A Functional Analytic Approach to Understanding Disordered Gambling.
Problem Gambling Severity Index, PGSI; Ferris & Wynne, 2001) to identifying the functions sustaining the behavior (i.e., Gambling Functional Assessment, GFA; Dixon & Johnson, 2007; Gambling Functional Assessment--Revised; GFA-R; Weatherly, Miller, & Terrell, 2011). Function-based assessments are rather unique in their quest to discover what types of consequences sustain gambling (e.g., social, escape, neurological), instead of relying solely on corollary environmental events that may occasion gambling (e.g., advertisements, access to games, prior losing/winning history). The GFA (Dixon & Johnson, 2007) was the first functional behavioral assessment developed to determine which of four possible consequences maintains a person's gambling behaviors. The GFA was similar to other behavioral assessments designed to measure aberrant behaviors, particularly self-injurious behaviors (Durand & Crimmins, 1988), which have been used in clinical psychology and by behavior analysts. The 20-item GFA included questions regarding the contributions of social attention (e.g., enjoyment of interacting with peers), psychological/physical escape (e.g., ability to forget about stress at home, leave a troubled work environment), access to tangible rewards (e.g., money, comps, vouchers), and sensory (e.g., feeling a rush or buzz). For example, a question targeting an escape function stated, "I often gamble when I feel stressed or anxious," while a question targeting a sensory function stated, "I like the sounds, the lights, and the excitement that often go along with gambling." Questions are rated on a 7-point Likert scale, from never (0) to always (6). Afterward, each function is summed, and the highest category is identified as the most likely maintaining function of gambling behavior.
In research that followed, the utility of the GFA appeared statistically promising. For example, Miller, Meier, Muehlenkamp, and Weatherly (2009) tested the construct validity of the GFA with 949 undergraduate university students, all considered to be non-disordered gamblers. A factor analysis yielded two factor loadings consistently across participants, which reported to be representative of positive and negative reinforcement (Miller et al., 2009). Escape items were reported to be representative of negative reinforcement, while items related to tangible, sensory, and attention were reported as positive reinforcement. Subsequently, Weatherly et al., (2011) revised the items on the GFA to construct the Gambling Functional Assessment--Revised (GFA-R) that included only negative and positive functional categories.
To test the factor structure of the GFA-R (Weatherly et al., 2011), 728 undergraduate university students completed the 22-item assessment, and responses were subjected to explanatory and confirmatory factor analyses. A principal component analysis suggested a two-factor solution accounting for 53.86% of the variance in participant responses. A total of six items were removed due to inaccurate loading, leaving eight items representing positive reinforcement and eight items representing negative reinforcement. A confirmatory structural equation model was used to asses multiple fit indices and suggested a moderate correlation (r = 0.043) between the two factors with "all items loading significantly onto their respective factor" (Weatherly et al., 2011, p. 562). To date, the GFA-R has been found to have good test-retest (Weatherly, Miller, Montes, & Rost, 2012), internal consistency and construct validity (Weatherly, Terrell, & Bogenreif, 2013), and temporal reliability (Weatherly, Miller, Montes, & Rost, 2012). Further, the GFA-R has shown similar outcomes with populations in both the USA, the UK (Weatherly, Dymond, Samuels, Austin, & Terrell, 2014), and Japan (Weatherly, Aoyama, Terrall, & Berry, 2014).
Although research collected on the GFA-R has been shown to have good reliability and validity, the assessment endorses a two-factor model to account for the possible functions maintaining gambling behavior (positive and negative reinforcement). This approach deviates from a common theoretical approach in behavior analysis, which accounts for separate and distinct environmental events that can be categorized as positive reinforcement (social attention, tangible items/foods, and sensory experiences). Behavioral function is idiosyncratic and reliant on both antecedent and consequential variables beyond delineating reinforcement types (Iwata, Dorsey, Slifer, Bauman, & Richman, 1994). Further, positive and negative reinforcement are strictly categorical concepts that identify the addition or removal of stimulus events. Identification of general concepts (positive or negative reinforcement) may not assist clinicians in identifying specific classes that maintain disordered gambling. Further, in the decades of behavior analytic research on behavioral function, there have been no reported conceptualizations of challenging behavior analyzed solely by a description of positive or negative reinforcement. Function, as a construct, has always been reported with more precision and with details of the antecedent conditions, not just positive/negative reinforcement topographies. As a result, conceptually, the GFA-R identifies the topography of the consequences of gambling, namely that of reinforcement. While it is unknown if this approach as an indirect behavioral assessment has the critical external validity needed to be of value to a practicing clinician, a four-factor model is needed to adhere to a behavior analytic theoretical foundation.
