Pathways of Youth Gambling Problem Severity
Ken C. Winters, Randy D. Stinchfield, and
University of Minnesota, Twin Cities Campus
Wendy S. Slutske
University of Missouri—Columbia
Prospective studies are needed to advance knowledge of the developmental features of gambling
involvement and associated problems. Developmental pathways of youth gambling problem severity (no
problem gambling, at-risk gambling, and problem gambling) are described on the basis of a 3-wave data
set that spans midadolescence to young adulthood (N ? 305). The most prevalent group was the resistors
(no problem gambling at all data points); 60% of study participants were in this group. New incidence
cases (no problem gambling followed by at-risk or problem gambling) and desistors (at-risk or problem
gambling followed by no problem gambling) were found among 21% and 13% of participants, respec-
tively. Only 4% of cases were persistors, that is, at-risk or problem gambling at all 3 data waves. Findings
are discussed in light of extant research on adolescent gambling that heretofore has not benefited from
a developmental pathway perspective.
Cross-sectional studies have indicated that youth gambling oc-
curs on a frequency continuum, ranging from no involvement to
experimentation, occasional gambling, regular gambling, and pre-
occupation with serious adverse consequences (see review by
Stinchfield & Winters, 1998). Indeed, adolescents may be partic-
ularly vulnerable to addictions, such as problem gambling, owing
to neurodevelopmental characteristics of this age group (Cham-
bers, Taylor, & Potenza, 2003). A range of adolescents have been
identified as at-risk, problem, or pathological gamblers, various
terms used to reflect gambling-related problems (Winters &
Anderson, 2000). Shaffer and Hall (1996), in their comprehensive
review of all adolescent-prevalence studies, estimated that between
4% and 7% of youth display a serious gambling problem. Jacobs
(1989) earlier reached a similar conclusion; he placed the rate of
youth problem gamblers at 4%–6%. In comparison, adults have
prevalence rates of pathological gambling between 1% and 3%
(American Psychiatric Association, 1994). The differences in these
rates between adults and youth may reflect true differences in
prevalence rates or differences in definitions and measures of
problem and pathological gambling (National Research Council,
Prospective studies have described changes in adult problem
gambling over time (e.g., Slutske, Jackson, & Sher, 2003; Volberg,
1994), and one study (Winters, Stinchfield, Botzet, & Anderson,
2002; Winters, Stinchfield, & Kim, 1995) has described changes in
adolescent problem gambling. The Slutske et al. (2003) study
found that aggregate levels of problem gambling were fairly stable
but that on an individual level, problem gambling was relatively
transitory and episodic. Shaffer and Hall (2002) also observed that
many frequent gamblers who meet criteria for at-risk gambling
return to nonproblem gambling rather than progress to problem
gambling. The Winters et al. (2002) study, which was based on the
results of a three-wave study of 305 participants that spanned
midadolescence to young adulthood, reported two main findings:
(a) The prevalence at Wave 3 (young adulthood) of problem
gambling, defined as 4? on the South Oaks Gambling Screen
(SOGS; Lesieur & Blume, 1987), remained unchanged from the
previous two waves (adolescence), and (b) there was a significant
increase from adolescence to young adulthood of at-risk gambling,
defined as a SOGS score of 2–3. However, individual-level
changes in gambling groups (at-risk and problem) across time
were not reported in the study. Thus, it is not known how patterns
of gambling severity changed over time or the extent to which new
nonproblem cases (i.e., at-risk gamblers or problem gamblers)
emerged in young adulthood.
The present study extends our earlier aggregate-level analysis
(Winters et al., 2002) by describing developmental problem gam-
bling groups. Specifically, we characterize the frequency of four
mutually exclusive groups on the basis of three waves of data: (a)
resistance from at-risk or problem gambling, (b) persistence of
at-risk or problem cases, (c) desistance to less severe gambling,
and (d) incidence of new at-risk or problem cases. The analyses are
based on gambling data collected across an 8-year interval span-
ning the adolescent to young adult years (ages 16–24).
As described elsewhere (Winters et al., 2002), the longitudinal cohort
consists of 305 young adults who received all three assessments (Time 1
[T1] in 1992, Time 2 [T2] in 1994, and Time 3 [T3] in 1997–1998). This
sample represents 87% of the individuals eligible (N ? 350) for all three
assessments. Nonparticipants were comparable to participants on all de-
mographic and T1 gambling variables except that nonparticipants were
Ken C. Winters, Randy D. Stinchfield, and Andria Botzet, Department
of Psychiatry, University of Minnesota, Twin Cities Campus; Wendy S.
