Advertisement

Reliability of Instruments Measuring At-Risk and Problem Gambling Among Young Individuals: A Systematic Review Covering Years 2009–2015

  • Robert Edgren
    Correspondence
    Address correspondence to: Robert Edgren, Department of Tobacco, Gambling and Addiction, National Institute for Health and Welfare, P.O. Box 30, FI-00271, University of Helsinki, Helsinki, Finland.
    Affiliations
    Institute of Behavioural Sciences, Faculty of Behavioural Sciences, University of Helsinki, Helsinki, Finland

    Department of Tobacco, Gambling and Addiction, National Institute for Health and Welfare, Helsinki, Finland
    Search for articles by this author
  • Sari Castrén
    Affiliations
    Department of Tobacco, Gambling and Addiction, National Institute for Health and Welfare, Helsinki, Finland

    Institute of Clinical Medicine, Faculty of Medicine, University of Helsinki, Helsinki, Finland
    Search for articles by this author
  • Marjukka Mäkelä
    Affiliations
    Finnish Office for HTA (FINOHTA) at National Institute for Health and Welfare, Helsinki, Finland

    Department of General Practice, Institute of Public Health, University of Copenhagen, Copenhagen, Denmark
    Search for articles by this author
  • Pia Pörtfors
    Affiliations
    Department of Information Services, National Institute for Health and Welfare, Helsinki, Finland
    Search for articles by this author
  • Hannu Alho
    Affiliations
    Department of Tobacco, Gambling and Addiction, National Institute for Health and Welfare, Helsinki, Finland

    Institute of Clinical Medicine, Faculty of Medicine, University of Helsinki, Helsinki, Finland
    Search for articles by this author
  • Anne H. Salonen
    Affiliations
    Department of Tobacco, Gambling and Addiction, National Institute for Health and Welfare, Helsinki, Finland

    Institute of Clinical Medicine, Faculty of Medicine, University of Helsinki, Helsinki, Finland
    Search for articles by this author
Open AccessPublished:April 14, 2016DOI:https://doi.org/10.1016/j.jadohealth.2016.03.007

      Abstract

      This review aims to clarify which instruments measuring at-risk and problem gambling (ARPG) among youth are reliable and valid in light of reported estimates of internal consistency, classification accuracy, and psychometric properties. A systematic search was conducted in PubMed, Medline, and PsycInfo covering the years 2009–2015. In total, 50 original research articles fulfilled the inclusion criteria: target age under 29 years, using an instrument designed for youth, and reporting a reliability estimate. Articles were evaluated with the revised Quality Assessment of Diagnostic Accuracy Studies tool. Reliability estimates were reported for five ARPG instruments. Most studies (66%) evaluated the South Oaks Gambling Screen Revised for Adolescents. The Gambling Addictive Behavior Scale for Adolescents was the only novel instrument. In general, the evaluation of instrument reliability was superficial. Despite its rare use, the Canadian Adolescent Gambling Inventory (CAGI) had a strong theoretical and methodological base. The Gambling Addictive Behavior Scale for Adolescents and the CAGI were the only instruments originally developed for youth. All studies, except the CAGI study, were population based. ARPG instruments for youth have not been rigorously evaluated yet. Further research is needed especially concerning instruments designed for clinical use.

      Keywords

      Implications and Contribution
      Rigorous psychometric evaluation of instruments measuring at-risk and problem gambling in young people has not been accomplished yet. Reporting estimates of internal consistency from previous articles is not enough, considering the weak theoretical foundation of these instruments. Reliability testing in population-based studies and validation for clinical use are needed.
      Adolescents have persistently been reported to have higher problem gambling rates than the adult population [
      • Blinn-Pike A.
      • Worthy S.L.
      • Jonkman J.N.
      Adolescent gambling: A review of an emerging field of research.
      ,
      • Volberg R.A.
      • Gupta R.
      • Griffiths M.D.
      • et al.
      An international perspective on youth gambling prevalence studies.
      ]. Furthermore, initiation of gambling at a young age has consistently been identified as a risk factor for developing gambling-related problems [
      • Castrén S.
      • Basnet S.
      • Pankakoski M.
      • et al.
      An analysis of problem gambling among the Finnish working-age population: A population survey.
      ,
      • Johansson A.
      • Grant J.E.
      • Kim S.W.
      • et al.
      Risk factors for problematic gambling: A critical literature review.
      ]. Therefore, adequate tools for identifying both at-risk and problem gambling (ARPG) among adolescents are of utmost importance [
      • Derevensky J.L.
      • Gupta R.
      Prevalence estimates of adolescent gambling: A comparison of the SOGS-RA, DSM-IV-J, and the GA 20 questions.
      ]. Here, ARPG refers to a wider spectrum of problematic adolescent gambling [
      • Potenza M.N.
      • Wareham J.D.
      • Steinberg M.A.
      • et al.
      Correlates of at-risk/problem internet gambling in adolescents.
      ,
      • Edgren R.
      • Castrén S.
      • Jokela M.
      • Salonen A.H.
      At-risk and problem gambling among Finnish youth: The examination of risky alcohol consumption, tobacco smoking, mental health and loneliness as gender-specific correlates.
      ]. The prevalence rates of adolescent problem gambling in relation to adult prevalence rates have raised concerns about the validity of screening instruments, notably whether adolescent rates are exaggerated [
      • Blinn-Pike A.
      • Worthy S.L.
      • Jonkman J.N.
      Adolescent gambling: A review of an emerging field of research.
      ,
      • Winters K.C.
      • Stinchfield R.
      Youth gambling: Prevalence, risk and protective factors and clinical issues.
      ].
      In 2010, Stinchfield [
      • Stinchfield R.
      A critical review of adolescent problem gambling assessment instruments.
      ] conducted a critical review of youth problem gambling assessment instruments, identifying four instruments. Two of these were simply adaptations of adult instruments where item wording was modified to better represent potential adolescent-specific adverse consequences [
      • Winters K.C.
      • Stinchfield R.
      Youth gambling: Prevalence, risk and protective factors and clinical issues.
      ,
      • Stinchfield R.
      A critical review of adolescent problem gambling assessment instruments.
      ]. These were the South Oaks Gambling Screen Revised for Adolescents (SOGS-RA) and the Diagnostic Statistical Manual IV (Multiple Response format) adapted for Juveniles (DSM-IV-J; DSM-IV-MR-J; Table 1). The Massachusetts Gambling Screen (MAGS) was developed on a sample of high school students, but according to developers, it could be used for both adolescents and adults [
      • Stinchfield R.
      A critical review of adolescent problem gambling assessment instruments.
      ,
      • Shaffer H.
      • Labrie R.
      • Scanlan K.M.
      • et al.
      Pathological gambling among adolescents: Massachusetts Gambling Screen (MAGS).
      ]. The Canadian Adolescent Gambling Inventory (CAGI) was the only instrument purposefully developed for an adolescent population.
      Table 1Properties of instruments for measurement of gambling problems among young people
      Instrument [ref]Content and structureItems and time frameClassification cutoff scoreStrengths and weaknesses
      SOGS-RA
      • Winters K.C.
      • Stinchfield R.
      • Fulkerson J.
      Towards the development of an adolescent gambling problem severity scale.
      Signs and symptoms of problem gambling and its negative consequences.12 Items with two response options (yes/no) scored 0–1. Four additional items provide insight to an individual's gambling, but not used in scoring

      Time frame: Past year
      Recent studies have preferred the narrow criteria: sum score 2–3 = at-risk gambling; ≥4 = problem gambling
      • Widespread use.
      • Confusing due to the two scoring procedures (“broad” and “narrow”)
        • Stinchfield R.
        A critical review of adolescent problem gambling assessment instruments.
        .
      • Calculates a sum score instead of weighting items (e.g.,
        • Wiebe J.M.
        • Cox B.J.
        • Mehmel B.G.
        The South Oaks Gambling Screen Revised for Adolescents (SOGS-RA): Further psychometric findings from a community sample.
        ).
      • Produces exaggerated prevalence rates compared to other instruments (e.g.
        • Langhinrichsen-Rohling J.
        • Rohling M.L.
        • Rohde P.
        • Seeley J.R.
        The SOGS-RA vs. the MAGS-7: Prevalence estimates and classification congruence.
        ,
        • Welte J.W.
        • Barnes G.M.
        • Tidwell M.C.O.
        • Hoffman J.H.
        The prevalence of problem gambling among US adolescents and young adults: Results from a national survey.
        ,
        • Ladouceur R.
        • Bouchard C.
        • Rhéaume N.
        • et al.
        Is the SOGS an accurate measure of pathological gambling among children, adolescents and adults?.
        ), possibly due to misinterpretation of items
        • Ladouceur R.
        • Bouchard C.
        • Rhéaume N.
        • et al.
        Is the SOGS an accurate measure of pathological gambling among children, adolescents and adults?.
        .
      DSM-IV-J
      • Fisher S.
      Measuring pathological gambling in children: The case of fruit machines in the UK.


      DSM-IV-MR-J
      • Fisher S.
      Developing the DSM-IV-DSM-IV criteria to identify adolescent problem gambling in non-clinical populations.
      Based on the DSM-IV criteria. The DSM-IV-MR-J is a modified version of the DSM-IV-J featuring simpler language, fewer details and multiple response options.12 Items with two response options (yes/no) scored 0–1

      Time frame: Past year
      Sum score ≥4 = problem gambling
      • The DSM-IV-J is considered more conservative than the SOGS-RA.
      • It fails to measure the DSM-IV criterion “loss of control.” The exclusion criterion is claimed to be premature. The multiple-response format of the DSM-IV-MR-J is collapsed when calculating the scores.
      • Insufficient evidence of the classification accuracy of the instrument
        • Stinchfield R.
        A critical review of adolescent problem gambling assessment instruments.
        .
      MAGS
      • Fisher S.
      Measuring pathological gambling in children: The case of fruit machines in the UK.
      Psychological and social problems related to gambling. Developed using items from the Short Michigan Alcoholism Screening Test
      • Selzer M.L.
      • Vonokur A.
      • van Roojien L.
      A self-administered Short Michigan Alcoholism Screening Test (SMAST).
      . DSM-IV was used as a reference standard in the development process.
      14 Items; seven items are scored in a scale based on item weights from a discriminant function analysis; yes/no

      Time frame: Past year
      Each item is scored 0 for no and 1 for yes. Each item score is multiplied by a weight and then summed along with constant using a weighted scoring algorithm derived from a discriminant function analysis. The instrument classifies respondents into nonpathologic gambling, transitional gambling, or pathologic gambling
      • Stinchfield R.
      A critical review of adolescent problem gambling assessment instruments.
      ,
      • Shaffer H.
      • Labrie R.
      • Scanlan K.M.
      • et al.
      Pathological gambling among adolescents: Massachusetts Gambling Screen (MAGS).
      .
      • Strength of the MAGS is its brevity.
      • The generalizability of the weighting procedure is unknown
        • Stinchfield R.
        A critical review of adolescent problem gambling assessment instruments.
        .
      • The MAGS is a seven-item scale with item weights used to score it
        • Shaffer H.
        • Labrie R.
        • Scanlan K.M.
        • et al.
        Pathological gambling among adolescents: Massachusetts Gambling Screen (MAGS).
        . This subscale includes an item “arrested for gambling,” the appropriateness of which is questionable
        • Stinchfield R.
        A critical review of adolescent problem gambling assessment instruments.
        .
      • Both the MAGS-7 subscale and the DSM-IV subscale are used independently of one another within the field, which is a likely source of confusion and complicates comparability across studies.
      • All information up to year 2015 is based on three articles.
      CAGI
      • Wiebe J.
      • Wynne H.
      • Stinchfield R.
      • Tremblay J.
      Measuring problem gambling in adolescent populations: Phase I report.
      ,
      • Wiebe J.
      • Wynne H.
      • Stinchfield R.
      • Tremblay J.
      The Canadian Adolescent Gambling Inventory (CAGI): Phase II final report.
      Five domains:
      • 1)
        gambling problem severity,
      • 2)
        psychological consequences,
      • 3)
        social consequences,
      • 4)
        financial consequences, and
      • 5)
        loss of control
      45 Items, using a four-point multiple response format

      Time frame: Past 3 months
      Three categories: no problem (score 0–1), low to moderate severity (score 2–5), high severity (score ≤6)
      • Includes an item inquiring about gambling activity of a fictitious gambling form, to test the validity of self-report.
      • Includes a nine-item subscale (GPSS) consisting of the discriminatively best items from the five domains.
      • Administration time is up to 20 minutes while the SOGS-RA, DSM-IV-J/DSM-IV-MR-J and MAGS have been estimated to take a maximum of 10 minutes
        • Stinchfield R.
        A critical review of adolescent problem gambling assessment instruments.
        .
      • All information up to year 2015 is based on three reports.
      GABSA
      • Park H.S.
      • Jung S.Y.
      Development of a Gambling Addictive Behavior Scale for adolescents in Korea.
      Four domains:
      • 1)
        loss of control
      • 2)
        life dysfunction from problem gambling
      • 3)
        gambling experience
      • 4)
        social dysfunction from problem gambling
      25 Items

      Time frame: Not specified
      Three categories: nongambling, nonproblem gambling, and problem gambling;

      no cutoff scores for classification specified
      • Concludes to have a good validity and reliability among adolescents in Korea.
      • Information about the instrument is limited.
      • All information up to year 2015 is based on a single article.
      CAGI = Canadian Adolescent Gambling Inventory; DSM-IV-J/DSM-IV-(MR)-J = the Diagnostic Statistical Manual IV (Multiple Response format) adapted for Juveniles; GABSA = Gambling Addictive Behavior Scale for Adolescents; GPSS = Gambling Problem Severity Subscale; MAGS = Massachusetts Gambling Screen; SOGS-RA = the South Oaks Gambling Screen–Revised for Adolescents.
      Previous studies on adolescent gambling have mainly been nontheoretical, quantitative, prevalence based, descriptive, and school based [
      • Blinn-Pike A.
      • Worthy S.L.
      • Jonkman J.N.
      Adolescent gambling: A review of an emerging field of research.
      ]. ARPG instruments are used to make judgements at the individual level, and for decision-making at a societal level, making it essential for such tools to be valid and reliable [
      • Stinchfield R.
      A critical review of adolescent problem gambling assessment instruments.
      ]. Overall, there is a lack of validated instruments to assess ARPG among youth [
      • Blinn-Pike A.
      • Worthy S.L.
      • Jonkman J.N.
      Adolescent gambling: A review of an emerging field of research.
      ,
      • Stinchfield R.
      A critical review of adolescent problem gambling assessment instruments.
      ,
      • Hayer T.
      • Meyer G.
      • Petermann F.
      Gambling-related problems among youths: A critical review of current screening instruments.
      ]. To our knowledge, a systematic review has not yet been conducted on this subject.
      This systematic review aims to summarize existing evidence on the psychometric evaluations of instruments designed to evaluate ARPG among youth. The transition from being an adolescent to a young adult, and simultaneously becoming able to gamble legally, is a critical phase concerning gambling behavior. The age range of youth is limited to a maximum of 28 years [

      Nuorisolaki 72/2006. [Youth Law]. Finlex Data Bank. Helsinki, Finland: Finland's Ministry of Justice, 2006. Available at: http://www.finlex.fi/fi/laki/alkup/2006/20060072. Accessed September 28, 2015.

      ]. Specifically, we wish to clarify which instruments measuring ARPG among youth are reliable and valid for both population-based and clinical studies in light of reported estimates of internal consistency, classification accuracy, and psychometric properties. Consequently, the aim was to identify suitable instruments presently available and provide insight on what branches of the field require further investigation.

      Methods

      Search strategy

      A structured electronic search was conducted (November 2014) in PubMed, Medline, and PsycInfo databases, covering articles published between 2009 and 2014. The search result was processed and reported according to methods recommended in the PRISMA statement [
      • Liberati A.
      • Altman D.G.
      • Tetzlaff J.
      • et al.
      The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: Explanation and elaboration.
      ,
      • Moher D.
      • Liberati A.
      • Tetzlaff J.
      • Altman D.G.
      Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement.
      ]. This study complements the earlier reviews [
      • Blinn-Pike A.
      • Worthy S.L.
      • Jonkman J.N.
      Adolescent gambling: A review of an emerging field of research.
      ,
      • Stinchfield R.
      A critical review of adolescent problem gambling assessment instruments.
      ,
      • Derevensky J.
      • Gupta R.
      The measurement of youth gambling problems: Current instruments, methodological issues and future directions.
      ,
      • Derevensky J.
      • Gupta R.
      Measuring gambling problems amongst adolescents: Current status and future directions.
      ]. The latest two reviews [
      • Blinn-Pike A.
      • Worthy S.L.
      • Jonkman J.N.
      Adolescent gambling: A review of an emerging field of research.
      ,
      • Stinchfield R.
      A critical review of adolescent problem gambling assessment instruments.
      ] were published in 2010; therefore, to fill in the gap between preparing the articles and publishing them, the year 2009 was included in our search.
      The search terms were categorized using the Cochrane handbook guidelines for formulating review questions and inclusion criteria PICO [

      O'Connor D, Green S, Higgins JPT. Chapter 5: Defining the review question and developing criteria for including studies. In: Higgins JPT, Green S, eds. Cochrane handbook for systematic reviews of interventions version 5.0.0 [updated February 2008]. The Cochrane Collaboration, 2008. Available at: http://handbook.cochrane.org/chapter_5/5_defining_the_review_question_and_developing_criteria_for.htm. Accessed March 31, 2016.