Therefore, the purpose of the current study was to determine the extent to which a thoroughgoing functional assessment of gambling behaviors was possible using a four-factor model, as originally suggested by Dixon and Johnson. That, if we could construct an internally valid model containing four factors, there may be utility in developing a more psychometrically valid assessment instrument than that of Dixon and Johnson containing the same four functions. As well, if such a model could be developed that deviates from results reported by this and other research teams, this outcome could call for more precise research in this area, and potentially conducted by teams not invested in validating an assessment tool that they, themselves, created. To this end, 365 adults across the USA completed a demographic survey, the SOGS (gambling severity measure), and the original GFA as developed by Dixon and Johnson. The sample was randomly divided into two groups of respondents. The first sample (n =182) was used for the exploratory factor analysis in an attempt to identify items that would best assess the four functions maintaining gambling behavior (e.g., social attention, tangible, escape, and sensory). Items that cross-loaded on multiple factors or did not load onto any factors were discarded. The resulting model was tested with the second sample (N =183) through a confirmatory factor analysis using structural equation modeling. The goal of these analyses was to construct an empirically validated version of a behavioral assessment that contained four factors, to guide the development of future instruments that adhere to functional behavior analytic approaches to understanding disordered gambling.
Participants and Setting
Three hundred and sixty-five adults (45% female) were recruited from across the USA to participate in the study. The average age of the participants was 28.2 years (SD = 13.46).
Sixty-five percent of the total sample reported European American ethnicity (21% African American, 5% Latin American, 4% Asian American). About half of the sample had some college (46%) or a Bachelor's degree (19%), and 16% had a high school diploma.
All participants completed the study in the same location where they were recruited, in public spaces and university campuses across the USA. All participants received a chance to win a $25.00 gift card upon completion of the surveys, while some received extra course credit for completing the study. Both the Southern Illinois University Human Subjects Committee and the Saint Louis University Human Subjects Committee approved all study procedures.
Materials and Procedure
Participants completed the following surveys either on a computer or via paper-pencil: a demographic survey, the SOGS (Lesieur & Blume, 1987), and the GFA (Dixon & Johnson, 2007). Demographic questions included participant age, ethnicity, and education history. Participants were screened for potential gambling pathology using the SOGS (Lesieur & Blume, 1987). This 26-item assessment (20 of which are scored) includes questions as to the type of gambling typically engaged in, the frequency of gambling events, the severity of the amount of money gambled and borrowed from friends and family, and the subjective interpretation of the emotional aspects of gambling. Scores range from 0 to 20, where scores between 3 and 4 indicate at risk for disordered gambling, and scores of 5 or higher are characterized as indicators of disordered gambling. This assessment has good internal and convergent validity for both clinical and general populations (Gambino & Lesieur, 2006; Stinchfield, 2002; Stinchfield & Winters, 2001).
Participants were also assessed on potential maintaining variables of gambling behaviors using the GFA (Dixon & Johnson, 2007). The GFA is a 20-item assessment identifying four categories of maintaining reinforcers: removal from aversive tasks or states (escape), access to reinforcers derived from social mediation (attention), access to reinforcing items/objects (tangible), and physiological reinforcers not socially mediated (sensory). Participants respond to items using a Likert scale based on frequency, from 0 (not at all) to 6 (always).
The order of the surveys was counterbalanced across participants to control for sequence effects, except for the demographic survey, which was always presented first. The criterion for inclusion in the study was simply being over the age of 18.
All respondents (n = 365) were randomly divided into two samples for the following analyses. The exploratory factor analysis (n = 182) was conducted first to identify the number of factors that best accounted for the obtained data, and to isolate assessment items for each function that loaded onto a common factor. The confirmatory analysis (n = 183) was then conducted to confirm the results obtained from the first analysis and to refine the four-factor model in the GFA to provide a good model fit for the data. This was done by eliminating assessment items that increased model error terms, and covarying error terms for select items in the same factor. As noted by Schreiber, Nora, Stage, Barlow, and King (2006), when assessment items are modified based on the results of a confirmatory factor analysis, the revised model should be considered preliminary and may serve as the basis for further exploratory factor analyses if the assessment is subsequently revised. In the context of the present study, the results of the confirmatory factor analysis were intended to guide the development of new functional assessment instruments containing distinct functional categories for use with gamblers. Exploratory factor analyses (EFA) were conducted in IBM's SPSS version 19, while confirmatory factor analyses (CFA) were conducted using IBM's AMOS statistical software extension.
Exploratory Factor Analysis An initial EFA was conducted to identify the number of factors that best fit the obtained data, with component inclusion criteria of eigenvalues > 1. All participants were retained in this analysis, with the exception of participants who had incomplete demographic information or incomplete sections of the SOGS or GFA (n = 8). Responses from the 174 (age M = 30.5, SD = 14.7) participants were analyzed (for specific demographic information across groups, see Table 1). All skewness values were below an absolute value of 1 and kurtosis values were less than an absolute value of 1.5, which have been noted as the appropriate parameters for EFA and CFA (Curran, West, & Finch, 1996).