Slutske, Department of Psychological Sciences, University of Missouri—
Funding for this study was provided by the National Institute on Re-
sponsible Gaming and the State of Minnesota Department of Human
Services to Ken C. Winters.
Correspondence concerning this article should be addressed to Ken C.
Winters, Department of Psychiatry, University of Minnesota, F282/2A
West, 2450 Riverside Avenue, Minneapolis, MN 55454. E-mail:
Psychology of Addictive Behaviors
2005, Vol. 19, No. 1, 104–107
Copyright 2005 by the Educational Publishing Foundation
0893-164X/05/$12.00 DOI: 10.1037/0893-164X.19.1.104
slightly older than participants (p ? .08). The background characteristics
of participants are as follows: Mean ages were 16.0, 17.6, and 23.8,
respectively; 51% were male; 96% were White; 95% had a high school
degree (at T3); and 86% resided in Minnesota (at T3).
The sample was administered comparable structured telephone inter-
views at each time point. The interview provided measures of the following
youth gambling variables: prior-year gambling frequency for 11 activities,
prior-year signs and symptoms of gambling-related problems (South Oaks
Gambling Screen—Revised Adolescent [SOGS–RA; Winters, Stinchfield,
& Fulkerson, 1993] at T1 and T2; SOGS [Lesieur & Blume, 1987] at T3),
grade of first gambling experience, prior year alcohol and other drug use,
mental health status, school achievement, delinquent behavior, and parental
history of gambling behavior. The SOGS–RA and SOGS each contain a set
of problem severity items reflecting criteria for pathological gambling
similar to those in the Diagnostic and Statistical Manual of Mental Dis-
orders (American Psychiatric Association, 1994), such as loss of control,
preoccupation, and negative consequences associated with gambling in-
volvement (e.g., “Have you felt like you would like to stop gambling but
didn’t think you could?” and “Have people criticized your gambling?”).
The adult SOGS has 20 problem severity items (Lesieur & Blume, 1987),
whereas the adolescent SOGS–RA, which was developmentally adjusted
for use with youth, has 12 problem severity items (Winters et al., 1993).
The 8 SOGS items that are not a part of the SOGS–RA pertain to various
sources from which the gambler may have borrowed money to finance his
or her gambling habit. To minimize a confound with the data analysis, we
based group classification of problem severity status (described below) on
the 12 items that were identical in the SOGS–RA and SOGS.
Problem Severity Groups
No problem gambling (in the prior year) was defined as a score of 0 or
1 on the 12-item SOGS–RA/SOGS. A nongambler for a given prior year
would automatically receive a score of 0. At-risk gambling (in the prior
year) was defined as a score of 2 or 3 on the SOGS–RA/SOGS. Problem
gambling (in the prior year) was defined as a score of 4 or more on the
We recognize that there is a lack of consensus in the gambling research
literature as to how to best categorically define adolescent gambling
problem severity (Derevensky, Gupta, & Winters, 2003; Shaffer & Hall,
1996). However, given the lack of a benchmark standard for defining youth
gambling problem severity, and given that we used the present definition
of gambling severity groups in our prior prospective reports (Winters et al.,
1995, 2002), we chose to retain this grouping strategy.
Developmental Gambling Groups
We identified developmental gambling groups using a procedure similar
to the one used by Slutske et al. (2003) in their adult longitudinal study.
Membership in a gambling group was created for each participant by
characterizing his or her gambling status at each of the three waves (T1,
T2, and T3) with the following three codes: N (no problem gambling), A
(at-risk gambling), or P (problem gambling). First, second, and third letters
in the series represented T1, T2, and T3 status, respectively. The number
of different possible three-letter subject codes or pathways is 27 (33). The
three-letter codes were further grouped into these four mutually exclusive
developmental gambling groups: stable, no-problem gambling at all three
waves (resistors); stable at-risk or problem gambling at all three waves
(persistors); change from either at-risk or problem gambling to no problem
gambling without a return to at-risk or problem gambling (desistors); and
new incidence cases, that is, no problem gambling at the first wave and
at-risk or problem gambling at both the second and third waves, or no
problem gambling at both the first and second waves and at-risk or problem
gambling at the third wave. Only six of the letter codes (8 participants)
could not be placed into one of these four groups.
As described in Winters et al. (2002), participants were administered an
interview over the telephone at each time point. Interviewers were well-
trained undergraduate or graduate research assistants. A target case was
considered unreachable if contact and consent could not be obtained after
20 callbacks, spread over a 4-week period. For minors (relevant at T1 and
T2), parental consent was required. We screened data for both missing
cases and outlier scores on all study measures.