      ]: P (patient, i.e., population), I (intervention, i.e., instrument), C (comparator, i.e., reference instrument), and O (outcome, i.e., reliability). The search strategy for PubMed is in the Supplementary Data; further search strategy details across bibliographic databases are available on request. On completion, the searches from each database were documented and references imported into RefWorks, where duplicates were eliminated.
      Reference lists of the included articles and of identified review articles were scrutinized to find articles unrecognized in initial searches, resulting in eight additional articles for the systematic review. Database searches were updated in November 2015. The flow chart of the article selection process is presented in Figure 1.
      Figure thumbnail gr1
      Figure 1Flow chart of the retrieval and inclusion/exclusion process for articles used in review.

      Eligibility criteria

      Original research articles written in English were accepted. Publications in peer-reviewed journals, doctoral theses, and institutional reports were accepted. Both population-based samples and clinical samples were eligible. First, 822 abstracts were evaluated using the following exclusion criteria: (1) non–gambling-related research topic; (2) sample age >28 years; (3) no gambling instrument employed; (4) case study, commentary, editorial, or letter; and (5) other (specification required; for example, papers involving only families of young gamblers or youth whose parents gambled were excluded).
      Second, 445 full texts of the articles, doctoral theses [
      • Mooss A.V.
      Gambling behaviours among youth involved in juvenile and family courts.
      ,
      • West R.L.
      The effect of extended family gambling behavior and family functioning on African American adolescent gambling.
      ], and reports [
      ,
      • Dowling N.
      • Jackson A.C.
      • Thomas S.A.
      • Frydenberg E.
      Children at risk of developing problem gambling: Final report. A report to gambling research Australia in fulfilment of Tender 103/06.
      ,
      • Tremblay J.
      • Stinchfield R.
      • Wiebe J.
      • Wynne H.
      Canadian Adolescent Gambling Inventory (CAGI) phase III final report.
      ] (later referred as articles) were evaluated using inclusion criteria matching the PICO:
      • P: ≤28 years of age;
      • I: instrument designed to assess youth gambling; and
      • O: instrument reliability reported.
      A comparator criterion was not required for inclusion but was included for labeling the articles. It referred to whether the study used a reference standard in addition to the primary instrument for gambling assessment. Six articles repeating the same reliability estimates as the primary source were excluded. Altogether, 50 articles were included.
      In all phases, two researchers independently assessed all articles. The joint probability of agreement was 90.9% for the exclusion of articles based on the abstracts. The joint probability of agreement for the inclusion of articles based on full-text evaluation was 92.4%. Disagreement in evaluation was resolved by discussion and with a third independent researcher.

      Quality assessment

      The articles were appraised using the revised Quality Assessment of Diagnostic Accuracy Studies tool (QUADAS-2) [
      • Whiting P.F.
      • Rutjes A.W.
      • Westwood M.E.
      • et al.
      QUADAS-2: A revised tool for the quality assessment of diagnostic accuracy studies.
      ], which assesses the risk of bias and the applicability of articles. QUADAS-2 questions were tailored for our review, tested on a subsample of articles and modified to ensure unambiguous assessment between researchers. An important change was omitting the evaluation of risk of bias regarding the reference standard. This decision was made because both the index test and reference standard (if measured) were applied identically within the gambling context. Information pertaining to the reference standard was inferred from the applicability assessment. The risk of bias was assessed by five questions (Table 2), and each guiding question required an answer of “high,” “low,” or “unclear” risk of bias, where unclear risk refers to suboptimal reporting. The applicability of articles was evaluated with three questions (Table 2), and each item was scored as either “good,” “poor,” or “unclear” applicability. The joint probability of agreement was 89.8%. Disagreement was resolved by discussion (see Eligibility criteria section).
      Table 2Quality assessment of included articles using the revised Quality Assessment of Diagnostic Accuracy Studies tool
      ArticlesRisk of bias n (%)Applicability n (%)
      Patient selection
      Were participants selected randomly? Was the sample representative of the general population? Were inappropriate exclusions avoided?
      Index test
      Did the conduct of the test avoid introducing bias?
      Flow and timing
      Were all participants included in the analysis? Applicability to the review was assessed with the questions:
      Patient selection
      Does the sample match the review question?
      Index test
      Is the validity of the index test a research question of the study?
      Reference test
      Is a reference standard used to make judgements about the validity or reliability of the index test?
      All (n = 50)
       Low risk/good applicability24 (48.0)38 (76.0)43 (86.0)43 (86.0)9 (18.0)12 (24.0)
       High risk/poor applicability25 (50.0)3 (6.0)2 (4.0)6 (12.0)41 (82.0)38 (76.0)
       Unclear1 (2.0)9 (18.0)5 (10.0)1 (2)0 (0)0 (0)
      SOGS-RA (n = 33)
       Low risk/good applicability14 (42.4)24 (72.7)30 (90.9)27 (81.8)3 (9.1)5 (15.2)
       High risk/poor applicability19 (57.6)3 (9.1)1 (3.0)5 (15.2)30 (90.9)28 (84.8)
       Unclear0 (0)6 (18.2)2 (6.1)1 (3.0)0 (0)0 (0)
      DSM-IV-J/DSM-IV-MR-J (n = 12)
       Low risk/good applicability5 (41.7)11 (91.7)9 (75.0)11 (91.7)2 (16.7)4 (33.3)
       High risk/poor applicability6 (50.0)0 (0)0 (0)1 (8.3)10 (83.3)8 (66.7)
       Unclear1 (8.3)1 (8.3)3 (25.0)0 (0)0 (0)0 (0)
      MAGS (n = 3)
       Low risk/good applicability3 (100.0)2 (66.7)2 (66.7)3 (100.0)2 (66.7)1 (33.3)
       High risk/poor applicability0 (0)1 (33.3)1 (33.3)0 (0)1 (33.3)2 (66.7)
       Unclear0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)
      CAGI (n = 1)
       Low risk/good applicability1 (100.0)1 (100.0)1 (100.0)1 (100.0)1 (100.0)1 (100.0)
       High risk/poor applicability0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)
       Unclear0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)
      GABSA (n = 1)
       Low risk/good applicability1 (100.0)0 (0)1 (100.0)1 (100.0)1 (100.0)1 (100.0)
       High risk/poor applicability0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)
       Unclear0 (0)1 (100.0)0 (0)0 (0)0 (0)0 (0)
      CAGI = Canadian Adolescent Gambling Inventory; DSM-IV-J/DSM-IV-(MR)-J = the Diagnostic Statistical Manual IV (Multiple Response format) adapted for Juveniles; GABSA = Gambling Addictive Behavior Scale for Adolescents; GPSS = Gambling Problem Severity Subscale; MAGS = Massachusetts Gambling Screen; SOGS-RA = the South Oaks Gambling Screen–Revised for Adolescents.
      Risk of bias was assessed with the following questions:
      a Were participants selected randomly? Was the sample representative of the general population? Were inappropriate exclusions avoided?
      b Did the conduct of the test avoid introducing bias?
      c Were all participants included in the analysis? Applicability to the review was assessed with the questions:
      d Does the sample match the review question?
      e Is the validity of the index test a research question of the study?
      f Is a reference standard used to make judgements about the validity or reliability of the index test?

      Data extraction

      Information from the articles was compiled into tables (Tables 3 and 4). Data from external articles (e.g., methodological reports) were used when necessary to complete study information.
      Table 3Characteristics, applicability, and risk of bias of included studies on instruments for measurement of gambling problems among young people
      Ref.Country (year)Instrument (classification cutoff score)Study characteristics
      a) Design, b) sampling method, c) administration method, d) statistical methods related to reliability and validity, e) references of ARPG: comparator measurement.
      Sample characteristics
      Sample size in analysis and in total, gender distribution from analyzed sample (unless otherwise specified as total sample), mean age (range) in years.
      Applicability
      QUADAS-2 assessment for applicability concerns: patient selection; index test; reference standard.
      Risk of bias
      QUADAS-2 assessment of risk of bias concerning patient selection, index test, and flow and timing.
      • Barnes G.M.
      • Welte J.W.
      • Hoffman J.H.
      • Tidwell M.-C.O.
      Gambling, alcohol, and other substance use among youth in the United States.
      USA (2009)SOGS-RA;

      DSM-IV-MR-J;

      DIS-IV adapted for adolescents (≥3 gambling problems in the past year = gambling problems)
      • a)
        Cross-sectional
      • b)
        Random nationally representative sample
      • c)
        Computer-assisted telephone interviews
      • d)
        Total sum of instruments
      • e)
        Correlation with principal component
      n = 2,274 (4,467)

      female: 49.5%

      age: (14–21)
      Good

      Poor

      Poor
      Low

      Low

      Low
      • Mooss A.V.
      Gambling behaviours among youth involved in juvenile and family courts.
      USA (2009)SOGS-RA (2–3 = at risk; ≥4 = problem gambling)
      • a)
        Cross-sectional
      • b)
        Nonrandom sample from nine local juvenile courts
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha
      • e)
        Correlation with scope of gambling activities and gambling-related crime
      n = 145

      female: 31%

      age: 15.45 (12–18)
      Good

      Poor

      Poor
      High

      Low

      Low
      • Taylor L.M.
      • Hillyard P.
      Gambling awareness for youth: An analysis of the “Don’t gamble away our future™” program.
      USA (2009)MSOGST (1–4 = at risk; ≥5 = probable pathological)
      • a)
        Experimental
      • b)
        Convenience sample from school settings
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha
      n = 8,455

      female: 52%

      age: not specified
      Unclear

      Poor

      Poor
      High

      Unclear

      Low
      • Wanner B.
      • Vitaro F.
      • Carbonneau R.
      • Tremblay R.E.
      Cross-lagged links among gambling, substance use, and delinquency from midadolescence to young adulthood: Additive and moderating effects of common risk factors.
      Canada (2009)SOGS-RA;

      SOGS (no categories used)
      • a)
        Longitudinal
      • b)
        Two community samples; Sample A: all French-speaking school boys in economically disadvantaged area; sample B: representative sample of kindergarteners
      • c)
        Self-report questionnaires
      • d)
        Cronbach alpha
      • e)
        Correlation with gambling problems at age 16 (SOGS-RA) and 23 (SOGS).
      n = sample A: 502 (1,037); sample B: 663 (2,000)

      female: 0%

      age: middle adolescence 16.2/16.2; early adulthood 22.8/22.5
      Good

      Good

      Good
      High

      Low

      Low
      • Welte J.W.
      • Barnes G.M.
      • Tidwell M.-C.O.
      • Hoffman J.H.
      Association between problem gambling and conduct disorder in a national survey of adolescent and young adults in the Unites States.
      USA (2009)SOGS-RA (2–3 = at risk, ≥4 = problem gambling); DSM-IV-MR-J
      • a)
        Cross-sectional
      • b)
        National random sample stratified by county
      • c)
        Computer-assisted telephone interview
      • d)
        Cronbach alpha
      • e)
        Convergent validity with DSM instrument
      n = 2,258 (2,274)

      female: not specified

      age: (14–21)
      Good

      Poor

      Good
      Low

      Low

      Low
      • Barnes G.M.
      • Welte J.W.
      • Hoffman J.H.
      • Tidwell M.-C.O.
      Comparisons of gambling and alcohol use among college students and noncollege young people in the United States.
      USA (2010)SOGS-RA (≥2 symptoms)
      • a)
        Cross-sectional
      • b)
        Random nationwide household sample
      • c)
        Telephone interview
      • d)
        Cronbach alpha
      n = 1,000 (2,274)

      females: 51.5%

      age: 18–21
      Good

      Poor

      Poor
      Low

      High

      Low
      • Hansen M.
      • Rossow I.
      Limited cash flow on slot machines: Effects of prohibition of note acceptors on adolescent gambling behavior.
      Norway (2010)SOGS-RA (2–3 = at risk; ≥4 = problem gambling)
      • a)
        Cross-sectional over three time waves (2004; 2005; 2006)
      • b)
        National high school sample
      • c)
        Self-report questionnaire
      • d)
        Instrument sensitivity to change after intervention
      n = 20,648; 21,260; 20,573 (25,037; 24,560; 24,137)

      female: 50.7%; 49.9%; 50.4%

      age: 15 (13–19)
      Good

      Poor

      Poor
      Low

      Low

      Low
      • Barnes G.M.
      • Welte J.W.
      • Hoffman J.H.
      • Tidwell M.-C.O.
      The co-occurrence of gambling with substance use and conduct disorder among youth in the United States.
      USA (2011)SOGS-RA;

      DSM-IV-MR-J;

      DIS-IV (≥3 gambling problems in the past year = gambling problems)
      • a)
        Cross-sectional
      • b)
        Random nationwide household sample
      • c)
        Telephone interview
      • d)
        Cronbach alpha
      n = 2,258 (2,274)

      females: 49.5% (total sample)

      age: (14–21)
      Good

      Poor

      Good
      Low

      High

      Low
      • Dussault F.
      • Brendgen M.
      • Vitaro F.
      • et al.
      Longitudinal links between impulsivity, gambling problems and depressive symptoms: A transactional model from adolescence to early adulthood.
      Canada (2011)SOGS-RA;

      SOGS (no categories used)
      • a)
        Longitudinal
      • b)
        Local school sample
      • c)
        Self-report
      • d)
        Cronbach alpha
      n = 1,004 (1,162)

      females: 0%

      age: measurements at age 10, 14, 17 (SOGS-RA), 23 (SOGS)
      Poor

      Poor

      Poor
      High

      Low

      Low
      • Villella C.
      • Martinotti G.
      • Nicola M.D.
      • et al.
      Behavioural addictions in adolescents and young adults: Results from a prevalence study.
      Italy (2011)SOGS-RA (≥5 = probable pathologic gambling)
      • a)
        Cross-sectional
      • b)
        Local high school sample
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha
      n = 2,853

      female: 40%

      age: 16.7 (13–20)
      Good

      Poor

      Poor
      High

      Low

      Unclear
      • West R.L.
      The effect of extended family gambling behavior and family functioning on African American adolescent gambling.
      USA (2011)SOGS-RA (≥2 = at risk or problem gambler)
      • a)
        Cross-sectional
      • b)
        African-American student sample from three local high schools
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha
      n = 634 (749)

      female: 37.1%

      age: 15.8 (SD 1.4)
      Poor

      Poor

      Poor
      High

      Low

      Low
      • Ashrafioun L.
      • McCarthy A.
      • Rosenberg H.
      Assessing the impact of cue exposure on craving to gamble in university students.
      USA (2012)SOGS-RA (no categories used)
      • a)
        Cross-sectional
      • b)
        Convenience sample from one university
      • c)
        Web-based self-report questionnaire
      • d)
        Cronbach alpha
      • e)
        SOGS-RA correlation with Gambling Urge Scale
      n = 48 (56)

      female: 37%

      age: 21.1 (SD = 2.2; participants aged ≤30 excluded)
      Poor

      Poor

      Poor
      High

      Low

      Low
      • Chiu E.Y.-W.
      • Woo K.
      Problem gambling in Chinese American adolescents: Characteristics and risk factors.
      USA (2012)SOGS-RA (2–3 = at risk; ≥4 = problem gambling/probable pathological)
      • a)
        Cross-sectional
      • b)
        Sample from three selected local high schools
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha
      n = 183 (192)

      female: 48.4%

      age: 15.9 (13–19)
      Good

      Poor

      Poor
      High

      Low

      Low
      • Liu W.
      • Lee G.P.
      • Goldweber A.
      • et al.
      Impulsivity trajectories and gambling in adolescence among urban male youth.
      USA (2012)SOGS-RA (2–3 = at risk; ≥4 = problem)
      • a)
        Longitudinal
      • b)
        Random sampling from local primary schools
      • c)
        Self-reported questionnaire
      • d)
        Cronbach alpha
      n = 310 (678)

      female: 0%

      age: 11–15 (impulsivity) 17, 19, 20 (SOGS-RA; highest score used for analyses)
      Poor

      Poor

      Poor
      Low

      Unclear

      Low
      • Walther B.
      • Morgenstern M.
      • Hanewinkel R.
      Co-occurrence of addictive behaviours: Personality factors related to substance use, gambling and computer gaming.
      Germany (2012)SOGS-RA (2–3 = at risk; 4–5 = problem; >5 = probable pathologic gambling; classifications combined for analysis)
      • a)
        Cross-sectional
      • b)
        Randomly selected local high school sample
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha
      n = 2,553 (2,640)

      female: 49.3%

      age: 16.7 (12–25)
      Good

      Poor

      Poor
      Low

      Low

      Low
      • Wong S.S.K.
      • Tsang S.K.M.
      Validation of the Chinese version of the Gamblers’ Belief Questionnaire (GBQ-C).
      Hong Kong (2012)SOGS-RA (classification not specified)