An exploratory principal component analysis was conducted with a varimax rotation to optimize the solution. The first analysis yielded three components with eigenvalues greater than 1, although only slightly for the third component (1.22). The shape of the scree plot was also suggestive of a three-factor solution, accounting for 63.3% of GFA item variance. Rotated item scores were then examined and ranked based on best fit in the component and variance score. The items and component matrix loading values for each identified factor are summarized in Table 2. Items were excluded based on the following criteria: (a) component matrix loading below 0.4, (b) component matrix loading across each of the three factors (i.e., matrix loading above 0.4), (c) highest component score at least 0.10 units higher than other component scores, and (d) component item ranking of 1 (see Costello & Osborne, 2005, p. 4 for discussion of similar item extraction criterion). For example, if item 1 was loaded at 0.32, 0.4, and 0.38 on components two, three, and four, item 1 would be thrown out of the confirmatory analysis. Likewise, if item 2 was loaded at 0.42 and 0.58 on components one and four, item 2 would be retained under component four, as 0.58 > 0.42. No items were eliminated from the analysis based on this elimination criterion. Analysis of the items in each component shows that component one consisted of the majority of sensory (n = 3) and escape (n = 5) items, component two consisted of the majority of attention items (n = 5), and component three consisted of the majority of tangible items (n = 3). Only the GFA items that measured the same function and loaded into the same component were retained, and included three sensory items (items 1, 5,19), five escape items (items 4, 7, 11, 14, 17), five attention items (items 3, 6, 8, 13, 16), and three tangible items (items 2, 9, 18). This new model with the 16 retained items is shown in Fig. 1.
Confirmatory Factor Analysis To validate a four-factor structure, a Confirmatory Factor Analysis (CFA) was used to determine if the identified model would fit the second data set. Participants who failed to complete the demographic data, SOGS, or items in the GFA were excluded from the analysis, and only participants with a SOGS score > 0 were retained to ensure that the final model was predictive of the gambling function of actual gamblers (n = 55). The CFA was then conducted with the remaining 128 participants (age M =27.8, SD= 13.6) and the model fit was determined by using absolute (e.g., Goodness-of-Fit Index), Standardized Root Mean Squared Residual, and incremental fit statistics (see also Miller et al., 2009). Figure 2 illustrates the model as proposed in the EFA with associated error terms. The four factors derived from the exploratory analysis, tangible ([F.sub.1]), escape ([F.sub.2]), sensory (F), and attention (F4), served as latent variables for the confirmatory analysis. A first CFA was conducted to evaluate the model fit with all items from the EFA retained, and without covarying associated error terms. Fit indices suggest this model failed to provide a good fit to the data, [chi square] (98)= 193.4, p < 0.05 (in CFA, a good fit requires no significant difference between the proposed model and the obtained data represented by p > 0.05 (Barrett, 2007)), RMSEA = 0.09, and GFI = 0.84.
As such, we systematically eliminated items from the model that produced low factor loadings onto the factor (i.e., factor loadings <0.8) or covaried with two or more items from another factor. Items 6 (attention), 8 (attention), 13 (attention), 14 (escape), and 18 (tangible) were eliminated from the model. In addition, we covaried error terms for items in the same functional category that appeared to covary, resulting in covariation of the error terms for items 4 and 11 (escape). The CFA was then conducted with the remaining 10 items and covaried error terms, which is displayed graphically in Fig. 3. Fit indices suggested that this model produced a good or acceptable fit for the data [chi square] (28) = 38.7, p >0.05, RMSEA = 0.06, and GFI = 0.94 (see also Fig. 4 for a scree plot representing eigenvalues). Remaining functional categories and their respective mean, standard deviation, skewness, and kurtosis outcomes across items are represented in Table 3. According to Hooper, Coughlan, and Mullen (2008), RMSEA values < 0.07 represent a good fit and a traditional omnibus cutoff point for GFI has been > 0.90. The remaining items were renumbered to establish the GFA-II (see Fig. 5 for an overview) to guide the development of future instruments.
A final set of analyses were conducted to evaluate the correlation between participant scores on the GFA, the 10-item revised GFA model (GFA-II), and SOGS total scores. All participants who had completed the demographics information, the GFA, and the SOGS were retained in the analysis (n = 360, age M =28, age SD= 13.8). A correlation matrix is shown in Table 4, and a scatterplot of the relationship between both versions of the GFA and SOGS results is shown in Fig. 6. The results show a strong, positive correlation between the full GFA and the GFA total score derived from the new model (r = 0.96, p < 0.01). The results also suggest strong, significant correlations between participant scores on the SOGS and both versions of the GFA, where the reduced 10-item version produced a slightly stronger correlation (r = 0.79, p < 0.01). A linear regression analysis was additionally conducted for both the full GFA and reduced GFA to evaluate their relationship with SOGS, producing [R.sup.2] values of 0.49 and 0.56, respectively. Both of these analyses therefore suggest that the reduced 10-item model may be more predictive of the problem or disordered gambling.
The purpose of the current study was to determine the extent to which a four-factor functional assessment of gambling could be developed using questions contained in the original GFA developed by Dixon and Johnson (2007). Three hundred and sixty-five adult gamblers were recruited to complete the SOGS and the GFA (n = 360 retained for subsequent analyses). Results from the exploratory analysis suggested a three-factor model, accounting for 63.3% of GFA item response variance. After removing four original GFA items, 16 items were subjected to a confirmatory analysis, which failed to provide a good fit to the data. Therefore, after eliminating low loading factor items and items that covaried with other items, a 10-item four-factor model was subjected to a confirmatory analysis. The items were subsequently reorganized into the GFA-II (see also Fig. 4). These results suggest a behavior analytic approach to understanding the controlling factors for problem gambling that mirrors those that have been reported in the behavior analytic literature over the past three decades (i.e., Beavers, Iwata, & Lerman, 2013; Iwata et al., 1994; Hanley, Iwata, & McCord, 2003).