Table 1 provides a list of the letter codes associated with the
target gambling groups (resistors, persistors, desistors, new inci-
dence cases, and other) and group frequencies. The data reveal
Frequencies of Three-Letter Codes and Developmental
Gambling Groups (N ? 305)
Resistors: N N N
A A A
A A P
A P A
A P P
P A A
P A P
P P A
P P P
A N N
A A N
A P N
P A N
P P N
P N N
New incidence cases
N N A
N N P
N A P
N P P
N P A
N A A
N A N
N P N
A N A
A N P
P N A
P N P
gambling); A ? at-risk gambling (2 or 3 on the SOGS–RA/SOGS); P ?
problem gambling (4 or more on the SOGS–RA/SOGS). SOGS–RA ?
South Oaks Gambling Screen—Revised Adolescent; SOGS ? South Oaks
N ? no problem gambling (0 or 1 on the SOGS–RA/SOGS, or no
several patterns. The majority of cases (60%) revealed a stable
resistor pattern (NNN); that is, no problem gambling occurred at
all three waves. No problem gambling typically occurred in the
presence of some gambling involvement although rarely at a
regular level (i.e., weekly or daily frequency of at least one game).
Among these 182 resistors, only 29 (16%) reported no recent
gambling at all three data points, and only 6 (3%) reported no
recent gambling for two data waves. On the other hand, no resis-
tors were regular gamblers at two or three data waves, and only 3
resistors (2%) reported regular gambling at one data wave.
The least frequent target pattern observed was the persistors.
Maintaining either at-risk or problem gambling was observed in
only 4% of cases. Among the persistors, only 2 participants (18%)
maintained problem gambling over the course of the study (PPP).
Both of them were regular gamblers at all data waves. However,
29% (7 out of 24 cases) of individuals who reported problem
gambling (P) at any single wave reported at-risk or problem
gambling at both other waves. It was relatively unusual for a
participant to report problem gambling at one wave and no prob-
lem gambling at the other two waves: 3 participants revealed the
NPN letter code, 1 participant had the NNP letter code, and no
participants had the PNN letter code. Among these 4 participants
with a single P letter code, it was always the case that the data
wave associated with problem gambling was linked with regular
gambling and the data wave associated with no problem gambling
was never linked with regular gambling.
Thirteen percent of the participants (n ? 41) exhibited the
desistor pattern. Nearly all desistors had three-letter codes of either
ANN or AAN (38 of 41). Thus, it was rare for desistors to include
problem gambling at any of the three study waves. The other 3
desistors had letter codes of PAN (2) or ANP (1); thus, there were
no desistors with letter codes of PNN or PPN. New incidence cases
were relatively frequent. Twenty-one percent of cases revealed an
at-risk or a problem gambling status at either the second or third
wave following no problem gambling at the first wave. Nearly two
thirds of new incidence cases (45 of 63) had the NNA letter code,
and another 9 individuals revealed an NAA letter code. New
incidence cases were more often characterized by at-risk gambling
status (NNA, NAA) than by problem gambling status (NNP, NAP,
NPP) (17% and 2%, respectively).
The 8 cases in the other category were NAN (3), NPN (3), and
ANP (2). The NAN and NPN patterns combine notions of both
incidence (at the second data wave) and desistance (a return to no
problem gambling at the third data wave).
This study provides a more detailed picture of gambling behav-
ior pathways than prior publications of our longitudinal results. In
light of the finding that resistors constituted the most prevalent
pathway, coupled with the result that nearly all of these individuals
had engaged in some form of gambling at one or more waves (and
77% reported prior 12 months gambling at each wave), the data
further support the argument we advanced in our 2002 report
(Winters et al., 2002)—that is, gambling involvement by young
people does not reliably contribute to at-risk or problem gambling.
The developmental pathway analysis also clarified what accounted
for the significant increase at young adulthood of at-risk gambling.
We observed 45 cases with an NNA pattern, which was the second
most frequent pattern, yet only 9 cases showed an NAA pattern.
Thus, young adulthood seems to be a particularly important age
period, when gambling-related problems emerge in the form of
Another finding was that early problem gambling, although rare,
was moderately associated with later problem gambling. Among
the 7 problem gamblers at T1, 4 (57%) were problem gamblers at
either or both T2 and T3, and 6 of 16 (38%) T2 problem gamblers
were problem gamblers at T3. On the other hand, early at-risk
gambling was not a common harbinger of later problem gambling.