      Chinese version of the Gamblers Belief Questionnaire; Chinese version of the Gambling Urge scale
      • a)
        Cross-sectional
      • b)
        Convenience sampling from Integrated Children and Youth Services Centers
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha
      • e)
        Correlations between scales
      n = 258

      female: 25.2%

      age: 16.13 (12–19)
      Good

      Poor

      Poor
      High

      Low

      Low
      • Chiesi F.
      • Donati M.A.
      • Galli S.
      • Primi C.
      The suitability of the South Oaks Gambling Screen-Revised for Adolescents (SOGS-RA) as a screening tool: Irt-based evidence.
      Italy (2013)SOGS-RA (2–3 = at risk; ≥4 = problem gambler)
      • a)
        Cross-sectional
      • b)
        Randomly selected local high school sample
      • c)
        Self-report
      • d)
        Test information function
      n = 871 (981)

      females: 36% (total sample)

      age: 16.57 (14–20)
      Good

      Good

      Poor
      High

      Low

      Low
      • Colasante E.
      • Gori M.
      • Bastiani L.
      • et al.
      Italian adolescent gambling behavior: Psychometric evaluation of the South Oaks Gambling Screen: Revised for Adolescents (SOGS-RA) among a sample of Italian students.
      Italy (2013)SOGS-RA (2–3 = at risk; ≥4 = problem gambler)
      • a)
        Cross-sectional
      • b)
        Nationwide high school sample
      • c)
        Self-report
      • d)
        Multiple Correspondence Analysis, kappa coefficient, Cronbach alphas, item endorsement, factor structure
      • e)
        Gambling frequency
      n = 5,930 (n = 14,910)

      females: 48.6% (total sample)

      age: 17 (15–19)
      Good

      Good

      Good
      High

      Low

      Low
      • Donati M.A.
      • Chiesi F.
      • Primi C.
      A model to explain at-risk/problem gambling among male and female adolescents: Gender similarities and differences.
      Italy (2013)SOGS-RA (broad definition (see
      • Poulin C.
      Problem gambling among adolescent students in the Atlantic provinces of Canada.
      ): no problem, at risk, problem gambling)
      • a)
        Cross-sectional
      • b)
        Randomly selected local high school sample
      • c)
        In-class self-report questionnaire
      • d)
        Cronbach alpha
      n = 943 (994)

      females: 46% (total sample)

      age: 16.57 (first- to fifth-year students)
      Good

      Poor

      Poor
      High

      Low

      Low
      • Lee G.P.
      • Stuart E.A.
      • Ialongo N.S.
      • Martins S.S.
      Parental monitoring trajectories and gambling among a longitudinal cohort of urban youth.
      USA (2013)SOGS-RA;

      SOGS (2–3 = at risk; ≥4 = problem; cumulative measures made for gambling variable by using participants' highest involvement at any year.)
      • a)
        Longitudinal
      • b)
        Random sampling from local primary schools
      • c)
        Self-reported questionnaire
      • d)
        Cronbach alpha
      n = 514 (678)

      female: 47.1% (total sample)

      age: 11–14 (parental monitoring) 16, 18, 19 (SOGS-RA), 20–22 (SOGS)
      Poor

      Poor

      Poor
      Low

      Unclear

      Low
      • Parker J.D.A.
      • Summerfeldt L.J.
      • Kloosterman P.H.
      Gambling behavior in adolescents with learning disorders.
      Canada (2013)SOGS-RA;

      DSM-IV-J (2–3 = at risk; ≥4 = probable pathologic gambling)
      • a)
        Cross-sectional
      • b)
        Community-based high school sample
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha
      • e)
        Correlation between instruments
      n = 532 (2,004)

      female: 36.5%

      age: 16.29 (14–18)
      Good

      Poor

      Good
      High

      Low

      Unclear
      • Primi C.
      • Donati M.A.
      • Bellini I.
      • et al.
      Measuring the attitude towards the profitability of gambling: The psychometric properties of the Gambling Attitude Scale.
      Italy (2013)SOGS-RA (Broad definition (see
      • Poulin C.
      Problem gambling among adolescent students in the Atlantic provinces of Canada.
      ) no problem, at risk, problem gambling);

      Gambling Attitude Scale
      • a)
        Cross-sectional
      • b)
        Sample from four local high schools
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha
      n = 960 (981)

      female: 36% (total sample)

      age: 16.57 (13–23)
      Good

      Poor

      Poor
      High

      Low

      Low
      • Tozzi L.
      • Akre C.
      • Fleury-Schubert A.
      • Suris J.-C.
      Gambling among youths in Switzerland and its association with other addictive behaviours.
      Switzerland (2013)SOGS-RA (French version; adapted: 8 of 12 items used) (2–3 = at risk; ≥4 = problem gambling)
      • a)
        Cross-sectional
      • b)
        Local noncompulsory secondary school sample
      • c)
        Online questionnaire
      • d)
        Cronbach alpha
      n = 1,102 (1,126)

      female: 48.7%

      age: 15–20 (73.7% under 18 years of age)
      Good

      Poor

      Poor
      High

      High

      Low
      • Wong G.
      • Zane N.
      • Saw A.
      • Chan A.K.K.
      Examining gender differences for gambling engagement and gambling problems among emerging adults.
      USA (2013)SOGS-RA (≥2 = problem with gambling)
      • a)
        Cross-sectional
      • b)
        Sampled from two high schools in different regions
      • c)
        Online self-report questionnaire
      • d)
        Cronbach alpha
      n = 743

      female: 57.9%

      age: 18.7 (18–20)
      Good

      Poor

      Poor
      High

      Unclear

      Low
      Canada (2014)SOGS-RA (six items; ≥2 = gambling problem)
      • a)
        Cross-sectional survey over several time waves
      • b)
        Two-stage stratified cluster selection school samples in Ontario (classes selected randomly)
      • c)
        Self-report questionnaires
      • d)
        Cronbach alpha
      n ≈ 4,000 to 10,000 (range) age: (12–18)

      2013:

      n = 10,272 (10,398)

      females: 48.2%

      age: (grades 7–12)
      Good

      Poor

      Poor
      Low

      Low

      Low
      • Bray B.C.
      • Lee G.P.
      • Liu W.
      • et al.
      Transitions in gambling participation during late adolescence and young adulthood.
      USA (2014)SOGS-RA (no categories used)
      • a)
        Longitudinal
      • b)
        Randomized block design of local schools
      • c)
        Self-reported questionnaire
      • d)
        Cronbach alpha
      n = 515 (678)

      females: 45%

      age: 17–22
      Good

      Poor

      Poor
      High

      Low

      Low
      • Cook S.
      • Turner N.E.
      • Ballon B.
      • et al.
      Problem gambling among Ontario students: Associations with substance abuse, mental health problems, suicide attempts, and delinquent behaviours.
      Canada (2014)SOGS-RA (six items; ≥2 = gambling problem)
      • a)
        Cross-sectional
      • b)
        Stratified cluster sample of Ontario students
      • c)
        Self-report questionnaires
      • d)
        Cronbach alpha, ROC analysis
      n = 4,851 (4,980)

      females: 53% (total sample)

      age: 14.6 (grades 7–12)
      Good

      Poor

      Poor
      Low

      Unclear

      Low
      • Donati M.A.
      • Primi C.
      • Chiesi F.
      Prevention of problematic gambling behavior among adolescents: Testing the efficacy of an integrative intervention.
      Italy (2014)SOGS-RA (broad definition see
      • Poulin C.
      Problem gambling among adolescent students in the Atlantic provinces of Canada.
      ): no problem, at risk, problem gambling)
      • a)
        Experimental
      • b)
        Sample from two local high schools
      • c)
        In-class self-report questionnaire before and after intervention
      • d)
        Prevalence of at risk and problem gambling before and after intervention (McNemar tests)
      n = 181

      female: 36%

      age: 15.95 (15–18)

      (Training group: n = 119; female: 17%)
      Good

      Poor

      Poor
      Low

      Low

      High
      • Canale N.
      • Vieno A.
      • Griffiths M.D.
      • et al.
      Trait urgency and gambling in young people by age: The mediating role of decision making processes.
      Italy (2015)SOGS-RA (2–3 = at risk; ≥4 = problem)
      • a)
        Cross-sectional
      • b)
        Local school sample
      • c)
        In-class self-report questionnaire
      • d)
        Cronbach alpha
      n = 986

      female: 36%

      age: 19.51 (16–25)
      Good

      Poor

      Poor
      High

      Unclear

      Low
      • Carbonneau R.
      • Vitaro F.
      • Brendgen M.
      • Tremblay R.E.
      Variety of gambling activities from adolescence to age 30 and association with gambling problems: A 15-year longitudinal study of a general population sample.
      Canada (2015)SOGS-RA;

      SOGS (1–4 = some problems with gambling; ≥5 = probable pathologic gambler)
      • a)
        Longitudinal
      • b)
        Random stratified sample of Quebec students
      • c)
        Structured interview (age 15 and 22), self-report questionnaire (age 30)
      • d)
        Cronbach alpha
      At ages 15, 22, and 30:

      n = 1,882, 1,785, and 1,358

      female: 50.2%, 55.5%, and 59.8%

      age: 15, 22, 30
      Good

      Poor

      Poor
      Low

      Low

      Low
      • Chan A.K.K.
      • Zane N.
      • Wong G.M.
      • Song A.V.
      Personal gambling expectancies among Asian American and White American college students.
      USA (2015)SOGS-RA (≥2 = problem gambling)
      • a)
        Cross-sectional
      • b)
        Convenience school sample
      • c)
        Self-report online questionnaire completed at school computer laboratory
      • d)
        Cronbach alpha, omega
        • Revelle W.
        • Zinbarg R.E.
        Coefficients alpha, beta, omega, and the glb: Comments on Sijtsma.
      n = 813

      female: 50.6%

      age: 19.5 (18–25)
      Good

      Poor

      Poor
      High

      Low

      Low
      • Míguez M.C.
      • Becona E.
      Do cigarette smoking and alcohol consumption associate with cannabis use and problem gambling among Spanish adolescents?.
      Spain (2015)SOGS-RA (2–3 = at risk; ≥4 = problem)
      • a)
        Cross-sectional
      • b)
        Random clustered sample from 17 randomly selected schools
      • c)
        In-class self-report questionnaire
      • d)
        Cronbach alpha
      n = 1,447

      female: 44.9%

      age: 12.8 (11–16)
      Good

      Poor

      Poor
      Low

      Low

      Low
      • Sheela P.S.
      • Choo W.-Y.
      • Goh L.Y.
      • Tan C.P.L.
      Gambling risk amongst adolescents: Evidence from a school-based survey in the Malaysian setting.
      Malaysia (2015)SOGS-RA (≥4 = problem gambling)
      • a)
        Cross-sectional
      • b)
        Random school sample
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha
      n = 2,262

      female: 57.6%

      age: 14.2 (12–17)
      Good

      Poor

      Poor
      Low

      Low

      Low
      • Delfabbro P.
      • King D.
      • Lambos C.
      • Puglies S.
      Is video-gaming playing a risk factor for pathological gambling in Australian adolescents?.
      Australia (2009)DSM-IV-J (0 = not at risk; 1–3 = at risk; ≥4 = problem gambling)
      • a)
        Cross-sectional
      • b)
        Representative sample from six schools
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha
      n = 2,669

      female: 49.2%

      age: 14.63 (12–17)
      Good

      Poor

      Poor
      Low

      Low

      Unclear
      • Skokauskas N.
      • Burba B.
      • Freedman D.
      An assessment of the psychometric properties of Lithuanian versions of DSM-IV-MR-J and SOGS-RA.
      Lithuania (2009)DSM-IV-MR-J (2–3 = at risk; ≥4 = problem gambling);

      SOGS-RA (2–3 = at risk; ≥4 = problem gambling)
      • a)
        Cross-sectional
      • b)
        Sample selected from all schools within city of Kaunas
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha, classification accuracy, discrimination function analysis
      • e)
        Comparison to external references, correlation between instruments
      n = 835

      female: 52.7%

      age: 14.5 (10–18)
      Good

      Good

      Good
      Low

      Low

      Low
      • Dowling N.
      • Jackson A.C.
      • Thomas S.A.
      • Frydenberg E.
      Children at risk of developing problem gambling: Final report. A report to gambling research Australia in fulfilment of Tender 103/06.
      Australia (2010)DSM-IV-MR-J (2–3 = at-risk gambling; ≥4 = problem gambling)
      • a)
        Cross-sectional
      • b)
        Sample of secondary schools from metropolitan and regional areas of Victoria
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha
      n = 612

      female: 60.6%

      age: 16 (12–18)
      Good

      Poor

      Poor
      Low

      Low

      Low
      • Brunelle N.
      • Leclerc D.
      • Cousineau M.-M.
      • et al.
      Internet gambling, substance use, and delinquent behavior: An adolescent deviant behavior involvement pattern.
      Canada (2012)DSM-IV-MR-J (2–3 = at risk; ≥4 = probable pathologic gambling)
      • a)
        Cross-sectional
      • b)
        Convenience sample from local schools
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha
      n = 1,870

      females: 54.1% (total sample)

      age: 15.43 (14–18)
      Poor

      Poor

      Poor
      High

      Low

      Low
      • Floros G.D.
      • Siomos K.
      • Fisoun V.
      • Geroukalis D.
      Adolescent online gambling: The impact of parental practices and correlates with online activities.
      Greece (2013)DSM-IV-MR-J (≥4 = probable pathologic gambling)
      • a)
        Cross-sectional
      • b)
        Entire student population aged 12–19 on island of Kos
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha
      n = 2,017

      females: 48.2%

      age: 15.08 (12–19)
      Good

      Poor

      Poor
      High

      Low

      Low
      • Lepper J.
      • Haden B.
      Testing NLCLiP: Validation of estimates of rates of non-problematic and problematic gambling in a sample of British schoolchildren.
      England (2013)NL-CLiP (0–2 = nonproblem);

      DSM-IV-MR-J (2–3 = at risk; ≥4 = problem gambling)
      • a)
        Cross-sectional
      • b)
        Schools sampled from England, Scotland, and Wales, classes selected randomly
      • c)
        Self-report questionnaire
      • d)
        Cohen kappa
      n = 1,425 (8,958)

      females: 49.6% (total sample)

      age: 11–15
      Good

      Poor

      Good
      Low

      Low

      Unclear
      • Tolchard B.
      • Delfabbro P.
      The Victorian Gambling Screen: Validity and reliability in an adolescent population.
      Australia (2013)Victorian Gambling Screen (0–7 = nonproblem gambling; 8–20 = borderline problem; ≥21 = problem gambling);

      DSM-IV-J (≥4 = pathologic gambling)
      • a)
        Cross-sectional
      • b)
        Sample from 18 schools selected from Australian Capital Territory
      • c)
        Self-report questionnaire (not directly specified)
      • d)
        Cronbach alpha
      • e)
        Instrument correlation, classification comparison between instruments
      n = 926

      female: 48.4%

      age: 14.46 (approximately 11–19)
      Good

      Poor

      Good
      Unclear

      Unclear

      Unclear
      • Cheung N.W.T.
      Low self-control and co-occurrence of gambling with substance use and delinquency among Chinese adolescents.
      China (2014)DSM-IV-J (2–3 = at risk; ≥4 = probable pathologic gambling)
      • a)
        Cross-sectional
      • b)
        Random sampling of high schools in Hong Kong
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha
      n = 4,734 (5,523)

      females: 49.3%

      age: 16.39 (12–23)
      Good

      Poor

      Poor
      Low

      Low

      Low
      • Taylor R.N.
      • Parker J.D.A.
      • Keefer K.V.
      • et al.
      Are gambling related cognitions in adolescence multidimensional? Factor structure of the Gambling Related Cognitions Scale.
      Canada (2014)DSM-IV-J (2–3 = at risk, ≥4 = problem gambling);

      GRCS
      • a)
        Cross-sectional
      • b)
        Several high schools sampled from a school district
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha, DSM -IV-J classification predicted with GRCS scores
      n = 1,490

      female: 57.7%

      age: 17.10 (16–18)
      Good

      Poor

      Poor
      High

      Low

      Low
      • Taylor R.N.
      • Parker J.D.A.
      • Keefer K.V.
      • et al.
      Gambling related cognitive distortions in adolescence: Relationships with gambling problems in typically developing and special needs students.
      Canada (2014)DSM-IV-J (2–3 = at risk; ≥4 = problem gambling);