The current study supports the use ofa four-factor model that corresponds with traditional behavior analytic perspectives of maintaining functions of challenging behavior, when identifying conditions that control gambling. With the use of a four-factor model, the controlling conditions can be systematically narrowed to specify types of reinforcers that are positive (e.g., access to tangibles or socially mediated), negative (e.g., escape or avoidance from aversive stimulation), or sensory/automatically reinforcing (e.g., physiological or non-socially mediated) (see also Iwata, Pace, Kalsher, Cowdery, & Cataldo, 1990). This variation is important theoretically for identifying sources of reinforcement believed to maintain problem or disordered gambling behavior, rather than merely referencing the broader topographical reinforcing classes suggested by prior GFA researchers and assessed in updated and revised instruments (Weatherly et al., 2011). By narrowing a specific category of reinforcers, clinicians may be more likely to provide targeted functional interventions, where functional interventions for problem gamblers have been shown to be effective at both a behavioral (e.g., Guercio, Johnson, & Dixon, 2012) and neurological level (e.g., Dixon, Wilson, & Habib, 2016).
Without the precision of function, treatment approaches may not compete with environmental factors, and therefore remain minimally effective. Further, while it is arguable that sensory or automatic reinforcement can be positively or negatively reinforcing (i.e., behavior may either increase to receive more of the sensory input, or behavior that removes or avoids the sensory input may increase), it is critical that a functional assessment measure has a categorical distinction between reinforcement that is not delivered via social mediation (Iwata et al., 1990). Following the same logic, it would be just as critical to distinguish between reinforcement delivered via social mediation and reinforcement delivered from food items, objects, or people in the environment. Thus, adopting an assessment that includes four factors is parsimonious with Iwata et al.'s (1990) rationale for the inclusion of general contingency classes rather than broad classes that include "stimuli that apply in any given case" (p. 17). In addition to identification, the current study also showed positive relationships between total scores on the four-factor GFA-II model and total scores on the SOGS, an indicator of gambling proclivity and severity. Regression analyses found positive correlations between total GFA-II scores and total SOGS scores ([R.sup.2] = 0.55, p <0.001), suggesting that higher proclivity toward gambling leads to higher total scores on the GFA-II.
Perhaps the most important outcome of the current study, and in the context of prior research on the GFA, is the degree to which results reported by different research teams using the same 20-item GFA (e.g., Weatherly et al., 2011, the current study) have yielded different results. For example, Miller, Meier, Muehlenkamp, and Weatherly (Miller et al., 2009) found that participant response results were best described in a two-factor model, whereas, we found that a three-factor model best described the data in the EFA, and that a modified version of the GFA (i.e., GFA-II) could provide a good fit for the data within a four-factor model. Over the past decade, two different research teams have now developed three unique assessments from the same pool of 20 questions. In light of differing results, ascertaining the sources of these differences is essential in understanding the environmental conditions that cause and maintain disordered gambling, and in developing functional analytic treatment strategies that operate within a behavioral framework. One potential explanation for the observed differences could be demographic inconsistencies in the participant samples, in terms of both age and education (Miller et al. utilized an exclusive sample of college students), as well as regional differences. Another could be differences in the purpose of the investigation. The current attempt was aimed at developing an internally valid four-function model that is consistent with behavioral theory and approaches to resolving challenging behavior. The degree to which results reported by various research teams differ undoubtedly calls for more precise research on the functional events that maintain gambling behavior, and perhaps conducted by research teams that have not participated in the development of a given assessment, to further distance any potential for conflict of interest in instrument analysis.
While these results are promising, they should be taken in light of the following limitations. The relatively small total sample size of the present study may be a limitation, given the importance of large sample sizes in exploratory factor analysis research. However, the recommended sample size to factor item ratio for our current investigation is actually within a range of 10 to 20 people per measured variable (as discussed in Thompson, 2004). Under this criterion, with a four-factor model, 80 people per analysis would have been sufficient. Further, prior researchers tend to consider the value of the structure coefficients > 0.6 regardless of the size of the sample (Gorsuch, 1990; as discussed in Thompson, 2004, p. 24) as a metric of sample appropriateness.
Although not a limitation of the current study, a limitation of this line of research in general is that despite the development psychometrically sound (Miller et al., 2009; Weatherly et al., 2011) and conceptually systematic (Dixon & Johnson, 2007) gambling functional assessments, no research to date has evaluated the effectiveness of a behavioral intervention that directly targets the function identified in the assessment. A key feature in determining the validity of existing functional behavioral assessments is that the results lead to effective treatments in application with clients that the instrument is intended to serve. Therefore, ultimately in a pragmatic science, the true utility of any gambling functional assessment, utilizing either a two-factor or four-factor model, will be empirically supported effective application of applied technologies developed from assessment results.
Specifically, researchers may consider the extent to which the GFA instruments assist with clinical case conceptualization, and if the measure is sensitive enough to accurately identify controlling mechanisms of disordered gambling. Future research should also consider comparing structured functional assessment interviews with functional assessment outcomes. In this way, clinicians could determine if a more refined and targeted functional assessment, like the GFA-II, can identify controlling variables of their gambling clients.