At-risk gambling likely emerged first during young adulthood
rather than preceding the later emergence of problem gambling.
Among the 47 at-risk gamblers at T1, only 5 (11%) were problem
gamblers at either or both T2 and T3; among the 36 at-risk
gamblers at T2, only 2 (6%) were problem gamblers at T3. Even
among the 20 cases that revealed at-risk gambling at both T1 and
T2, only 1 individual revealed problem gambling at T3. Yet 70%
of T3 at-risk gamblers were non-problem gamblers at both T1 and
T2. Of course, a more extended prospective examination of the
data is needed to see the extent that at-risk gambling during youth
is a reliable predictor of later adult problem gambling.
The findings must be viewed in light of study limitations. First,
all data are solely based on self-report. Second, whereas our
sample of 305 participants represents cases with data at all three
waves, the eligible sample for the longitudinal study was 350.
Whereas the attrition analysis indicated that cases lost to attrition
(13%) did not differ on any gambling or psychosocial functioning
variables except age as compared with the retained cases, it is
important to keep in mind that our study suffers from some
attrition. Third, as already noted, our grouping strategy for the
designations of no problem, at-risk, and problem gambling is open
for debate. Fourth, given that the adult SOGS was administered at
T3 whereas the adolescent SOGS–RA was administered at T1 and
T2, it is possible that a measurement confound occurred. Finally,
the sample is limited to Minnesota youth, and the sample size is
relatively small; thus, one needs to be cautious when generalizing
the findings to youth in general.
American Psychiatric Association. (1994). Diagnostic and statistical man-
ual of mental disorders (4th ed.). Washington, DC: Author.
Chambers, R. A., Taylor, J. R., & Potenza, M. N. (2003). Developmental
neurocircuitry of motivation in adolescence: A critical period of addic-
tion vulnerability. American Journal of Psychiatry, 160, 1041–1052.
Derevensky, J., Gupta, R., & Winters, K. C. (2003). Prevalence rates of
youth gambling: Are the current rates inflated? Journal of Gambling
Studies, 19, 405–426.
Jacobs, D. F. (1989). Illegal and undocumented: A review of teenage
gamblers in America. In H. J. Shaffer, S. A. Stein, B. Gambino, & T. N.
Cummings (Eds.), Compulsive gambling: Theory, research and practice
(pp. 35–64). Lexington, MA: Lexington Books.
Lesieur, H. R., & Blume, S. B. (1987). The South Oaks Gambling Screen
(SOGS): A new instrument for the identification of pathological gam-
blers. American Journal of Psychiatry, 144, 1184–1188.
National Research Council. (1999). Pathological gambling: A critical
review. Washington, DC: National Academy Press.
Shaffer, H. J., & Hall, M. N. (1996). Estimating the prevalence of adoles-
cent gambling disorders: A quantitative synthesis and guide toward
standard gambling nomenclature. Journal of Gambling Studies, 12,
Shaffer, H. J., & Hall, M. N. (2002). The natural history of gambling and Download full-text
drinking problems among casino employees. Journal of Social Psychol-
ogy, 142, 405–424.
Slutske, W. S., Jackson, K. M., & Sher, K. J. (2003). The natural history
of problem gambling from age 18 to 29. Journal of Abnormal Psychol-
ogy, 112, 263–274.
Stinchfield, R. D., & Winters, K. C. (1998, March). Adolescent gambling:
A review of prevalence, risk factors and health implications. Annals of
the American Academy of Political and Social Science, 556, 172–185.
Volberg, R. A. (1994). The prevalence and demographics of pathological
gamblers: Implications for public health. American Journal of Public
Health, 84, 237–241.
Winters, K. C., & Anderson, N. (2000). Gambling involvement and drug
use among adolescents. Journal of Gambling Studies, 16, 175–198.
Winters, K. C., Stinchfield, R. D., Botzet, A., & Anderson, N. (2002).
Prospective study of youth gambling behaviors. Psychology of Addictive
Behaviors, 16, 3–9.
Winters, K. C., Stinchfield, R. D., & Fulkerson, J. (1993). Toward the
development of an adolescent gambling problem severity scale. Journal
of Gambling Studies, 9, 63–84.
Winters, K. C., Stinchfield, R. D., & Kim, L. (1995). Monitoring adoles-
cent gambling in Minnesota. Journal of Gambling Studies, 11, 165–183.
Received August 19, 2003
Revision received December 15, 2003
Accepted December 22, 2003 ?