      GRCS
      • a)
        Cross-sectional
      • b)
        Several secondary schools sampled from a school district
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha
      n = 2,004

      female: 57.7%

      age: 16.51 (14–18)
      Good

      Poor

      Poor
      High

      Low

      Low
      • Castrén S.
      • Grainger M.
      • Lahti T.
      • et al.
      At-risk and problem gambling among adolescents: A convenience sample of first-year junior high school students in Finland.
      Finland (2015)DSM-IV-MR-J (≥2 = at risk and problem gambling)
      • a)
        Cross-sectional
      • b)
        Convenience sample from 11 schools
      • c)
        In-class self-report questionnaire
      • d)
        Cronbach alpha, factor analysis, classification accuracy
      • e)
        Gambling frequency
      n = 988

      female: 46.8%

      age: 13.41 (12–15)
      Good

      Good

      Good
      High

      Low

      Low
      • Gavriel-Fried B.
      • Ronen T.
      Contribution of positivity ratio and self-control to reduce gambling severity among adolescents.
      Israel (2015)DSM-IV-MR-J (2–3 = at risk; ≥4 = probable pathologic gambling)
      • a)
        Cross-sectional
      • b)
        Convenience sample from six schools
      • c)
        In-class self-report questionnaire
      • d)
        Cronbach alpha
      n = 595

      female: 60%

      age: 15.13 (13–19)
      Good

      Poor

      Poor
      High

      Low

      Low
      • Molde H.
      • Pallesen S.
      • Bartone P.
      • et al.
      Prevalence and correlates of gambling among 16 to 19-year-old adolescents in Norway.
      Norway (2009)MAGS (3–4.5 = problem gambler, ≥5 = pathologic gambler)
      • a)
        Cross-sectional
      • b)
        Randomly selected (at the class level) high school sample
      • c)
        Online self-report questionnaire during designated school time
      • d)
        Item response theory: Differential Item Functioning
      n = 1,285 (1,351)

      female: 47.5%

      age: 17.3 (16–19)
      Good

      Good

      Poor
      Low

      Low

      Low
      • Kong G.
      • Tsai J.
      • Krishnan-Sarin S.
      • et al.
      A latent class analysis of pathological-gambling criteria among high school students: Associations with gambling, risk and health/functioning characteristics.
      USA (2014)MAGS (used according to DSM-IV and DSM-5 criteria)
      • a)
        Cross-sectional
      • b)
        Convenience sample from 10 local high schools
      • c)
        Self-report questionnaire
      • d)
        Latent class analysis
      • e)
        OR for spending >1 hour/week gambling
      n = 3,901 (4,523)

      female: 51.5%

      age: (<14–>18)
      Good

      Good

      Good
      Low

      Low

      Low
      • Foster D.W.
      • Hoff R.A.
      • Pilver C.E.
      • et al.
      Correlates of gambling on high-school grounds.
      USA (2015)MAGS (≥1 = at risk and problem gambling)
      • a)
        Cross-sectional
      • b)
        Convenience high school sample
      • c)
        Administration method not specified
      • d)
        Cronbach alpha
      n = 1,988

      female: 39.2%

      age: (9th to 12th grade)
      Good

      Poor

      Poor
      Low

      Unclear

      High
      • Tremblay J.
      • Stinchfield R.
      • Wiebe J.
      • Wynne H.
      Canadian Adolescent Gambling Inventory (CAGI) phase III final report.
      Canada (2010)CAGI (GPSS 0–1 = no problem; 2–5 = low-to-moderate severity; ≥6 = high severity);

      DSM-IV (1–3 = low gambling problem; ≥4 = problem gambling); SOGS-RA (2–3 = at-risk gambling; ≥4 = problem gambling); clinical interview (DSM-IV [same thresholds];

      CRAGS [four classes; ≥5 = problem gambling])
      • a)
        Cross-sectional
      • b)
        Phase II sample from schools, Phase III sample from local substance abuse and detention treatment centers
      • c)
        Self-report questionnaires/clinical interviews
      • d)
        Cronbach alpha, Intraclass correlations, principal component analysis, confirmatory factor analysis, discriminant function analysis, ROC analysis,
      • e)
        Correlations with references
      Phases II and III:

      n = 105 (66 and 39)

      female: 46.7% (51.5% and 38.5%)

      age: 14.9 and 15.6 (12–>18)

      Phase II school sample n = 864
      Good

      Good

      Good
      Low

      Low

      Low
      • Park H.S.
      • Jung S.Y.
      Development of a Gambling Addictive Behavior Scale for adolescents in Korea.
      South Korea (2012)Gambling Addictive Behavior Scale for Adolescents (no classifications specified)
      • a)
        Cross-sectional
      • b)
        Convenience stratified school sample classes selected randomly
      • c)
        Self-report questionnaire
      • d)
        Cronbach alpha, principal component analysis, ROC analysis
      • e)
        Correlation with other variables
      n = 299 (320)

      female: 40.8%

      age: not specified
      Good

      Good

      Good
      Low

      Unclear

      Low
      CAGI = Canadian Adolescent Gambling Inventory; CRAGS = Clinician Rating of Adolescent's Gambling Severity; DIS-IV = Diagnostic Interview Schedule, Version IV; DSM-IV-J/DSM-IV-(MR)-J = the Diagnostic Statistical Manual IV (Multiple Response format) adapted for Juveniles; GPSS = Gambling Problem Severity Subscale; GRCS = Gambling-Related Cognitions Scale; MAGS = Massachusetts Gambling Screen; MSOGST = Modified South Oaks Gambling Screen for Teens; NL-CLiP = short screen for problem gambling among children based on criteria identified in NODS-CLiP (NODS-CLiP: Diagnostic Screening for Gambling Disorders [NODS] Loss of Control, Lying and Preoccupation [CLiP]); ROC = receiver operating characteristic; SOGS-RA = the South Oaks Gambling Screen–Revised for Adolescents.
      a a) Design, b) sampling method, c) administration method, d) statistical methods related to reliability and validity, e) references of ARPG: comparator measurement.
      b Sample size in analysis and in total, gender distribution from analyzed sample (unless otherwise specified as total sample), mean age (range) in years.
      c QUADAS-2 assessment for applicability concerns: patient selection; index test; reference standard.
      d QUADAS-2 assessment of risk of bias concerning patient selection, index test, and flow and timing.
      Table 4Findings of studies on instruments for measurement of gambling problems among young people
      Ref.Country (year)Statistical results and author report of reliability/validity findings
      • Barnes G.M.
      • Welte J.W.
      • Hoffman J.H.
      • Tidwell M.-C.O.
      Gambling, alcohol, and other substance use among youth in the United States.
      USA (2009)Total number of items endorsed (on SOGS-RA, DSM-IV-MR-J, and DIS-IV) correlation with factor from principal component analysis r = .97
      • Mooss A.V.
      Gambling behaviours among youth involved in juvenile and family courts.
      USA (2009)SOGS-RA α = .85; correlation with gambling activities r = .57 and crime r = .26 (p < .001)
      • Taylor L.M.
      • Hillyard P.
      Gambling awareness for youth: An analysis of the “Don’t gamble away our future™” program.
      USA (2009)MSOGST α = .87
      • Wanner B.
      • Vitaro F.
      • Carbonneau R.
      • Tremblay R.E.
      Cross-lagged links among gambling, substance use, and delinquency from midadolescence to young adulthood: Additive and moderating effects of common risk factors.
      Canada (2009)SOGS-RA α = .78 (sample A) and α = .78 (sample B); correlation with SOGS-RA and SOGS r = .22 (p < .05) and .28 (p < .05);

      SOGS and SOGS-RA were metrically invariant, and thus psychologically comparable.
      • Welte J.W.
      • Barnes G.M.
      • Tidwell M.-C.O.
      • Hoffman J.H.
      Association between problem gambling and conduct disorder in a national survey of adolescent and young adults in the Unites States.
      USA (2009)SOGS-RA α = .74; SOGS-RA correlation with DSM-IV-MR-J r = .76
      • Barnes G.M.
      • Welte J.W.
      • Hoffman J.H.
      • Tidwell M.-C.O.
      Comparisons of gambling and alcohol use among college students and noncollege young people in the United States.
      USA (2010)SOGS-RA α = .74 (n = 2,274; age 14–21)
      • Hansen M.
      • Rossow I.
      Limited cash flow on slot machines: Effects of prohibition of note acceptors on adolescent gambling behavior.
      Norway (2010)At-risk and problem gambling displayed stability during 2004 and 2005 (preintervention), with a significant decrease after the removal bank note acceptors in 2006 (postintervention). Thus, the SOGS-RA displayed stability and sensitivity to change. No gender- or age-related differences were evident in the reduction of gambling problem prevalence (i.e., 2005–2006).
      • Barnes G.M.
      • Welte J.W.
      • Hoffman J.H.
      • Tidwell M.-C.O.
      The co-occurrence of gambling with substance use and conduct disorder among youth in the United States.
      USA (2011)SOGS-RA α = .72; DSM-IV-MR-J α = .71; DIS-IV α = .77; combined α = .89
      • Dussault F.
      • Brendgen M.
      • Vitaro F.
      • et al.
      Longitudinal links between impulsivity, gambling problems and depressive symptoms: A transactional model from adolescence to early adulthood.
      Canada (2011)SOGS-RA α = .76
      • Villella C.
      • Martinotti G.
      • Nicola M.D.
      • et al.
      Behavioural addictions in adolescents and young adults: Results from a prevalence study.
      Italy (2011)SOGS-RA α = .80
      • West R.L.
      The effect of extended family gambling behavior and family functioning on African American adolescent gambling.
      USA (2011)SOGS-RA α = .83; SOGS-RA correlation with gambling frequency r = .59
      • Ashrafioun L.
      • McCarthy A.
      • Rosenberg H.
      Assessing the impact of cue exposure on craving to gamble in university students.
      USA (2012)SOGS-RA α = .84; Gambling Urge Scale and SOGS-RA correlation r = .60 (p ≤ .001);

      Postexposure Gambling Urge Scale scores correlated significantly with SOGS-RA scores
      • Chiu E.Y.-W.
      • Woo K.
      Problem gambling in Chinese American adolescents: Characteristics and risk factors.
      USA (2012)SOGS-RA α = .80
      • Liu W.
      • Lee G.P.
      • Goldweber A.
      • et al.
      Impulsivity trajectories and gambling in adolescence among urban male youth.
      USA (2012)SOGS-RA α = .71
      • Walther B.
      • Morgenstern M.
      • Hanewinkel R.
      Co-occurrence of addictive behaviours: Personality factors related to substance use, gambling and computer gaming.
      Germany (2012)SOGS-RA α = .77
      • Wong S.S.K.
      • Tsang S.K.M.
      Validation of the Chinese version of the Gamblers’ Belief Questionnaire (GBQ-C).
      Hong Kong (2012)Chinese version of the Gamblers Belief Questionnaire α = .91; Chinese version of the Gamblers Belief Questionnaire correlation range with other scales .40–.75
      • Chiesi F.
      • Donati M.A.
      • Galli S.
      • Primi C.
      The suitability of the South Oaks Gambling Screen-Revised for Adolescents (SOGS-RA) as a screening tool: Irt-based evidence.
      Italy (2013)Factor loadings range .53–.83 (p < .001), CFI = .96; TLI = .97; RMSEA = .03;

      Suggests single factor structure for SOGS-RA. Items “feeling bad about money lost” and “gambling more than planned” had the highest endorsement rate. Absence from school due to betting, borrowing money, and stealing for betting were the most discriminative. Items “lying about winning” and “wanting to stop gambling” had the lowest discrimination. Majority of items had good discrimination. Screen accurately measures medium to high levels of problem gambling (i.e., item severity located along intended range).
      • Colasante E.
      • Gori M.
      • Bastiani L.
      • et al.
      Italian adolescent gambling behavior: Psychometric evaluation of the South Oaks Gambling Screen: Revised for Adolescents (SOGS-RA) among a sample of Italian students.
      Italy (2013)Multiple correspondence analysis principle component (eigenvalue = 3.875) explained 32.3% of variance; test–retest κ coefficient range = .53–.80; α (males) = .786; α (females) = .707;

      Suggested single factor structure for SOGS-RA. The least endorsed items were “absent from school due to betting” and “borrowed or stolen for bets or debts.” The most endorsed items were “gambling more than intended” and “feeling bad about the amount bet.” The SOGS-RA score was positively associated to gambling frequency. No gender differences were evident in item endorsements. The SOGS-RA seems to be stable over time.
      • Donati M.A.
      • Chiesi F.
      • Primi C.
      A model to explain at-risk/problem gambling among male and female adolescents: Gender similarities and differences.
      Italy (2013)SOGS-RA α = .73
      • Lee G.P.
      • Stuart E.A.
      • Ialongo N.S.
      • Martins S.S.
      Parental monitoring trajectories and gambling among a longitudinal cohort of urban youth.
      USA (2013)SOGS-RA α = .61–.72
      • Parker J.D.A.
      • Summerfeldt L.J.
      • Kloosterman P.H.
      Gambling behavior in adolescents with learning disorders.
      Canada (2013)SOGS-RA α = .94; DSM-IV-J α = .93; SOGS-RA and DSM-IV-J correlation r = .64;

      DSM-IV-J is a more conservative instrument for measuring pathologic gambling than SOGS-RA.
      • Primi C.
      • Donati M.A.
      • Bellini I.
      • et al.
      Measuring the attitude towards the profitability of gambling: The psychometric properties of the Gambling Attitude Scale.
      Italy (2013)SOGS-RA α = .73; Gambling Attitude Scale α = .80;

      Gambling Attitude Scale discriminated nonproblem gamblers (more cautious perception) from at-risk and problem gamblers. Problem gamblers scored higher on the items from the profitability factor than at-risk and nonproblem gamblers.
      • Tozzi L.
      • Akre C.
      • Fleury-Schubert A.
      • Suris J.-C.
      Gambling among youths in Switzerland and its association with other addictive behaviours.
      Switzerland (2013)SOGS-RA (eight items) α = .70
      • Wong G.
      • Zane N.
      • Saw A.
      • Chan A.K.K.
      Examining gender differences for gambling engagement and gambling problems among emerging adults.
      USA (2013)SOGS-RA α = .82
      Canada (2014)SOGS-RA (6 items) α = .70;

      Most endorsed items in 2013 were “gambling more than planned” (1.7%) and “experiencing problems with family or school due to gambling” (1.5%). Females were significantly less likely than males to endorse the previously mentioned items (p < .05).
      • Bray B.C.
      • Lee G.P.
      • Liu W.
      • et al.
      Transitions in gambling participation during late adolescence and young adulthood.
      USA (2014)SOGS-RA α range during different years of administration .60–.72
      • Cook S.
      • Turner N.E.
      • Ballon B.
      • et al.
      Problem gambling among Ontario students: Associations with substance abuse, mental health problems, suicide attempts, and delinquent behaviours.
      Canada (2014)SOGS-RA (six items) α = .71, AUC = .80 (concordance of short version and full version of SOGS-RA);

      The short version of the SOGS-RA may overestimate prevalence rates.
      • Donati M.A.
      • Primi C.
      • Chiesi F.
      Prevention of problematic gambling behavior among adolescents: Testing the efficacy of an integrative intervention.
      Italy (2014)McNemar χ2 (1, N = 88) = 8.77, p ˂ .05;

      At baseline the prevalence of ARPG (measured with SOGS-RA) in the training group was 41%. After training (approximately 6 months after pretest) a significant and medium in size reduction in the prevalence of ARPG was evident (prevalence at follow up: 28%). The intervention intended to increase correct knowledge about gambling, reduce misconceptions, economic optimistic view of gambling profitability, and superstitious beliefs. The intervention contained activities, PowerPoint slides, a video, and collective discussions.
      • Canale N.
      • Vieno A.
      • Griffiths M.D.
      • et al.
      Trait urgency and gambling in young people by age: The mediating role of decision making processes.
      Italy (2015)SOGS-RA α = .73 (CI = .70/.75)
      • Carbonneau R.
      • Vitaro F.
      • Brendgen M.
      • Tremblay R.E.
      Variety of gambling activities from adolescence to age 30 and association with gambling problems: A 15-year longitudinal study of a general population sample.
      Canada (2015)SOGS-RA (at age 15) α = .76
      • Chan A.K.K.
      • Zane N.
      • Wong G.M.
      • Song A.V.
      Personal gambling expectancies among Asian American and White American college students.
      USA (2015)SOGS-RA α = .67; omega = .68;

      Asian-Americans were more likely to endorse the following items compared to white Americans: lying about winning; gambling more than intended; felt bad about money bet; hidden any signs of gambling.
      • Míguez M.C.
      • Becona E.
      Do cigarette smoking and alcohol consumption associate with cannabis use and problem gambling among Spanish adolescents?.
      Spain (2015)SOGS-RA α = .83
      • Sheela P.S.
      • Choo W.-Y.
      • Goh L.Y.
      • Tan C.P.L.
      Gambling risk amongst adolescents: Evidence from a school-based survey in the Malaysian setting.
      Malaysia (2015)SOGS-RA α = .77
      • Delfabbro P.
      • King D.
      • Lambos C.
      • Puglies S.
      Is video-gaming playing a risk factor for pathological gambling in Australian adolescents?.
      Australia (2009)DSM-IV-J α = .82
      • Skokauskas N.
      • Burba B.
      • Freedman D.
      An assessment of the psychometric properties of Lithuanian versions of DSM-IV-MR-J and SOGS-RA.
      Lithuania (2009)DSM α = .80; SOGS-RA α = .75; DSM-IV-MR-J and SOGS-RA correlation r = .892 (p < .001); SOGS-RA κ = .833 (p < .001);