The current study is a first step in identifying the utility of a four-factor model, aligned with a behavior analytic conceptualization. Current researchers outside of the behavior analytic tradition tend to believe the best approach to treatment of disordered gambling is traditional standardized cognitive-behavioral counseling or psychotherapy (Ladouceur, Lachance, & Fournier, 2009), yet the current work speaks to the importance of developing individualized treatment plans, particularly in the identification of maintaining functional relations. Further research is clearly needed to statistically identify the construct validity, test-retest reliability, and clinical significance of GFA instruments when in use in developing treatment of clients with gambling problems. Furthermore, it remains to be seen if this indirect self-report measure of functional relations will match observed functional relations when clinicians engage in more observational or experimental functional analytic analyses, and when conducting direct behavioral assessments with disordered gamblers.
Published online: 18 April 2018
Compliance with Ethical Standards
Conflict of Interest The authors declare that they have no conflict of interest.
Ethical Approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed Consent Informed consent was obtained from all individual participants included in the study.
Barrett, P. (2007). Structural equation modelling: adjudging model fit. Personality and Individual Differences, 42, 815-824. https://doi.org/ 10.1016/j.paid.2006.09.018.
Beavers, G. A., Iwata, B. A., & Lerman, D. C. (2013). Thirty years of research on the functional analysis of problem behavior. Journal of Applied Behavior Analysis, 46, 1-21. https://doi.org/10.1002/jaba.30.
Blaszcznski, A., & Nower, L. (2002). A pathways model of problem and pathological gambling. Addiction, 97, 487-499. https://doi.org/10. 1046/j.1360-0443.2002.00015.x.
Costello, A., & Osborne, J. (2005). Best practices in exploratory factor analysis: four recommendations for getting the most from your analysis. Practical Assessment, Research and Evaluation, 10, 1-9.
Curran, P. J., West, S. G., & Finch, J. F. (1996). The robustness of test statistics to normality and specification error in confirmatory factor analysis. Psychological Methods, 1, 16-29.
Dixon, M. R., & Johnson, T. E. (2007). The gambling functional assessment (GFA): an assessment device for identification of the maintaining variables of pathological gambling. Analysis of Gambling Behavior, 1, 44-49.
Dixon, M. R., Wilson, A. N., & Habib, R. (2016). Neurological evidence of acceptance and commitment therapy effectiveness in college-age gamblers. Journal of Contextual Behavioral Science, 5, 80-88. https://doi.org/10.1016/jjcbs.2016.04.004.
Durand, V. M., & Crimmins, D. B. (1988). Identifying the variables maintaining self-injurious behavior. Journal of Autism and Developmental Disabilities, 18, 99-117. https://doi.org/10.1007/ BF02211821.
Ferris, J., & Wynne, H. (2001). The Canadian problem gambling index. Ottawa, ON: Canadian Centre on Substance Abuse.
Gambino, B., & Lesieur, H. (2006). The south oaks gambling screen (SOGS): a rebuttal to critics. Journal of Gambling Issues, 17, 116. https://doi.org/10.4309/jgi.2006.17.10.
Gorsuch, R. L. (1990). Common factor analysis versus component analysis: Some well and little known facts. Multivariate Behavioral Research, 25, 33-39.
Guercio, J. M., Johnson, T., & Dixon, M. R. (2012). Behavioral treatment for pathological gambling in persons with acquired brain injury. Journal of Applied Behavior Analysis, 45, 485-495. https://doi. org/10.1901/jaba.2012.45-485.
Hanley, G. P., Iwata, B. A., & McCord, B. E. (2003). Functional analysis of problem behavior: a review. Journal of Applied Behavior Analysis, 36, 147-185. https://doi.org/10.1901/jaba.2003.36-147.
Hooper, D., Coughlan, J., & Mullen, M. (2008). Structural equation modelling: Guidelines for determining model fit. Electronic Journal of Business Research Methods, 6(1), 53-60.
Iwata, B. A., Dorsey, M. F., Slifer, K. J., Bauman, K. E., & Richman, G. S. (1994). Toward a functional analysis of self-injury. Journal of Applied Behavior Analysis, 27, 197-209. https://doi.org/10.1901/ jaba.1994.27-197.
Iwata, B. A., Pace, G. M., Kalsher, M. J., Cowdery, G. E., & Cataldo, M. F. (1990). Experimental analysis and extinction of self-injurious escape behavior. Journal of Applied Behavior Analysis, 23, 11-27. https://doi.org/10.1901/jaba. 1990.23-11.
Johnson, E. E., Hamer, R. M., & Nora, R. M. (1998). The lie/bet questionnaire for screening pathological gamblers: a follow-up study. Psychological Reports, 83, 1219-1224.
Kim, S. W., Grant, J. E., Adson, D. E., & Young, C. S. (2001). Double-blind naltrexone and placebo comparison study in the treatment of pathological gambling. Biological Psychiatry, 49, 914-921.
Ladouceur, R., Lachance, S., & Fournier, P. M. (2009). Is control a viable goal in the treatment of pathological gambling? Behaviour Research and Therapy, 47, 189-197. https://doi.org/10.1016Zj.brat.2008.11.004.
Lesieur, H. R., & Blume, S. B. (1987). The south oaks gambling screen (the SOGS): a new instrument for the identification of pathological gamblers. American Journal of Psychiatry, 144, 1184-1188.
