, M. A. M., Derevensky, J., & Gupta, R. (2007b). The utility of outcome expectancies in the
prediction of adolescent gambling behaviour. Journal of Gambling Issues, 19, 69–85.
Goldman, M. S. (2002). Expectancy and risk for alcoholism. The unfortunate exploitation of a fundamental
characteristic of neurobehavioral adaptation. Alcoholism, Clinical and Experimental Research, 26, 737–746.
Goldman, M. S., Darkes, J., & Del Boca, F. K. (1999). Expectancy mediation of biopsychosocial risk for
alcohol use and alcoholism. In I. Kirsch (Ed.), How expectancies shape experience (pp. 223–262).
Washington, DC: APA Books.
Goldman, M. S., & Rather, B. C. (1993). Substance use disorders: Cognitive models and architectures. In P. Kendall
&K.S.Dobson(Eds.),Psychopathology and cognition (pp. 245–291). Orlando, FL: Academic Press.
Goldman, M. S., Reich, R. R., & Darkes, J. (2006). Expectancy as a unifying construct in alcohol- related
cognition. In R. W. Wiers & A. W. Stacy (Eds.), Handbook of implicit cognition and addiction
(pp. 105–120). Thousand Oaks, CA: Sage Publications.
Grady, C. L., McIntosh, A. R., Rajah, M. N., & Craik, F. I. M. (1998). Neural correlates of the episodic
encoding of pictures and words. Proceedings of the National Academy of Sciences of the United States
of America, 95, 2703–2708.
Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998). Measuring individual differences in implicit
cognition: The implicit association task. Journal of Personality and Social Psychology, 74, 1464–1480.
Houben, K., Havermans, R., & Wiers, R. (2010). Learning to dislike alcohol: Conditioning negative implicit
attitudes toward alcohol and its effect on drinking behavior. Psychopharmacology (Berl), 211, 79–86.
Jajodia, A., & Earleywine, M. (2003). Measuring alcohol expectancies with the implicit association task.
Psychology of Addictive Behaviors, 17, 126–133.
Jarvis, B. G. (2010). DirectRT (version 2010) [computer software]. New York, NY: Empirisoft Corporation.
Jones, B. T., Corbin, W., & Fromme, K. (2001). A review of expectancy theory and alcohol consumption.
Addiction, 96, 57–72.
Jones, B. T., & McMahon, J. (1996). A comparison of positive and negative alcohol expectancy and value
and their multiplicative composite as predictors of post-treatment abstinence survivorship. Addiction,
91, 89–99.
J Gambl Stud
123
Kramer, D. A., & Goldman, M. S. (2003). Using a modified Stroop task to implicitly discern the cognitive
organization of alcohol expectancies. Journal of Abnormal Psychology, 112, 171–175.
McMillen, J., & Wenzel, M. (2006). Measuring problem gambling: Assessment of three prevalence screens.
International Gambling Studies, 6, 147–174.
Palfai, T. P., & Ostafin, B. D. (2003). The influence of alcohol on the activation of outcome expectancies:
The role of evaluative expectancy activation in drinking behavior. Journal of Studies on Alcohol, 64,
111–119.
Potenza, M. N. (2006). Should addictive disorders include non-substance related conditions? Addiction, 101,
142–151.
Sayette, M. A. (1999). Cognitive theory and research. In K. E. Leonard & H. T. Blane (Eds.), Psychological
theories of drinking and alcoholism (pp. 247–291). New York: Guilford Press.
Seifert, L. S. (1997). Activating representations in permanent memory: Different benefits for pictures and
words. Journal of Experimental Psychology. Learning, Memory, and Cognition, 23, 1106–1121.
Shead, N. W., & Hodgins, D. C. (2009). Affect-regulation expectancies among gamblers. Journal of
Gambling Studies, 25, 357–375.
Stacy, A. W. (1997). Memory activation and expectancy as prospective predictors of alcohol and marijuana
use. Journal of Abnormal Psychology, 106, 61–73.
Stacy, A. W., Widaman, K. F., & Marlatt, G. A. (1990). Expectancy models of alcohol use. Journal of
Personality and Social Psychology, 58, 918–928.
Strack, F., & Deutsch, R. (2004). Reflective and impulsive determinants of social behavior. Personality and
Social Psychology Review, 8, 220–247.
Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Boston, MA: Allyn and
Bacon.
Wiers, R. W., & Stacy, A. W. (2006). Handbook of implicit cognition and addiction. Thousand Oaks, CA:
Sage Publications.
Wiers, R. W., Stacy, A. W., Ames, S. L., Noll, J. A., Sayette, M. A., Zack, M., et al. (2002). Implicit and
explicit alcohol-related cognitions. Alcoholism: Clinical and Experimental Research, 26, 129–137.
Wiers, R. W., Van de Luitgaarden, J., van den Wildenberg, E., & Smulders, F. T. Y. (2005). Challenging
implicit and explicit alcohol-related cognitions in young heavy drinkers. Addiction, 100, 806–819.
J Gambl Stud
123