      SOGS-RA is more liberal in classifying gambling pathology. SOGS-RA classified 34 of 35 pathologic gamblers correctly, using the DSM-IV-MR-J as the reference. DSM distinguished between social, at-risk, and pathologic gamblers. DSM item on “Escape” identified as best discriminator. Those who gambled at least 1 per week scored significantly higher than participants who gambled more seldom. SOGS-RA sensitivity = .97 (34/35), specificity .986, false-positive rate = .20 and false-negative rate = .00015. DSM pathologic gamblers were likely to spend more money on gambling than nonpathologic gamblers.
      • Dowling N.
      • Jackson A.C.
      • Thomas S.A.
      • Frydenberg E.
      Children at risk of developing problem gambling: Final report. A report to gambling research Australia in fulfilment of Tender 103/06.
      Australia (2010)DSM-IV-MR-J α = .78
      • Brunelle N.
      • Leclerc D.
      • Cousineau M.-M.
      • et al.
      Internet gambling, substance use, and delinquent behavior: An adolescent deviant behavior involvement pattern.
      Canada (2012)DSM-IV-MR-J α = .75
      • Floros G.D.
      • Siomos K.
      • Fisoun V.
      • Geroukalis D.
      Adolescent online gambling: The impact of parental practices and correlates with online activities.
      Greece (2013)DSM-IV-MR-J α = .91
      • Lepper J.
      • Haden B.
      Testing NLCLiP: Validation of estimates of rates of non-problematic and problematic gambling in a sample of British schoolchildren.
      England (2013)Nonproblem/problem κ = .633 (only gamblers) and .778 (gamblers and nongamblers);

      There was a lack of consistency in responses of comparable questions in the two instruments; NL-CLiP is accurate in classifying nonproblematic and problematic gamblers, but not in distinguishing between at-risk and problem gamblers (DSM-IV-MR-J as reference)
      • Tolchard B.
      • Delfabbro P.
      The Victorian Gambling Screen: Validity and reliability in an adolescent population.
      Australia (2013)VGS α = .95 (split half α = .92 and .88); DSM-IV-J α = .92 (split half α = .84 and .90); VGS-DSM correlation r = .65 (p < .001);

      VGS classified 31 participants as problem gamblers (nine unidentified by DSM). Similarly DSM classified 41 participants (19 unidentified by VGS).
      • Cheung N.W.T.
      Low self-control and co-occurrence of gambling with substance use and delinquency among Chinese adolescents.
      China (2014)DSM-IV-J α = .82
      • Taylor R.N.
      • Parker J.D.A.
      • Keefer K.V.
      • et al.
      Are gambling related cognitions in adolescence multidimensional? Factor structure of the Gambling Related Cognitions Scale.
      Canada (2014)DSM-IV-J α = .90; GRCS α = .97 (subscale α range .77–.91); Average correlation with five-factor model of GRCS = .82;

      At-risk and problem gamblers scored significantly higher than nonproblem gamblers on the entire GRCS scale and all its subscales separately with significant gender interaction evident throughout analyses (males scoring higher); 32% of the variance in DSM-IV-J was explained by the GRCS with hierarchical multiple regression. Inability to stop gambling, illusion of control and gambling related expectancies subscales were significant unique predictors of at-risk and problem gambling.
      • Taylor R.N.
      • Parker J.D.A.
      • Keefer K.V.
      • et al.
      Gambling related cognitive distortions in adolescence: Relationships with gambling problems in typically developing and special needs students.
      Canada (2014)DSM-IV-J α = .90; GRCS α = .97 (subscale α range .80–.91)
      • Castrén S.
      • Grainger M.
      • Lahti T.
      • et al.
      At-risk and problem gambling among adolescents: A convenience sample of first-year junior high school students in Finland.
      Finland (2015)DSM-IV-MR-J α = .86; sensitivity of items = .22–.78; specificity of items = .94–.99; gambling often or sometimes odds ratio (95% CI) for ARPG = 5.78 (3.0–11.0);

      Scree plot of exploratory factor analysis supports 1-factor solution, accounting for 40.1% of variance and correlated positively with the psychological states preoccupation, tolerance, withdrawal, loss of control, escape, and chasing. Illegal acts, tolerance, loss of control, and lies were the most commonly endorsed and most sensitive items in identifying ARPG. Item on illegal acts was the least specific. Lowest sensitivity was for items on escape, risked job/education/relationship, and withdrawal.
      • Gavriel-Fried B.
      • Ronen T.
      Contribution of positivity ratio and self-control to reduce gambling severity among adolescents.
      Israel (2015)DSM-IV-MR-J α = .91
      • Molde H.
      • Pallesen S.
      • Bartone P.
      • et al.
      Prevalence and correlates of gambling among 16 to 19-year-old adolescents in Norway.
      Norway (2009)All the MAGS items displayed different functioning between males and females. Males were more likely to endorse each item than females, given otherwise equal scores for the latent variable. This indicates that the criteria are more valid for males than females.
      • Kong G.
      • Tsai J.
      • Krishnan-Sarin S.
      • et al.
      A latent class analysis of pathological-gambling criteria among high school students: Associations with gambling, risk and health/functioning characteristics.
      USA (2014)Odds ratios (95% CI): ACG = 7.66 (4.34–13.53); ANCG = 70.84 (43.41–115.62); PrG = 11.34 (3.82–33.61);

      Using MAGS, Latent Class Analysis indicated a four-class solution to be optimal for DSM IV and five criteria: low-risk (LG; most common), at-risk chasing gambling (ACG), at-risk negative consequences gambling (ANCG), and problem gambling (PrG; least common). Inclusion/exclusion of item on illegal acts had little effect on the classification of gambling groups. LG was characterized by low probability of endorsement for all items. ACG was characterized by elevated probability of endorsement for “win back lost money” and “gambling more money over time.” ANCG was characterized by elevated probability to endorse “losing/jeopardizing relationship or career opportunities,” “committing illegal acts,” “turning to other financial sources,” and “unsuccessful attempts to reduce or quit.” PrG was characterized by elevated probability to endorse all 10 items. Compared with LG, other gambling classes were more likely to spend more than 1 hour/week gambling (p ≤ .025).
      • Foster D.W.
      • Hoff R.A.
      • Pilver C.E.
      • et al.
      Correlates of gambling on high-school grounds.
      USA (2015)MAGS α = .92
      • Tremblay J.
      • Stinchfield R.
      • Wiebe J.
      • Wynne H.
      Canadian Adolescent Gambling Inventory (CAGI) phase III final report.
      Canada (2010)CAGI α's for four factors = .90; .90; .83; .87; test–retest intraclass correlations = .77; .90; .83; .87; CRAGS and DSM-IV measures correlations r ≤ .89; CAGI subscale correlations with gambling involvement measures r = .14–.67;

      Endorsement of consequence items “stealing to gamble,” “feeling guilty about gambling behaviors,” and “gambling for longer periods than planned” were much higher for phase III sample than phase II. Principal component analysis suggested four-factor solution (psychological consequences; social consequences; financial consequences; loss of control) explaining 67.3% of variance, with a balanced weight among factors. Factor correlation between .62 and .69. Confirmatory factor analysis suggests reasonably good model fit. High congruency between classifications of gold standards (DSM-IV self-rated and clinical interview, and CRAGS). Discriminant function analysis and ROC analysis revealed nine-item subscale (GPSS) to be optimal for classification performance. Measures of cognitive distortions, decision-making, and self-efficacy correlated below r = .30 with CAGI subscales. Strongest correlates related to convergent validity of problem gambling were risk taking and self-control (all r > .30), followed by impulsivity.
      • Park H.S.
      • Jung S.Y.
      Development of a Gambling Addictive Behavior Scale for adolescents in Korea.
      South Korea (2012)Gambling Addictive Behavior Scale for Adolescents α = .94; α's for subscales = .90; .89; .88; .90;

      Assessment by expert panel yielded content validity index of 94.3%. Final scale composed of 25 items, loading onto four factors explaining 54.9% of variance (loss of control; life dysfunction from gambling addiction; gambling experience; social dysfunction from problem gambling). Scale categorizes individuals as nongambling (AUC = .71), nonproblem gambling (AUC = .75), and problem gambling (AUC = .74) group (p ≤ .001). Factors correlated significantly with irrational gambling beliefs, gambling behavior
      • Kim H.J.
      Study of the factors influencing adolescent gambling behavior.
      , and the Addictive Personality subscale of the Eysenck Personality Questionnaire and self-control (p < .001).
      ARPG = at-risk and problem gambling; AUC = Area Under ROC Curve analysis; CAGI = Canadian Adolescent Gambling Inventory; CFI = comparative fit index; CRAGS = Clinician Rating of Adolescent's Gambling Severity; DIS-IV = Diagnostic Interview Schedule, Version IV; DSM-IV-J/DSM-IV-(MR)-J = the Diagnostic Statistical Manual IV (Multiple Response format) adapted for Juveniles; GPSS = Gambling Problem Severity Subscale; GRCS = Gambling-Related Cognitions Scale; MAGS = Massachusetts Gambling Screen; MSOGST = Modified South Oaks Gambling Screen for Teens; NL-CLiP = short screen for problem gambling among children based on criteria identified in NODS-CLiP (NODS-CLiP: Diagnostic Screening for Gambling Disorders [NODS] Loss of Control, Lying and Preoccupation [CLiP]); ROC = receiver operating characteristic; SOGS-RA = the South Oaks Gambling Screen–Revised for Adolescents; RMSEA = root mean square error of approximation; TLI = Tucker-Lewis index; VGS = Victorian Gambling Screen.

      Results

      Of the 50 articles, 33 dealt with the SOGS-RA and 12 with the DSM-IV-J or DSM-IV-MR-J [collectively referred to as DSM-IV-(MR)-J] as the primary ARPG instrument. Three articles evaluated the MAGS, one evaluated the CAGI, and one evaluated the Gambling Addictive Behavior Scale for Adolescents (GABSA). Four studies primarily using the SOGS-RA also used one of the two DSM instruments, and one study primarily using the DSM-IV-MR-J also used the SOGS-RA. All samples were general population based, except the study on the CAGI, which used a sample from a substance use center. About half of the articles (n = 26) used cross-sectional data from local schools, including two pairs of articles based on the same data sets. Eight articles had a longitudinal study design. Five articles had cross-sectional nationally representative samples and another nine articles used cross-sectional data from nationally representative school samples, including one pair of articles using partially the same data set. Two studies had local school samples and used experimental designs. Articles originated from 17 different countries, including 16 from the United States, 10 from Canada, 7 from Italy, 3 from Australia, and 2 from Norway and 1 article from China, England, Finland, Germany, Greece, Hong Kong, Israel, Lithuania, Malaysia, South Korea, Spain, and Switzerland.
      Table 2 summarizes the risk of bias in the articles. Almost half (n = 24) of the articles had done patient selection well. Most (n = 38) had conducted the index test properly, and in nearly all articles (n = 43), participants were maintained in analyses.
      The applicability of the articles to this review was more variable (Table 2). In most (n = 43) of the articles, patient selection suited our study question, whereas only some had good applicability related to evaluating the index test (n = 9) or for using a reference measurement (n = 12).

      South Oaks Gambling Screen Revised for Adolescents

      In articles evaluating the SOGS-RA as the primary adolescent ARPG instrument, high risk of bias was often evident for patient selection but not for other items. Although the sample was often appropriate, only a few of the SOGS-RA articles primarily investigated instrument properties or used a reference (Table 2).
      Estimates of internal consistency (Cronbach α) for the SOGS-RA ranged from .60 [
      • Bray B.C.
      • Lee G.P.
      • Liu W.
      • et al.
      Transitions in gambling participation during late adolescence and young adulthood.
      ] to .94 [
      • Parker J.D.A.
      • Summerfeldt L.J.
      • Kloosterman P.H.
      Gambling behavior in adolescents with learning disorders.
      ] (M = .76). The instrument seems stable over time [
      • Colasante E.
      • Gori M.
      • Bastiani L.
      • et al.
      Italian adolescent gambling behavior: Psychometric evaluation of the South Oaks Gambling Screen: Revised for Adolescents (SOGS-RA) among a sample of Italian students.
      ]. Cronbach α for the Modified South Oaks Gambling Screen for Teens (MSOGST) was .87 [
      • Wanner B.
      • Vitaro F.
      • Carbonneau R.
      • Tremblay R.E.
      Cross-lagged links among gambling, substance use, and delinquency from midadolescence to young adulthood: Additive and moderating effects of common risk factors.
      ]. The MSOGST includes 20 items (12 items used for scoring; score range 0–20) with rewording for adolescents [
      • Taylor L.
      Evaluating an adolescent measure of gambling pathology.
      ].
      The SOGS-RA displayed good classification accuracy, with a sensitivity of .97, specificity of .99, false-positive rate of .20, and false-negative rate of .00015 when using the DSM-IV-MR-J as a reference [
      • Skokauskas N.
      • Burba B.
      • Freedman D.
      An assessment of the psychometric properties of Lithuanian versions of DSM-IV-MR-J and SOGS-RA.
      ]. “Feeling bad about money lost” and “Gambling more than planned” were the most commonly reported items in two studies [
      • Chiesi F.
      • Donati M.A.
      • Galli S.
      • Primi C.
      The suitability of the South Oaks Gambling Screen-Revised for Adolescents (SOGS-RA) as a screening tool: Irt-based evidence.
      ,
      • Colasante E.
      • Gori M.
      • Bastiani L.
      • et al.
      Italian adolescent gambling behavior: Psychometric evaluation of the South Oaks Gambling Screen: Revised for Adolescents (SOGS-RA) among a sample of Italian students.
      ], whereas another study found highest rates for items “Gambling more than planned” and “Experiencing problems with family or school due to gambling” [
      ]. For both items, females were less likely to endorse them than males [
      ], contradicting the finding that item endorsement differences would not exist between genders [
      • Colasante E.
      • Gori M.
      • Bastiani L.
      • et al.
      Italian adolescent gambling behavior: Psychometric evaluation of the South Oaks Gambling Screen: Revised for Adolescents (SOGS-RA) among a sample of Italian students.
      ]. The least often reported items were “Absent from school due to betting” and “Borrowed or stolen for bets or debts” [
      • Colasante E.
      • Gori M.
      • Bastiani L.
      • et al.
      Italian adolescent gambling behavior: Psychometric evaluation of the South Oaks Gambling Screen: Revised for Adolescents (SOGS-RA) among a sample of Italian students.
      ]. Items “Lying about winning,” “Gambling more than planned,” “Felt bad about money lost,” and “Hidden any signs of gambling” were more likely endorsed by Asian-Americans than by white Americans [
      • Chan A.K.K.
      • Zane N.
      • Wong G.M.
      • Song A.V.
      Personal gambling expectancies among Asian American and White American college students.
      ]. Overall, the SOGS-RA items had good discrimination, and the item severity was appropriate for screening ARPG [
      • Chiesi F.
      • Donati M.A.
      • Galli S.
      • Primi C.
      The suitability of the South Oaks Gambling Screen-Revised for Adolescents (SOGS-RA) as a screening tool: Irt-based evidence.
      ].
      Two Italian studies suggest a single factor structure of the SOGS-RA [
      • Chiesi F.
      • Donati M.A.
      • Galli S.
      • Primi C.
      The suitability of the South Oaks Gambling Screen-Revised for Adolescents (SOGS-RA) as a screening tool: Irt-based evidence.
      ,
      • Colasante E.
      • Gori M.
      • Bastiani L.
      • et al.
      Italian adolescent gambling behavior: Psychometric evaluation of the South Oaks Gambling Screen: Revised for Adolescents (SOGS-RA) among a sample of Italian students.
      ]. The SOGS-RA correlated positively with the scope of gambling activities (r = .57) [
      • Mooss A.V.
      Gambling behaviours among youth involved in juvenile and family courts.
      ] and gambling frequency (r = .59) [
      • West R.L.
      The effect of extended family gambling behavior and family functioning on African American adolescent gambling.
      ,
      • Colasante E.
      • Gori M.
      • Bastiani L.
      • et al.
      Italian adolescent gambling behavior: Psychometric evaluation of the South Oaks Gambling Screen: Revised for Adolescents (SOGS-RA) among a sample of Italian students.
      ]. The SOGS-RA score correlation with gambling related crimes was .26 [
      • Derevensky J.
      • Gupta R.
      Measuring gambling problems amongst adolescents: Current status and future directions.
      ]. Correlations with the SOGS-RA and the DSM-IV-J (r = .64) [
      • Parker J.D.A.
      • Summerfeldt L.J.
      • Kloosterman P.H.
      Gambling behavior in adolescents with learning disorders.
      ] and the DSM-IV-MR-J (r = .74–.892) [
      • Welte J.W.
      • Barnes G.M.
      • Tidwell M.-C.O.
      • Hoffman J.H.
      Association between problem gambling and conduct disorder in a national survey of adolescent and young adults in the Unites States.
      ,
      • Skokauskas N.
      • Burba B.
      • Freedman D.
      An assessment of the psychometric properties of Lithuanian versions of DSM-IV-MR-J and SOGS-RA.
      ] indicate convergence. The SOGS-RA correlated positively with the Chinese version of the Gambling Belief Questionnaire [
      • Ashrafioun L.
      • McCarthy A.
      • Rosenberg H.
      Assessing the impact of cue exposure on craving to gamble in university students.
      ] and the Gambling Urge Scale (r = .60) [
      • Ashrafioun L.
      • McCarthy A.
      • Rosenberg H.
      Assessing the impact of cue exposure on craving to gamble in university students.
      ]. In a longitudinal study [
      • Wanner B.
      • Vitaro F.
      • Carbonneau R.
      • Tremblay R.E.
      Cross-lagged links among gambling, substance use, and delinquency from midadolescence to young adulthood: Additive and moderating effects of common risk factors.
      ], the SOGS-RA correlated positively with the SOGS; the two were considered metrically invariant, thus psychologically comparable.
      Finally, the SOGS-RA indicated sensitivity to change, as the prevalence of ARPG decreased after the removal of banknote acceptors in slot machines, compared to the two preceding baseline years during which ARPG prevalence remained stable [
      • Hansen M.
      • Rossow I.
      Limited cash flow on slot machines: Effects of prohibition of note acceptors on adolescent gambling behavior.
      ]. The SOGS-RA also detected a decrease in ARPG prevalence following intervention [
      • Donati M.A.
      • Primi C.
      • Chiesi F.
      Prevention of problematic gambling behavior among adolescents: Testing the efficacy of an integrative intervention.
      ].