Miller, J. C., Meier, E., Muchlenkamp, J., & Weatherly, J. N. (2009). Testing the construct validity of Dixon and Johnson's (2007) Gambling Functional Assessment. Behavior Modification, 33, 156-174. https://doi.org/10.1177/0145445508320927.
Schreiber, J. B., Nora, A., Stage, F. K., Barlow, E. A., & King, J. (2006). Reporting structural equation modeling and confirmatory factor analysis results: A review. The Journal of Educational Research, 99, 323-338. https://doi.org/10.3200/JOER.99.6323-338.
Shaffer, H. J., & Kidman, R. (Eds.). (2003). Shifting perspectives on gambling and addiction. Journal of Gambling Studies, 19, 1-6. https://doi.org/10.1023/A:1021267028254.
Stinchfield, R. (2002). Reliability, validity, and classification accuracy of the south oaks gambling screen (SOGS). Addictive Behaviors, 27, 1-19. https://doi.org/10.1016/S0306-4603(00)00158-1.
Stinchfield, R., & Winters, K. C. (2001). Outcome of Minnesota's gambling treatment programs. Journal of Gambling Studies, 17, 217245. https://doi.org/10.1023/A:1012268322509.
Thompson, B. (2004). Exploratory and confirmatory factor analysis. Washington, DC: American Psychological Association.
Weatherly, J. N., Aoyama, K., Terrell, H. K., & Berry, J. C. (2014). Comparing the Japanese version of the Gambling Functional Assessment-Revised to an American sample. Journal of Gambling Issues, 29, 1-20. https://doi.org/10.4309/jgi.2014.29.4. Weatherly, J. N., Dymond, S., Samuels, L., Austin, J. L., & Terrell, H. K. (2014). Validating the gambling functional assessment--revised in a United Kingdom sample. Journal of Gambling Studies, 30, 335347. https://doi.org/10.1007/s10899-012-9354-5.
Weatherly, J. N., Terrell, H. K., & Bogenreif, D. (2013). Testing the internal consistency and construct validity of the gambling functional assessment-revised in a general-population sample. Modern Behavioral Science, 1, 37-48.
Weatherly, J. N., Miller, J. C., Montes, K. S., & Rost, C. (2012). Assessing the reliability of the gambling functional assessment: revised. Journal of Gambling Studies, 28, 217-223. https://doi.org/10. 1007/s10899-011-9275-8.
Weatherly, J. N., Miller, J. C., & Terrell, H. K. (2011). Testing the construct validity of the gambling functional assessment-revised. Behavior Modification, 35, 553-569.
Mark R. Dixon (1) * Alyssa N. Wilson (2) * Jordan Belisle (3) * James B. Schreiber (4)
([mail]) Mark R. Dixon firstname.lastname@example.org
(1) Behavior Analysis and Therapy Program, Rehabilitation Institute, Southern Illiniois University, Carbondale, IL 62901, USA
(2) St. Louis University, St Louis, MO, USA
(3) Southern Illiniois University, Carbondale, IL, USA
(4) Duquesne University, Pittsburgh, PA, USA
Caption: Fig. 1 Items retained from the EFA and corresponding functions as identified in the GFA
Caption: Fig. 2 Results of the initial CFA with 16 items retained based on the results of the initial EFA
Caption: Fig. 3 Results of the secondary CFA with 10 items retained based on the results of the initial CFA
Caption: Fig. 4 Scree plot of the confirmatory factor analysis
Caption: Fig. 6 Scatter plot of the relationship between SOGS total score and participant scores on the full GFA model (top panel) and the reduced GFA model (bottom panel) based on the EFA and CFA results. A linear regression was fit to both models
Table 1 Demographic information across exploratory and confirmatory analyses Demographics Exploratory Confirmatory analyses analyses % of sample % of sample Gender Male 55.2 65.6 Female 44.8 34.4 Race African 18.4 25.0 American European 69.5 60.1 American Latino 5.7 6.3 American Asian 4.5 1.6 American Other 0 0 Education Some high 5.7 4.0 school HS/GED 18.4 17.2 Some college, 44.8 47.7 no degree Associates 8.0 6.3 degree Bachelor's 17.8 18.0 degree Master's 4.0 6.3 degree Professional or 1.1 .8 doctorate degree Not answered 0 0 SOGS score 0-2 47.1 25.0 3-4 28.1 21.9 5-10 15.5 35.9 11-15 8.6 14.0 16-20 1.1 3.1 Average SOGS 4.0 6.0 score Table 2 Component item loadings and function identified in the GFA Component 1 Component 2 Component 3 Item Function Item Function Item Function 1 Sensory 3 Attention 2 Tangible 4 Escape 6 Attention 9 Tangible 5 Sensory 8 Attention 10 Sensory 7 Escape 12 Tangible 18 Tangible 11 Escape 13 Attention 14 Escape 15 Sensory 17 Escape 16 Attention 19 Sensory 20 Tangible Table 3 Rotated component matrix of items in exploratory factor analysis and descriptive statistics across variable examination GFA component Item number [R.sup.2] 1: Sensory + 1: I tend to gamble most frequently when 0.403 Escape there is nothing else going on or I have nothing better to do. 4: I often gamble after fighting with my 0.596 spouse or significant other. 5: I feel more alive when I am gambling 0.430 than when I am doing other types of activities. 7: I often gamble when I feel stressed or 0.646 anxious. 11: I gamble to get a break from work or 0.501 other difficult tasks. 14: I often gamble when I am feeling 0.598 depressed or sad. 17: If I have a hard day at work, I am 0.653 likely to gamble. 19: When I gamble, I am often unaware of 0.221 my surroundings. 2: Attention 3: I enjoy the social aspects of gambling 0.328 such as being with my friends or being around other people who are having a good time and cheering me on. 6: Even if I lose, I can always count on a 0.491 friend/loved one to help me through this difficult time. 8: After I gamble, I like to go out and 0.398 celebrate my winnings with others. 