      DSM-IV-(MR)-J

      Of the 12 articles evaluating the DSM-IV-(MR)-J, six had a high risk of bias concerning patient selection, while four instances of unclear risk of bias were evident. Otherwise, the risk of bias was low (Table 2). In patient selection, these studies had good applicability. Two studies investigated instrument properties, and in four cases, a reference standard was used.
      Cronbach α ranged from .75 [
      • Sheela P.S.
      • Choo W.-Y.
      • Goh L.Y.
      • Tan C.P.L.
      Gambling risk amongst adolescents: Evidence from a school-based survey in the Malaysian setting.
      ] to .93 [
      • Colasante E.
      • Gori M.
      • Bastiani L.
      • et al.
      Italian adolescent gambling behavior: Psychometric evaluation of the South Oaks Gambling Screen: Revised for Adolescents (SOGS-RA) among a sample of Italian students.
      ] (M = .85). Items on illegal acts, tolerance, loss of control, and lying had the highest endorsement rates and were the most sensitive in identifying ARPG, whereas items on escape, withdrawal, and risking job, education, or relationship were the least sensitive [
      • Castrén S.
      • Grainger M.
      • Lahti T.
      • et al.
      At-risk and problem gambling among adolescents: A convenience sample of first-year junior high school students in Finland.
      ]. One study suggested a single-factor structure for the DSM-IV-MR-J [
      • Castrén S.
      • Grainger M.
      • Lahti T.
      • et al.
      At-risk and problem gambling among adolescents: A convenience sample of first-year junior high school students in Finland.
      ]. The DSM-IV-MR-J distinguished between at-risk and pathologic gamblers, and the escape item was the best discriminator [
      • Skokauskas N.
      • Burba B.
      • Freedman D.
      An assessment of the psychometric properties of Lithuanian versions of DSM-IV-MR-J and SOGS-RA.
      ]. The DSM-IV-J correlated significantly with the Victorian Gambling Screen (r = .65) [
      • Tolchard B.
      • Delfabbro P.
      The Victorian Gambling Screen: Validity and reliability in an adolescent population.
      ], classifying 41 participants as problem gamblers (19 unidentified by the Victorian Gambling Screen) [
      • Tolchard B.
      • Delfabbro P.
      The Victorian Gambling Screen: Validity and reliability in an adolescent population.
      ]. The Gambling-Related Cognitions Scale explained 32% of the DSM-IV-J variance [
      • Taylor R.N.
      • Parker J.D.A.
      • Keefer K.V.
      • et al.
      Are gambling related cognitions in adolescence multidimensional? Factor structure of the Gambling Related Cognitions Scale.
      ]. The DSM-IV-MR-J score was positively associated with gambling frequency [
      • Castrén S.
      • Grainger M.
      • Lahti T.
      • et al.
      At-risk and problem gambling among adolescents: A convenience sample of first-year junior high school students in Finland.
      ]. In line with previous research, the DSM-IV-J was more conservative than the SOGS-RA [
      • Parker J.D.A.
      • Summerfeldt L.J.
      • Kloosterman P.H.
      Gambling behavior in adolescents with learning disorders.
      ].

      Massachusetts Gambling Screen

      Three articles concerning the MAGS had low risk of bias for patient selection, whereas two articles had low risk of bias for both the conduct of the index test and flow and timing. Two of the articles had high applicability for the index test (Table 2).
      Cronbach α for the MAGS was .92 [
      • Foster D.W.
      • Hoff R.A.
      • Pilver C.E.
      • et al.
      Correlates of gambling on high-school grounds.
      ]. Two studies [
      • Molde H.
      • Pallesen S.
      • Bartone P.
      • et al.
      Prevalence and correlates of gambling among 16 to 19-year-old adolescents in Norway.
      ,
      • Kong G.
      • Tsai J.
      • Krishnan-Sarin S.
      • et al.
      A latent class analysis of pathological-gambling criteria among high school students: Associations with gambling, risk and health/functioning characteristics.
      ] were concerned with instrument psychometric properties. ARPG classification was associated to a stronger likelihood of gambling at least 1–2 hours/week compared to low-risk gamblers [
      • Kong G.
      • Tsai J.
      • Krishnan-Sarin S.
      • et al.
      A latent class analysis of pathological-gambling criteria among high school students: Associations with gambling, risk and health/functioning characteristics.
      ]. All items had different functioning between genders, such that males were more likely to endorse each item, given otherwise equal scores for the latent variable [
      • Molde H.
      • Pallesen S.
      • Bartone P.
      • et al.
      Prevalence and correlates of gambling among 16 to 19-year-old adolescents in Norway.
      ]. Latent class analysis suggested a four-class solution for the DSM criteria (measured with the MAGS): low-risk gambling, at-risk chasing gambling, at-risk negative consequences gambling, and problem gambling [
      • Kong G.
      • Tsai J.
      • Krishnan-Sarin S.
      • et al.
      A latent class analysis of pathological-gambling criteria among high school students: Associations with gambling, risk and health/functioning characteristics.
      ].

      Canadian Adolescent Gambling Inventory

      The report on the CAGI [
      • Tremblay J.
      • Stinchfield R.
      • Wiebe J.
      • Wynne H.
      Canadian Adolescent Gambling Inventory (CAGI) phase III final report.
      ] had low risk of bias on all domains and good overall applicability. It is a comprehensive assessment of the scale beyond the scope of this systematic review, so we provide a brief summary. Cronbach α for the subscales ranged from .83 to .90. The correlation with the DSM-IV criteria and the Clinician Rating of Adolescent's Gambling Severity was ≥.89, and correlation with gambling involvement measures ranged from .14 to .67. Principal component analysis and confirmatory factor analysis suggested a four-factor solution (psychological consequences, social consequences, financial consequences, and loss of control). The nine-item Gambling Problem Severity Subscale (GPSS) was optimal for classification performance as defined by discriminant function analysis and receiver operating characteristic (ROC) analysis.

      Gambling Addictive Behavior Scale for Adolescents

      The article on the GABSA [
      • Park H.S.
      • Jung S.Y.
      Development of a Gambling Addictive Behavior Scale for adolescents in Korea.
      ] had an unclear risk of bias for conducting the index test but otherwise low risk and overall good applicability.
      The developmental report of the GABSA suggested the scale to have high internal consistency, with an overall Cronbach α of .94, and subscale α ranging from .88 to .90. The scale displayed significant (p ≤ .001) classification accuracy (area under ROC curve analysis [AUC]) of .71 for the nongambling group, .75 for the nonproblem gamblers, and .74 for the problem gamblers, using reference scores from the Addictive Personality subscale of the Eysenck Personality Questionnaire. As for the psychometric properties, the scale loaded onto four factors (loss of control, life dysfunction from gambling addiction, gambling experience, and social dysfunction from problem gambling) explaining 55% of variance. The scale further displayed convergent validity by significant positive correlations with addictive personality, irrational gambling beliefs, and gambling behavior.

      Discussion

      Our review aimed to clarify which instruments measuring ARPG among youth are reliable and valid in light of reported estimates of internal consistency, classification accuracy, and psychometric properties. Five ARPG instruments were examined; 3 of the 33 SOGS-RA articles and 2 of the 12 DSM-IV-(MR)-J articles investigated instrument properties. Two of the three MAGS articles looked at instrument properties. The remaining articles concerned the development of the CAGI and the GABSA.

      South Oaks Gambling Screen Revised for Adolescents

      Most (58%) of the SOGS-RA articles had potentially biased sampling procedures. The reviewed estimates for internal consistency were similar to earlier findings [
      • Winters K.C.
      • Stinchfield R.
      • Fulkerson J.
      Towards the development of an adolescent gambling problem severity scale.
      ,
      • Welte J.W.
      • Barnes G.M.
      • Tidwell M.C.O.
      • Hoffman J.H.
      The prevalence of problem gambling among US adolescents and young adults: Results from a national survey.
      ,
      • Poulin C.
      An assessment of the validity and reliability of the SOGS-RA.
      ], although variation was evident. Estimates of correlations with the SOGS-RA and external references (e.g., gambling frequency) and the corresponding DSM-IV-(MR)-J instruments parallel previous findings [
      • Derevensky J.L.
      • Gupta R.
      Prevalence estimates of adolescent gambling: A comparison of the SOGS-RA, DSM-IV-J, and the GA 20 questions.
      ,
      • Winters K.C.
      • Stinchfield R.
      • Fulkerson J.
      Towards the development of an adolescent gambling problem severity scale.
      ].
      Specific items are problematic in population-based studies because of uneven endorsement rates. Controversially, there was evidence suggesting no item endorsement differences between genders [
      • Colasante E.
      • Gori M.
      • Bastiani L.
      • et al.
      Italian adolescent gambling behavior: Psychometric evaluation of the South Oaks Gambling Screen: Revised for Adolescents (SOGS-RA) among a sample of Italian students.
      ] but also that females are less likely to endorse items than males [
      ]. The SOGS-RA was psychologically parallel to the SOGS [
      • Wanner B.
      • Vitaro F.
      • Carbonneau R.
      • Tremblay R.E.
      Cross-lagged links among gambling, substance use, and delinquency from midadolescence to young adulthood: Additive and moderating effects of common risk factors.
      ], contradicting the idea of uniqueness of youth gambling compared to adult gambling. The results on the SOGS-RA highlight the lack of a unique conceptualization of gambling problems for adolescents and are inconclusive concerning gender differences [
      • Poulin C.
      An assessment of the validity and reliability of the SOGS-RA.
      ].
      Almost half of the SOGS-RA articles used ethnically diverse samples, consisting African-Americans, Hispanics, Asian-Americans, and samples from Hong Kong and Malaysia. Although these studies were not primarily testing the SOGS-RA, studies with different ethnicities emerged, as previously deemed necessary [
      • Blinn-Pike A.
      • Worthy S.L.
      • Jonkman J.N.
      Adolescent gambling: A review of an emerging field of research.
      ]. Cronbach α estimates among samples with high rates of African-Americans ranged from .60 to .85, which is moderate at best, but within the same range as previous estimates. Two items endorsed more often by Asian-Americans than white Americans concerned minimizing other people's worries about the respondents gambling, suggesting a difference in attitudes toward the social aspect of gambling [
      • Chan A.K.K.
      • Zane N.
      • Wong G.M.
      • Song A.V.
      Personal gambling expectancies among Asian American and White American college students.
      ].

      DSM-IV-(MR)-J

      Estimates of internal consistency parallel previous findings of about .80 [
      • Fisher S.
      Measuring pathological gambling in children: The case of fruit machines in the UK.
      ,
      • Fisher S.
      Developing the DSM-IV-DSM-IV criteria to identify adolescent problem gambling in non-clinical populations.
      ], which is considered satisfactory. Two DSM-IV-MR-J articles were primarily concerned with instrument properties and indicated that items are not equivalent in their ability to detect problem gambling, and variabilities in item endorsement rates and sensitivity are evident [
      • Skokauskas N.
      • Burba B.
      • Freedman D.
      An assessment of the psychometric properties of Lithuanian versions of DSM-IV-MR-J and SOGS-RA.
      ,
      • Castrén S.
      • Grainger M.
      • Lahti T.
      • et al.
      At-risk and problem gambling among adolescents: A convenience sample of first-year junior high school students in Finland.
      ]. Gambling frequency, amount of money spent on gambling, and score on other problem gambling screening instruments were positively associated to the DSM-IV-MR-J score [
      • Skokauskas N.
      • Burba B.
      • Freedman D.
      An assessment of the psychometric properties of Lithuanian versions of DSM-IV-MR-J and SOGS-RA.
      ,
      • Tolchard B.
      • Delfabbro P.
      The Victorian Gambling Screen: Validity and reliability in an adolescent population.
      ,
      • Castrén S.
      • Grainger M.
      • Lahti T.
      • et al.
      At-risk and problem gambling among adolescents: A convenience sample of first-year junior high school students in Finland.
      ], in alignment with previous findings [
      • Derevensky J.L.
      • Gupta R.
      Prevalence estimates of adolescent gambling: A comparison of the SOGS-RA, DSM-IV-J, and the GA 20 questions.
      ,
      • Fisher S.
      A prevalence study of gambling and problem gambling in British adolescents.
      ]. Agreement between the SOGS-RA and the DSM-IV-MR-J has previously been found to be greater among males than females [
      • Derevensky J.L.
      • Gupta R.
      Prevalence estimates of adolescent gambling: A comparison of the SOGS-RA, DSM-IV-J, and the GA 20 questions.
      ]. Corresponding items of the DSM-IV-MR-J and the NLCLiP (short screen for problem gambling among children based on the criteria identified in the NODS-CLiP [Diagnostic Screening for Gambling Disorders - Loss of Control, Lying, and Preoccupation]) had deviating endorsement rates [
      • Tolchard B.
      • Delfabbro P.
      The Victorian Gambling Screen: Validity and reliability in an adolescent population.
      ], highlighting the importance of item wording, as misinterpretation may magnify prevalence rates [
      • Ladouceur R.
      • Bouchard C.
      • Rhéaume N.
      • et al.
      Is the SOGS an accurate measure of pathological gambling among children, adolescents and adults?.
      ,
      • Lepper J.
      • Haden B.
      Testing NLCLiP: Validation of estimates of rates of non-problematic and problematic gambling in a sample of British schoolchildren.
      ]. The concept of gambling to escape a negative state of mind may be foreign to and poorly understood by young individuals, leading to false-positive responses [
      • Dodig D.
      Assessment challenges and determinants of adolescents’ adverse psychosocial consequences of gambling.
      ].

      Massachusetts Gambling Screen

      This review provided one estimate of internal consistency (α = .92) [
      • Foster D.W.
      • Hoff R.A.
      • Pilver C.E.
      • et al.
      Correlates of gambling on high-school grounds.
      ] but no indices for classification accuracy of the MAGS DSM-IV subscale and no information on the MAGS subscale. The seven-item subscale of the MAGS (MAGS-7) Cronbach α was .83, and the DSM-IV subscale .87 [
      • Shaffer H.
      • Labrie R.
      • Scanlan K.M.
      • et al.
      Pathological gambling among adolescents: Massachusetts Gambling Screen (MAGS).
      ]. MAGS-7 correctly classified 96% of adolescents as problem gamblers, at-risk gamblers, and nonproblem gamblers when using the DSM-IV criteria as a reference. The MAGS-7 item “Are you always able to stop gambling when you want” was the least discriminating item, and low congruence for the classification of ARPGers was evident between the MAGS-7 and SOGS-RA [
      • Langhinrichsen-Rohling J.
      • Rohling M.L.
      • Rohde P.
      • Seeley J.R.
      The SOGS-RA vs. the MAGS-7: Prevalence estimates and classification congruence.
      ].
      We found new information concerning the psychometric properties of the DSM-IV subscale of the MAGS, suggesting there may be different types of at-risk gamblers: some mostly chase loses, whereas others more notably experience negative consequences. This finding supports the notion of multiple trajectories leading to problem gambling, as suggested by the Pathways Model [
      • Blaszczynski A.
      • Nower L.
      A pathways model of problem and pathological gambling.
      ]. The item on illegal acts was not useful for classifying gamblers [
      • Kong G.
      • Tsai J.
      • Krishnan-Sarin S.
      • et al.
      A latent class analysis of pathological-gambling criteria among high school students: Associations with gambling, risk and health/functioning characteristics.
      ]. Item endorsement differed significantly between genders, suggesting that the validity of the criteria is not equivalent between genders [
      • Molde H.
      • Pallesen S.
      • Bartone P.
      • et al.
      Prevalence and correlates of gambling among 16 to 19-year-old adolescents in Norway.
      ].