12: If it were not for the ability to win 0.590 a bunch of money, I would probably not gamble much at all. 13: I only gamble when my friends are 0.264 gambling with me. 15: I find myself feeling a rush, and 0.350 getting excited when I gamble. 16: After I gamble, I often find comfort 0.269 from other people to help me deal with my losses. 20: I gamble primarily for the money that I can win. 0.579 3: Tangible 2: I really enjoy the complementary perks 0.501 that come along with gambling, like free points, drinks, comp coupons, etc. 9: When I gamble, I like to accumulate 0.567 points at a casino so they will offer me incentives and bonuses. 10: I like the sounds, the lights, and the 0.412 excitement that often go along with gambling. 18: I gamble more often when I have been 0.635 offered complementary drinks, hotel rooms or other items. GFA component Item number [bar.x] 1: Sensory + 1: I tend to gamble most frequently when 1.95 Escape there is nothing else going on or I have nothing better to do. 4: I often gamble after fighting with my 0.63 spouse or significant other. 5: I feel more alive when I am gambling 1.65 than when I am doing other types of activities. 7: I often gamble when I feel stressed or 1.18 anxious. 11: I gamble to get a break from work or 1.49 other difficult tasks. 14: I often gamble when I am feeling 0.84 depressed or sad. 17: If I have a hard day at work, I am 0.94 likely to gamble. 19: When I gamble, I am often unaware of 1.21 my surroundings. 2: Attention 3: I enjoy the social aspects of gambling 3.33 such as being with my friends or being around other people who are having a good time and cheering me on. 6: Even if I lose, I can always count on a 2.30 friend/loved one to help me through this difficult time. 8: After I gamble, I like to go out and 2.41 celebrate my winnings with others. 12: If it were not for the ability to win 2.78 a bunch of money, I would probably not gamble much at all. 13: I only gamble when my friends are 2.48 gambling with me. 15: I find myself feeling a rush, and 2.76 getting excited when I gamble. 16: After I gamble, I often find comfort 1.22 from other people to help me deal with my losses. 20: I gamble primarily for the money that I can win. 3.36 3: Tangible 2: I really enjoy the complementary perks 2.56 that come along with gambling, like free points, drinks, comp coupons, etc. 9: When I gamble, I like to accumulate 1.44 points at a casino so they will offer me incentives and bonuses. 10: I like the sounds, the lights, and the 2.39 excitement that often go along with gambling. 18: I gamble more often when I have been 1.84 offered complementary drinks, hotel rooms or other items. GFA component Item number s 1: Sensory + 1: I tend to gamble most frequently when 1.96 Escape there is nothing else going on or I have nothing better to do. 4: I often gamble after fighting with my 1.34 spouse or significant other. 5: I feel more alive when I am gambling 1.89 than when I am doing other types of activities. 7: I often gamble when I feel stressed or 1.61 anxious. 11: I gamble to get a break from work or 1.94 other difficult tasks. 14: I often gamble when I am feeling 1.42 depressed or sad. 17: If I have a hard day at work, I am 1.61 likely to gamble. 19: When I gamble, I am often unaware of 1.72 my surroundings. 2: Attention 3: I enjoy the social aspects of gambling 2.24 such as being with my friends or being around other people who are having a good time and cheering me on. 6: Even if I lose, I can always count on a 2.33 friend/loved one to help me through this difficult time. 8: After I gamble, I like to go out and 2.13 celebrate my winnings with others. 12: If it were not for the ability to win 2.31 a bunch of money, I would probably not gamble much at all. 13: I only gamble when my friends are 2.25 gambling with me. 15: I find myself feeling a rush, and 2.22 getting excited when I gamble. 16: After I gamble, I often find comfort 1.84 from other people to help me deal with my losses. 20: I gamble primarily for the money that I can win. 2.40 3: Tangible 2: I really enjoy the complementary perks 2.22 that come along with gambling, like free points, drinks, comp coupons, etc. 9: When I gamble, I like to accumulate 1.99 points at a casino so they will offer me incentives and bonuses. 10: I like the sounds, the lights, and the 2.24 excitement that often go along with gambling. 18: I gamble more often when I have been 2.20 offered complementary drinks, hotel rooms or other items. GFA component Item number Skew 1: Sensory + 1: I tend to gamble most frequently when 0.58 Escape there is nothing else going on or I have nothing better to do. 4: I often gamble after fighting with my 2.49 spouse or significant other. 5: I feel more alive when I am gambling 0.79 than when I am doing other types of activities. 7: I often gamble when I feel stressed or 1.18 anxious. 11: I gamble to get a break from work or 1.01 other difficult tasks. 14: I often gamble when I am feeling 1.79 depressed or sad. 17: If I have a hard day at work, I am 1.83 likely to gamble. 19: When I gamble, I am often unaware of 1.30 my surroundings. 2: Attention 3: I enjoy the social aspects of gambling -0.36 such as being with my friends or being around other people who are having a good time and cheering me on. 6: Even if I lose, I can always count on a 0.42 friend/loved one to help me through this difficult time. 8: After I gamble, I like to go out and 0.24 celebrate my winnings with others. 12: If it were not for the ability to win 0.10 a bunch of money, I would probably not gamble much at all. 13: I only gamble when my friends are 0.28 gambling with me. 15: I find myself feeling a rush, and -0.01 getting excited when I gamble. 16: After I gamble, I often find comfort 1.36 from other people to help me deal with my losses. 20: I gamble primarily for the money that I can win. -0.35 3: Tangible 2: I really enjoy the complementary perks 0.24 that come along with gambling, like free points, drinks, comp coupons, etc. 9: When I gamble, I like to accumulate 1.07 points at a casino so they will offer me incentives and bonuses. 10: I like the sounds, the lights, and the 0.33 excitement that often go along with gambling. 18: I gamble more often when I have been 0.76 offered complementary drinks, hotel rooms or other items. GFA component Item number Kurtosis 1: Sensory + 1: I tend to gamble most frequently when -0.91 Escape there is nothing else going on or I have nothing better to do. 4: I often gamble after fighting with my 5.91 spouse or significant other. 5: I feel more alive when I am gambling -0.56 than when I am doing other types of activities. 7: I often gamble when I feel stressed or 0.42 anxious. 11: I gamble to get a break from work or -0.314 other difficult tasks. 14: I often gamble when I am feeling 2.59 depressed or sad. 17: If I have a hard day at work, I am 2.59 likely to gamble. 19: When I gamble, I am often unaware of 0.60 my surroundings. 2: Attention 3: I enjoy the social aspects of gambling -1.35 such as being with my friends or being around other people who are having a good time and cheering me on. 6: Even if I lose, I can always count on a -1.39 friend/loved one to help me through this difficult time. 8: After I gamble, I like to go out and -1.39 celebrate my winnings with others. 12: If it were not for the ability to win -1.50 a bunch of money, I would probably not gamble much at all. 13: I only gamble when my friends are -1.38 gambling with me. 15: I find myself feeling a rush, and - 1.43 getting excited when I gamble. 16: After I gamble, I often find comfort 0.54 from other people to help me deal with my losses. 20: I gamble primarily for the money that I can win. -1.47 3: Tangible 2: I really enjoy the complementary perks -1.4 that come along with gambling, like free points, drinks, comp coupons, etc. 9: When I gamble, I like to accumulate -0.28 points at a casino so they will offer me incentives and bonuses. 10: I like the sounds, the lights, and the -1.36 excitement that often go along with gambling. 18: I gamble more often when I have been -0.97 offered complementary drinks, hotel rooms or other items. Table 4 Pearson correlation matrix of participant scores on the SOGS, the original GFA, and the reduced GFA based on the results of the EFA and CFA SOGS total score GFA (20 items) GFA (10 items) SOGS total score -- GFA (20 items) 0.698 ** -- GFA (10 items) 0.745 ** 0.952 ** - -- ** p <0.01 Fig. 5 GFA-II assessment and scoring methods Gambling Functional Assessment II Answer the questions below using the provided scale. Write the corresponding number next to each question. Almost Half the Almost Never Never Seldom Time Usually Always Always 0 1 2 3 4 5 6 1. I tend to gamble most frequency when there is nothing else going on or I have nothing better to do.__ 2. I often gamble after fighting with my spouse or significant other.__ 3. I enjoy the social aspects of gambling such as being with my friends or being around other people who are having a good time and cheering me on.__ 4. If I have a hard day at work, I am likely to gamble.__ 5. When I gamble, I like to accumulate points at a casino so they will offer me incentives and bonuses.__ 6. I often gamble when I feel stressed or anxious.__ 7. I feel more alive when I am gambling than when I am doing other types of activities.__ 8. I really enjoy the complementary perks that come along with gambling, like free points, drinks, comp coupons, etc. __ 9. I gamble to get a break from work or other difficult tasks.__ 10. After I gamble, I often find comfort from other people to help me deal with my losses.__ SCORING: Write the number for each question in the following columns. Sum all items, and divide each by the total number of items, for the total score. Circle the column with the highest total score. SENSORY ESCAPE ATTENTIONTANGIBLE 1 2. 3. 5. 7. 4. 10. 8. 6. 9.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||ORIGINAL ARTICLE|
|Author:||Dixon, Mark R.; Wilson, Alyssa N.; Belisle, Jordan; Schreiber, James B.|
|Publication:||The Psychological Record|
|Article Type:||Case study|
|Date:||Jun 1, 2018|
|Previous Article:||Exploring Differential Trial-Type Effects and the Impact of a Read-Aloud Procedure on Deictic Relational Responding on the IRAP.|
|Next Article:||Kanji Writing Acquisition Through Sequential Stimulus Pairing in Japanese Students with Writing Difficulties.|