      Canadian Adolescent Gambling Inventory

      The CAGI was included to this study even though it is not a population-based instrument. During its development, the CAGI was tested with student samples and a clinical sample, all resulting in the same four-factor structure [

      Alberta Gambling Research Institute. Canadian Adolescent Gambling Inventory (CAGI). Available at: http://www.abgamblinginstitute.ualberta.ca/en/Research/StrategicPartnershipsCollabora/CanadianAdolescentGamblingInve.aspx. Accessed March 4, 2016.

      ]. Cronbach α for the factors ranged from .83 to .90 or within the range often estimated for other instruments. The CAGI displays high congruence to self-rated DSM-IV criteria, clinician-rated DSM-IV criteria, and to the Clinician Rating of Adolescent's Gambling Severity, although the validity of these comparators is unknown [
      • Tremblay J.
      • Stinchfield R.
      • Wiebe J.
      • Wynne H.
      Canadian Adolescent Gambling Inventory (CAGI) phase III final report.
      ].
      The administration time of the CAGI (20 minutes) [
      • Stinchfield R.
      A critical review of adolescent problem gambling assessment instruments.
      ] may inhibit its use, but the nine-item GPSS subscale is shorter and asks about frequencies of concrete gambling behaviors. A unique strength of the GPSS subscale is that items inquire about the effects of gambling on peer relationships and financial consequences in a developmentally appropriate manner [
      • Dodig D.
      Assessment challenges and determinants of adolescents’ adverse psychosocial consequences of gambling.
      ]. However, conclusions on CAGI need to be done cautiously since data are limited to one clinical study which was not reported in a peer-reviewed journal.

      Gambling Addictive Behavior Scale for Adolescents

      The article [
      • Park H.S.
      • Jung S.Y.
      Development of a Gambling Addictive Behavior Scale for adolescents in Korea.
      ] did not provide directions for scoring or classification thresholds, and the instrument lacks strong theoretical framework. The content of proposed factors is questionable, as similar items are distributed across various factors. For example, the factor named loss of control includes the item “Spend money on gambling without paying for necessary things,” whereas the social dysfunction factor includes the item “Bet money or prizes for gambling beyond my pocket money.” Likewise, items “Spend more and more time gambling,” “Do not leave a place all day to gamble,” “Have little time to play other things or do activities except for gambling,” and “It is hard to pass by a PC room, billiards room, or amusement arcade without stopping by” all load onto different factors. However, items do tap into concrete behaviors.

      Theoretical base of the instruments

      A theoretical framework should be a cornerstone for the instruments. In gambling research, surveys largely use adapted adult instruments, where DSM-criteria represent the gold standard. The DSM criteria were originally formulated for diagnostic purposes, not for classifying individuals in a survey [
      • Neal P.
      • Delfabbro P.
      • O'Neil M.
      Problem gambling and harm: Towards a national definition.
      ]. Although many instruments are based on the DSM-IV criteria [
      American Psychiatric Association
      Diagnostic and statistical manual of mental disorders, (DSM-IV).
      ], they lack sufficient empirical evidence from outside the clinical context [
      • Reilly C.
      • Smith N.
      From pathological gambling to gambling disorder: Changes in the DSM-5.
      ]. The use of the DSM-IV criteria as the gold standard has been criticized among adults [
      • Derevensky J.L.
      Measuring and addressing adolescent problem gambling.
      ]; however, the psychiatric criteria for pathologic gambling have never been clinically tested among adolescents [
      • Volberg R.A.
      • Gupta R.
      • Griffiths M.D.
      • et al.
      An international perspective on youth gambling prevalence studies.
      ]. Moreover, research does not recognize the unique developmental characteristics of young people. Developing the CAGI is a step in the right direction.
      Potenza et al. [
      • Potenza M.N.
      • Wareham J.D.
      • Steinberg M.A.
      • et al.
      Correlates of at-risk/problem internet gambling in adolescents.
      ] classified participants endorsing one or more criteria as ARPGers [
      • Potenza M.N.
      • Wareham J.D.
      • Steinberg M.A.
      • et al.
      Correlates of at-risk/problem internet gambling in adolescents.
      ]. Herein, the criteria for ARPG varied. Two or more criteria were used to define ARPGers, with the exception of few studies using one or more criteria [
      • Taylor L.M.
      • Hillyard P.
      Gambling awareness for youth: An analysis of the “Don’t gamble away our future™” program.
      ,
      • Carbonneau R.
      • Vitaro F.
      • Brendgen M.
      • Tremblay R.E.
      Variety of gambling activities from adolescence to age 30 and association with gambling problems: A 15-year longitudinal study of a general population sample.
      ,
      • Delfabbro P.
      • King D.
      • Lambos C.
      • Puglies S.
      Is video-gaming playing a risk factor for pathological gambling in Australian adolescents?.
      ,
      • Foster D.W.
      • Hoff R.A.
      • Pilver C.E.
      • et al.
      Correlates of gambling on high-school grounds.
      ], whereas some studies defined only problem gambling [
      • Barnes G.M.
      • Welte J.W.
      • Hoffman J.H.
      • Tidwell M.-C.O.
      Gambling, alcohol, and other substance use among youth in the United States.
      ,
      • Barnes G.M.
      • Welte J.W.
      • Hoffman J.H.
      • Tidwell M.-C.O.
      The co-occurrence of gambling with substance use and conduct disorder among youth in the United States.
      ,
      • Sheela P.S.
      • Choo W.-Y.
      • Goh L.Y.
      • Tan C.P.L.
      Gambling risk amongst adolescents: Evidence from a school-based survey in the Malaysian setting.
      ,
      • Molde H.
      • Pallesen S.
      • Bartone P.
      • et al.
      Prevalence and correlates of gambling among 16 to 19-year-old adolescents in Norway.
      ] and/or pathologic gambling [
      • Villella C.
      • Martinotti G.
      • Nicola M.D.
      • et al.
      Behavioural addictions in adolescents and young adults: Results from a prevalence study.
      ,
      • Floros G.D.
      • Siomos K.
      • Fisoun V.
      • Geroukalis D.
      Adolescent online gambling: The impact of parental practices and correlates with online activities.
      ]. Notably, there are clear differences of at-risk and problem gambling among adolescents, at-risk referring to individuals who are starting to develop a number of gambling-related problems, but do not meet the established criteria for the more severe form of gambling (i.e., gambling disorder) [
      • Derevensky J.L.
      Measuring and addressing adolescent problem gambling.
      ].
      Current instruments are focused on the negative consequences of gambling, paying less attention to actual behavior [
      • Dodig D.
      Assessment challenges and determinants of adolescents’ adverse psychosocial consequences of gambling.
      ]. Loss of control, however, is a behavior associated with problem gambling [
      • Bergen A.E.
      • Newby-Clark I.R.
      • Brown A.
      Low trait self-control in problem gamblers: Evidence from self-report and behavioral measures.
      ] taken into account to some degree in measurement. For example, pursuing lost money indicates loss of control and instruments included items for “chasing losses” [
      • Dodig D.
      Assessment challenges and determinants of adolescents’ adverse psychosocial consequences of gambling.
      ,
      • Breen R.B.
      • Zuckerman M.
      Chasing in gambling behavior: Personality and cognitive determinants.
      ] — this may even be related to a subtype of at-risk gambling occurring in absence of negative consequences of gambling [
      • Kong G.
      • Tsai J.
      • Krishnan-Sarin S.
      • et al.
      A latent class analysis of pathological-gambling criteria among high school students: Associations with gambling, risk and health/functioning characteristics.
      ]. Another concept that is a fundamental part of addiction, and intuitively linked to loss of control, is craving [
      • Miedl S.F.
      • Buchel C.
      • Peters J.
      Cue-induced craving increases impulsivity via changes in stratal value signals in problem gamblers.
      ,
      • Tavares H.
      • Zilberman M.L.
      • Hodgins D.C.
      • El Guebaly N.
      Comparison of craving between pathological gamblers and alcoholics.
      ]. Craving is not examined in the ARPG instruments.
      The conceptualization of problem gambling incorporates biological, psychological, and social aspects [
      • Blaszczynski A.
      • Nower L.
      A pathways model of problem and pathological gambling.
      ]. It is a major drawback that theory is not better integrated with measurement of ARPG. The individual-centered perspective to addiction research accounts for individual variance in a range of risk factors (i.e., impulsivity [personality trait] and social factors) [
      • Swendsen J.
      • Le Moal M.
      Individual vulnerability to addiction.
      ]. This approach may be helpful to broaden perspectives in gambling research, as the development of gambling problems is influenced by gambling related beliefs, personality traits, and the motivation to gamble [
      • Taylor R.N.
      • Parker J.D.A.
      • Keefer K.V.
      • et al.
      Are gambling related cognitions in adolescence multidimensional? Factor structure of the Gambling Related Cognitions Scale.
      ,
      • Gupta R.
      • Derevensky J.L.
      Adolescent gambling behavior: A prevalence study and examination of the correlates associated with pathological gambling.
      ,
      • Gupta R.
      • Nower L.
      • Derevensky J.L.
      • et al.
      Problem gambling in adolescents: An examination of the pathways model.
      ,
      • Moore S.M.
      • Ohtsuka K.
      Beliefs about control over gambling among young people and their relation to problem gambling.
      ].
      Different aspects of gambling are not equivalent indicators of problem gambling severity, so imputation of item-weighting procedures is necessary. The adult problem gambling instrument Problem and Pathological Gambling Measure [
      • Williams R.J.
      • Volberg R.A.
      Best practices in the population assessment of problem gambling.
      ] enables the detection of gambling problems even with lack of insight or denial of gambling problems, by formulating items accordingly. These aspects should also be noticed in the measurement of youth ARPG.

      Study limitations

      Most articles had poor applicability concerning the index test and reference measurements; only few had developing ARPG instruments as their primary objective. The low quality of the reviewed literature is a result in itself, but also a limitation. Measuring internal consistency exclusively with Cronbach α may be inappropriate because the assumptions are unlikely to be met [
      • Vehkalahti K.
      • Puntanen S.
      • Tarkkonen L.
      Estimation of reliability: A better alternative for Cronbach's alpha.
      ]. We propose that Tarkkonen rho may be a more appropriate estimate of internal consistency as it applies the general form for estimating reliability, whereas Cronbach α is a special case with restrictive assumptions [
      • Vehkalahti K.
      • Puntanen S.
      • Tarkkonen L.
      Estimation of reliability: A better alternative for Cronbach's alpha.
      ]. Regarding bias in the articles, the overall risk was mainly low. Further studies would benefit from even more in-depth quality assessment including criteria for psychometric properties of the instruments [
      American Educational Research Association, American Psychological Association and National Council on Measurement in Education
      Standards for educational and psychological testing.
      ,
      • Cicchetti D.V.
      Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology.
      ].
      Clinical assessment was used as a comparator only rarely. Gamblers with gambling-related problems have shown more discrepancies between self-report and actual outcomes (i.e., accuracy bias, underestimation of losses or overestimation of gains) than gamblers without gambling problems [
      • Bravermann J.
      • Tom M.A.
      • Shaffer H.J.
      Accuracy of self-reported versus actual online gambling wins and losses.
      ]. Attention, memory, and other cognitive systems [
      • Bravermann J.
      • Tom M.A.
      • Shaffer H.J.
      Accuracy of self-reported versus actual online gambling wins and losses.
      ], such as already recognized cognitive biases [
      • Ladouceur R.
      • Sylvain C.
      • Letarte H.
      • et al.
      Cognitive treatment of pathological gamblers.
      ,
      • Ladouceur R.
      • Walker M.
      The cognitive approach to understanding and treating pathological gambling.
      ], might be involved. Thus, in a clinical assessment discussion of possible discrepancies is recommended, as it enables a more detailed evaluation [
      • Bravermann J.
      • Tom M.A.
      • Shaffer H.J.
      Accuracy of self-reported versus actual online gambling wins and losses.
      ]. However, determining the performance of a tool compared to clinical assessment is challenging without a stronger base and defined criteria for ARPG. Without a “gold standard” test, the validation is problematic.
      Nearly all articles were population based, so results cannot be generalized into the clinical context [
      • Abbott M.W.
      • Volberg R.A.
      The measurement of adult problem and pathological gambling.
      ,
      • Salonen A.H.
      • Castrén S.
      • Raisamo S.
      • et al.
      Rahapeliriippuvuuden tunnistamiseen kehitetyt mittarit.
      ]. Short sensitive measures based on DSM criteria are usually used to screen at-risk individuals in the clinical setting as part of a wider clinical assessment [
      • Abbott M.W.
      • Volberg R.A.
      The measurement of adult problem and pathological gambling.
      ]. Screening tools and diagnostic instruments may not be interchangeable as the purposes of the two are fundamentally different [
      • Derevensky J.L.
      • Gupta R.
      The measurement of youth gambling problems.
      ]; in reality, instruments for assessing gambling are often used outside the context originally intended.
      Our material included methodologically diverse studies based on, for example, respondent age, ethnicity, and method of survey administration. To gain a comprehensive understanding of adolescent ARPG, gambling needs to be studied in an ecologically valid manner. This may be achieved with ambulatory methods (i.e., smart phones) [
      • van den Bos R.
      • Davies W.
      • Dellu-Hagedorn F.
      • et al.
      Cross-species approaches to pathological gambling: A review targeting sex differences, adolescent vulnerability and ecological validity of research tools.
      ].

      Summary and Implications

      An estimate of reliability was reported for five ARPG instruments. Most articles (66%) evaluated SOGS-RA. The GABSA was the only novel instrument developed from 2009. Generally, the evaluation of reliability and validity of the instruments was superficial. Despite a very modest publication base, the CAGI seems to have a strong theoretical and methodological base. Reviewed articles with high applicability [
      • Wanner B.
      • Vitaro F.
      • Carbonneau R.
      • Tremblay R.E.
      Cross-lagged links among gambling, substance use, and delinquency from midadolescence to young adulthood: Additive and moderating effects of common risk factors.
      ,
      • Chiesi F.
      • Donati M.A.
      • Galli S.
      • Primi C.
      The suitability of the South Oaks Gambling Screen-Revised for Adolescents (SOGS-RA) as a screening tool: Irt-based evidence.
      ,
      • Colasante E.
      • Gori M.
      • Bastiani L.
      • et al.
      Italian adolescent gambling behavior: Psychometric evaluation of the South Oaks Gambling Screen: Revised for Adolescents (SOGS-RA) among a sample of Italian students.
      ,
      • Skokauskas N.
      • Burba B.
      • Freedman D.
      An assessment of the psychometric properties of Lithuanian versions of DSM-IV-MR-J and SOGS-RA.
      ,
      • Castrén S.
      • Grainger M.
      • Lahti T.
      • et al.
      At-risk and problem gambling among adolescents: A convenience sample of first-year junior high school students in Finland.
      ,
      • Molde H.
      • Pallesen S.
      • Bartone P.
      • et al.
      Prevalence and correlates of gambling among 16 to 19-year-old adolescents in Norway.
      ,
      • Kong G.
      • Tsai J.
      • Krishnan-Sarin S.
      • et al.
      A latent class analysis of pathological-gambling criteria among high school students: Associations with gambling, risk and health/functioning characteristics.
      ,
      • Tremblay J.
      • Stinchfield R.
      • Wiebe J.
      • Wynne H.
      Canadian Adolescent Gambling Inventory (CAGI) phase III final report.
      ,
      • Park H.S.
      • Jung S.Y.
      Development of a Gambling Addictive Behavior Scale for adolescents in Korea.
      ] advocate that screening should include measures of risk taking, self-control or impulsivity, delinquent behavior, and social risk factors. The GABSA and the CAGI were the only instruments originally developed especially for youth. Studies were entirely population based, except the one concerning the CAGI. In 2010, Volberg et al. concluded that despite the questions raised regarding the validity of the SOGS-RA and DSM-IV-MR-J, these instruments are the best tools for evaluating adolescent gambling problems while waiting for a better-validated instrument [
      • Volberg R.A.
      • Gupta R.
      • Griffiths M.D.
      • et al.
      An international perspective on youth gambling prevalence studies.
      ]. In 2015, this conclusion still seems up to date.
      In the past 5 years, a variety of ethnicities were included in research [
      • Blinn-Pike A.
      • Worthy S.L.
      • Jonkman J.N.
      Adolescent gambling: A review of an emerging field of research.
      ], although cultural validation has not been an actual aim. It is encouraged that future studies look closer into gender differences, study clinical samples, and community-based samples. Attending to these issues in ARPG research may lead to a deeper understanding of the phenomena. In recognition of the similarities across addictions, it is important that accumulating knowledge from the field is integrated into measuring ARPG.
      Rigorous psychometric research of youth ARPG instruments, as recommended by earlier reviews [
      • Blinn-Pike A.
      • Worthy S.L.
      • Jonkman J.N.
      Adolescent gambling: A review of an emerging field of research.
      ,
      • Stinchfield R.
      A critical review of adolescent problem gambling assessment instruments.
      ], has not yet been accomplished. Thus, it is untimely to name the most suitable instrument presently available. Researchers are encouraged to test reliability in population-based studies, where samples are not derived from schools, and especially in the clinical context. Reporting alphas from previous articles is not enough, considering the weaknesses in the theoretical foundation of ARPG instruments. We hope that bringing recent work to light will nourish forthcoming studies. If researchers collectively strive to improve the accuracy of ARPG measurement, even the understanding of behavioral addictions in general may advance tremendously.

      Acknowledgments

      The authors thank the Information Service Designer Pirjo Vuorio, National Institute for Health and Welfare, for her substantial assistance in the management of articles used during the conduct of this systematic review. All authors contributed significantly to the planning of the work and drafting the manuscript or revising the work critically for important intellectual content. The authors of this manuscript and Pirjo Vuorio did not have any interests that might be interpreted as influencing the research. APA ethical standards were followed in the conduct of the study. The corresponding author affirms that all individuals who have contributed significantly to the preparation of this manuscript have been mentioned in the Acknowledgements. Please note that the coauthor Sari Castrén has presented the protocol of our report at the sixth International Gambling Conference in February 2016 (10.-12.2.2016), in Auckland, New Zealand.

      Funding Sources

      This study was financially supported by the Ministry of Social Affairs and Health, Helsinki, Finland (the §52 Appropriation of the Lotteries Act, contract STM/3189/2011).

      Supplementary Data

      References

        • Blinn-Pike A.
        • Worthy S.L.
        • Jonkman J.N.
        Adolescent gambling: A review of an emerging field of research.
        J Adolesc Health. 2010; 47: 223-236
        • Volberg R.A.
        • Gupta R.
        • Griffiths M.D.
        • et al.
        An international perspective on youth gambling prevalence studies.
        Int J Adolesc Med Health. 2010; 22: 3-38
        • Castrén S.
        • Basnet S.
        • Pankakoski M.
        • et al.
        An analysis of problem gambling among the Finnish working-age population: A population survey.
        BMC Public Health. 2013; 13: 519
        • Johansson A.
        • Grant J.E.
        • Kim S.W.
        • et al.
        Risk factors for problematic gambling: A critical literature review.
        J Gambl Stud. 2009; 25: 67-92
        • Derevensky J.L.
        • Gupta R.
        Prevalence estimates of adolescent gambling: A comparison of the SOGS-RA, DSM-IV-J, and the GA 20 questions.
        J Gambl Stud. 2000; 16: 227-251
        • Potenza M.N.
        • Wareham J.D.
        • Steinberg M.A.
        • et al.
        Correlates of at-risk/problem internet gambling in adolescents.
        J Am Acad Child Adolesc Psychiatry. 2011; 50: 150-159.e3
        • Edgren R.
        • Castrén S.
        • Jokela M.
        • Salonen A.H.
        At-risk and problem gambling among Finnish youth: The examination of risky alcohol consumption, tobacco smoking, mental health and loneliness as gender-specific correlates.
        Nordic Stud Alcohol Drugs. 2016; 11: 1-19
        • Winters K.C.
        • Stinchfield R.
        Youth gambling: Prevalence, risk and protective factors and clinical issues.
        in: Increasing The Odds: What Clinicians Need to Know About Gambling Disorders. Volume 7. National Center for Responsible Gaming (NCRG), Beverly, MA2012: 14-25 (Available at:) (Accessed October 1, 2015)
        • Stinchfield R.
        A critical review of adolescent problem gambling assessment instruments.
        Int J Adolesc Med Health. 2010; 22: 77-93
        • Winters K.C.
        • Stinchfield R.
        • Fulkerson J.
        Towards the development of an adolescent gambling problem severity scale.
        J Gambl Stud. 1993; 9: 63-84
        • Wiebe J.M.
        • Cox B.J.
        • Mehmel B.G.
        The South Oaks Gambling Screen Revised for Adolescents (SOGS-RA): Further psychometric findings from a community sample.
        J Gambl Stud. 2000; 16: 275-288
        • Langhinrichsen-Rohling J.
        • Rohling M.L.
        • Rohde P.
        • Seeley J.R.
        The SOGS-RA vs. the MAGS-7: Prevalence estimates and classification congruence.
        J Gambl Stud. 2004; 20: 259-281
        • Welte J.W.
        • Barnes G.M.
        • Tidwell M.C.O.
        • Hoffman J.H.
        The prevalence of problem gambling among US adolescents and young adults: Results from a national survey.
        J Gambl Stud. 2008; 24: 119-133
        • Ladouceur R.
        • Bouchard C.
        • Rhéaume N.
        • et al.
        Is the SOGS an accurate measure of pathological gambling among children, adolescents and adults?.
        J Gambl Stud. 2000; 16: 1-24
        • Fisher S.
        Measuring pathological gambling in children: The case of fruit machines in the UK.
        J Gambl Stud. 1992; 8: 263-285
        • Fisher S.
        Developing the DSM-IV-DSM-IV criteria to identify adolescent problem gambling in non-clinical populations.
        J Gambl Stud. 2000; 16: 253-273
        • Shaffer H.
        • Labrie R.
        • Scanlan K.M.
        • et al.
        Pathological gambling among adolescents: Massachusetts Gambling Screen (MAGS).
        J Gamb Stud. 1994; 10: 339-362
        • Wiebe J.
        • Wynne H.
        • Stinchfield R.
        • Tremblay J.
        Measuring problem gambling in adolescent populations: Phase I report.
        Canadian Centre on Substance Abuse, Ottawa, ON2005 (Available at:) (Accessed October 9, 2015)
        • Wiebe J.
        • Wynne H.
        • Stinchfield R.
        • Tremblay J.
        The Canadian Adolescent Gambling Inventory (CAGI): Phase II final report.
        Canadian Centre on Substance Abuse, Ottawa, ON2007 (Available at:) (Accessed October 9, 2015)
        • Hayer T.
        • Meyer G.
        • Petermann F.
        Gambling-related problems among youths: A critical review of current screening instruments.
        Kindheit und Entwicklung. 2014; 23: 174-183
      1. Nuorisolaki 72/2006. [Youth Law]. Finlex Data Bank. Helsinki, Finland: Finland's Ministry of Justice, 2006. Available at: http://www.finlex.fi/fi/laki/alkup/2006/20060072. Accessed September 28, 2015.

        • Liberati A.
        • Altman D.G.
        • Tetzlaff J.
        • et al.
        The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: Explanation and elaboration.
        Ann Intern Med. 2009; 151: W65-W94
        • Moher D.
        • Liberati A.
        • Tetzlaff J.
        • Altman D.G.
        Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement.
        Ann Intern Med. 2009; 151: 264-269
        • Derevensky J.
        • Gupta R.
        The measurement of youth gambling problems: Current instruments, methodological issues and future directions.
        in: Derevensky J. Gupta R. Gambling Problems in Youth: Theoretical and Applied Perspectives. Kluwer Academic/Plenum Publishers, New York2004: 121-144
        • Derevensky J.
        • Gupta R.
        Measuring gambling problems amongst adolescents: Current status and future directions.
        Int Gambl Stud. 2006; 6: 201-215
      2. O'Connor D, Green S, Higgins JPT. Chapter 5: Defining the review question and developing criteria for including studies. In: Higgins JPT, Green S, eds. Cochrane handbook for systematic reviews of interventions version 5.0.0 [updated February 2008]. The Cochrane Collaboration, 2008. Available at: http://handbook.cochrane.org/chapter_5/5_defining_the_review_question_and_developing_criteria_for.htm. Accessed March 31, 2016.

        • Whiting P.F.
        • Rutjes A.W.
        • Westwood M.E.
        • et al.
        QUADAS-2: A revised tool for the quality assessment of diagnostic accuracy studies.
        Ann Intern Med. 2011; 155: 529-536
        • Barnes G.M.
        • Welte J.W.
        • Hoffman J.H.
        • Tidwell M.-C.O.
        Gambling, alcohol, and other substance use among youth in the United States.
        J Stud Alcohol Drugs. 2009; 70: 134-142
        • Mooss A.V.
        Gambling behaviours among youth involved in juvenile and family courts.
        (Dissertation) Georgia State University, Atlanta, GA2009 (Available at:) (Accessed December 11, 2015)
        • Taylor L.M.
        • Hillyard P.
        Gambling awareness for youth: An analysis of the “Don’t gamble away our future™” program.
        Int J Ment Health Addict. 2009; 7: 250-261
        • Wanner B.
        • Vitaro F.
        • Carbonneau R.
        • Tremblay R.E.
        Cross-lagged links among gambling, substance use, and delinquency from midadolescence to young adulthood: Additive and moderating effects of common risk factors.
        Psychol Addict Behav. 2009; 23: 91-104
        • Welte J.W.
        • Barnes G.M.
        • Tidwell M.-C.O.
        • Hoffman J.H.
        Association between problem gambling and conduct disorder in a national survey of adolescent and young adults in the Unites States.
        J Adolesc Health. 2009; 45: 396-401
        • Barnes G.M.
        • Welte J.W.
        • Hoffman J.H.
        • Tidwell M.-C.O.
        Comparisons of gambling and alcohol use among college students and noncollege young people in the United States.
        J Am Coll Health. 2010; 58: 443-452
        • Hansen M.
        • Rossow I.
        Limited cash flow on slot machines: Effects of prohibition of note acceptors on adolescent gambling behavior.
        Int J Ment Health Addict. 2010; 8: 70-81
        • Barnes G.M.
        • Welte J.W.
        • Hoffman J.H.
        • Tidwell M.-C.O.
        The co-occurrence of gambling with substance use and conduct disorder among youth in the United States.
        Am J Addict. 2011; 20: 166-173
        • Dussault F.
        • Brendgen M.
        • Vitaro F.
        • et al.
        Longitudinal links between impulsivity, gambling problems and depressive symptoms: A transactional model from adolescence to early adulthood.
        J Child Psychol Psychiatry. 2011; 52: 130-138
        • Villella C.
        • Martinotti G.
        • Nicola M.D.
        • et al.
        Behavioural addictions in adolescents and young adults: Results from a prevalence study.
        J Gambl Stud. 2011; 27: 203-214
        • West R.L.
        The effect of extended family gambling behavior and family functioning on African American adolescent gambling.
        (Dissertation) The University of Memphis, Memphis, TN2011 (Available at:) (Accessed December 11, 2015)
        • Ashrafioun L.
        • McCarthy A.
        • Rosenberg H.
        Assessing the impact of cue exposure on craving to gamble in university students.
        J Gambl Stud. 2012; 28: 363-375
        • Chiu E.Y.-W.
        • Woo K.
        Problem gambling in Chinese American adolescents: Characteristics and risk factors.
        Int J Ment Health Addict. 2012; 10: 911-922
        • Liu W.
        • Lee G.P.
        • Goldweber A.
        • et al.
        Impulsivity trajectories and gambling in adolescence among urban male youth.
        Addiction. 2012; 108: 780-788
        • Walther B.
        • Morgenstern M.
        • Hanewinkel R.
        Co-occurrence of addictive behaviours: Personality factors related to substance use, gambling and computer gaming.
        Eur Addict Res. 2012; 18: 167-174
        • Wong S.S.K.
        • Tsang S.K.M.
        Validation of the Chinese version of the Gamblers’ Belief Questionnaire (GBQ-C).
        J Gambl Stud. 2012; 28: 561-572
        • Chiesi F.
        • Donati M.A.
        • Galli S.
        • Primi C.
        The suitability of the South Oaks Gambling Screen-Revised for Adolescents (SOGS-RA) as a screening tool: Irt-based evidence.
        Psychol Addict Behav. 2013; 27: 287-293
        • Colasante E.
        • Gori M.
        • Bastiani L.
        • et al.
        Italian adolescent gambling behavior: Psychometric evaluation of the South Oaks Gambling Screen: Revised for Adolescents (SOGS-RA) among a sample of Italian students.
        J Gambl Stud. 2014; 30: 789-801
        • Donati M.A.
        • Chiesi F.
        • Primi C.
        A model to explain at-risk/problem gambling among male and female adolescents: Gender similarities and differences.
        J Adolesc. 2013; 36: 129-137
        • Poulin C.
        Problem gambling among adolescent students in the Atlantic provinces of Canada.
        J Gambl Stud. 2000; 16: 53-78
        • Lee G.P.
        • Stuart E.A.
        • Ialongo N.S.
        • Martins S.S.
        Parental monitoring trajectories and gambling among a longitudinal cohort of urban youth.
        Addiction. 2013; 109: 977-985
        • Parker J.D.A.
        • Summerfeldt L.J.
        • Kloosterman P.H.
        Gambling behavior in adolescents with learning disorders.
        J Gambl Stud. 2013; 29: 231-239
        • Primi C.
        • Donati M.A.
        • Bellini I.
        • et al.
        Measuring the attitude towards the profitability of gambling: The psychometric properties of the Gambling Attitude Scale.
        Bollettino di Psicologia Applicata. 2013; 266: 49-58
        • Tozzi L.
        • Akre C.
        • Fleury-Schubert A.
        • Suris J.-C.
        Gambling among youths in Switzerland and its association with other addictive behaviours.
        Swiss Med Wkly. 2013; 143: w13768
        • Wong G.
        • Zane N.
        • Saw A.
        • Chan A.K.K.
        Examining gender differences for gambling engagement and gambling problems among emerging adults.
        J Gambl Stud. 2013; 29: 171-189
      3. The mental health and well-being of Ontario students 1991-2013: Detailed OSDUHS findings (CAMH Research Document Series No. 38). Center for Addiction and Mental Health, Toronto, ON2014
        • Bray B.C.
        • Lee G.P.
        • Liu W.
        • et al.
        Transitions in gambling participation during late adolescence and young adulthood.
        J Adolesc Health. 2014; 55: 188-194
        • Cook S.
        • Turner N.E.
        • Ballon B.
        • et al.
        Problem gambling among Ontario students: Associations with substance abuse, mental health problems, suicide attempts, and delinquent behaviours.
        J Gambl Stud. 2015; 31: 1121-1134
        • Donati M.A.
        • Primi C.
        • Chiesi F.
        Prevention of problematic gambling behavior among adolescents: Testing the efficacy of an integrative intervention.
        J Gambl Stud. 2014; 30: 803-818
        • Canale N.
        • Vieno A.
        • Griffiths M.D.
        • et al.
        Trait urgency and gambling in young people by age: The mediating role of decision making processes.
        Addict Behav. 2015; 46: 39-44
        • Carbonneau R.
        • Vitaro F.
        • Brendgen M.
        • Tremblay R.E.
        Variety of gambling activities from adolescence to age 30 and association with gambling problems: A 15-year longitudinal study of a general population sample.
        Addiction. 2015; 110: 1985-1993
        • Chan A.K.K.
        • Zane N.
        • Wong G.M.
        • Song A.V.
        Personal gambling expectancies among Asian American and White American college students.
        J Gambl Stud. 2015; 31: 33-57
        • Revelle W.
        • Zinbarg R.E.
        Coefficients alpha, beta, omega, and the glb: Comments on Sijtsma.
        Psychometrika. 2009; 74: 145-154
        • Míguez M.C.
        • Becona E.
        Do cigarette smoking and alcohol consumption associate with cannabis use and problem gambling among Spanish adolescents?.
        Addicciones. 2015; 27: 8-16
        • Sheela P.S.
        • Choo W.-Y.
        • Goh L.Y.
        • Tan C.P.L.
        Gambling risk amongst adolescents: Evidence from a school-based survey in the Malaysian setting.
        J Gambl Stud. 2015; ([E-pub ahead of print].)
        • Delfabbro P.
        • King D.
        • Lambos C.
        • Puglies S.
        Is video-gaming playing a risk factor for pathological gambling in Australian adolescents?.
        J Gambl Stud. 2009; 25: 391-405
        • Skokauskas N.
        • Burba B.
        • Freedman D.
        An assessment of the psychometric properties of Lithuanian versions of DSM-IV-MR-J and SOGS-RA.
        J Gambl Stud. 2009; 25: 263-271
        • Dowling N.
        • Jackson A.C.
        • Thomas S.A.
        • Frydenberg E.
        Children at risk of developing problem gambling: Final report. A report to gambling research Australia in fulfilment of Tender 103/06.
        The Problem Gambling Research and Treatment Centre (PGRTC), Victoria2010
        • Brunelle N.
        • Leclerc D.
        • Cousineau M.-M.
        • et al.
        Internet gambling, substance use, and delinquent behavior: An adolescent deviant behavior involvement pat