Validation of the global appraisal of individual needs’ general individual severity scale and its subscales against the Rasch measurement model

Abstract

s / Drug and Alcohol Dependence 140 (2014) e2–e85 e37 examined. The study was conducted using a laboratory model of risk taking to examine these differences in a controlled setting. Methods: The sample included 233 adolescents (42% Black) assessed longitudinally across 4 years (between 9 and 16 years old). To evaluate risk taking, we utilized the well-validated BARTY laboratory risk task (Lejuez et al., 2007). Parental monitoring was measured using an abbreviated version of the Stattin and Kerr (2000) questionnaire. Exposure to threats was assessed through a 21-item scale consisting of self-reported neighborhood, home, and school threats ( ’s across 4 years = .86–.89). Results: Hierarchical Linear Modeling analyses indicated that relative to White, Black adolescents engaged in lower risk taking over time (ˇ =−6.69; SE=1.62; p 1.33 for both Infit MNSQ and Outfit MNSQ. The criterion-related validity of all scales was consistent with hypotheses. Based on differential test functioning analysis, the full GISS scores had different meanings depending on age and gender of the respondents. Conclusions: This study supported the use of the four GISS subscales for their target constructs. As suggested by the DSF results, we urge caution in relying only on the GISS total score since variability in how one arrived at a total varied by age and gender. Financial support: NIDA grant no. R37 DA011323 and SAMHSA contract no. 270-12-0397. http://dx.doi.org/10.1016/j.drugalcdep.2014.02.122 Validation of the gain self-help involvement scale using the Rasch measurement model Karen M. Conrad1,2, K.J. Conrad3, L. Passetti 3, R.R. Funk3, Michael L. Dennis3 1 Program Metrics, Oak Park, IL, United States 2 University of Illinois at Chicago, Chicago, IL, United States 3 Chestnut Health Systems, Normal, IL, United States Aims: Self-help groups are increasingly recognized as an integral part of managing long term recovery from substance use disorders. The aims of this poster are to evaluate the psychometric properties of the original 21-item Self-Help Involvement Scale (SHIS) and of a short-form version based on traditionalmeasures of internal consistency (e.g., Cronbach’s alpha) and the more conservative Rasch measurement model (aka 1 parameter item response theory [IRT]). Methods: Data were collected using the Global Appraisal of Individual Needs during 2009–2010 from 801 clients in substance abuse treatment who completed the SHIS at 90-days post intake. The sample was predominantly over 18 years of age (71.4%) and male (54%). Over half were Caucasian (54%). Alcohol was reported as a primary drug by slightly over 20%,Marijuana andOpioidswere each reportedby18%, Cocainewas reportedby11%, Amphetamines was the least reported primary drug at about 8%. About a quarter were grouped into the other drug category rather than one of five primarydrugs listed.Winsteps softwarewasused toplacebothpersons and items on a common, linear, interval “ruler” that enabled examination of: person and item reliability; itemhierarchy; itemfit statistics; unidimensionality; and item invariance across primary drug of choice. Results: Both the original and short-form SHIS were internally consistent and met the requirements of the Rasch measurement model except for two items thatmisfit possibly due tomethod variance given that these two items had a different rating scale than the remaining items. Item invariance across drugs was observed for 10 items of the full 21-item version. This finding guided selection of items for the short-form version because of our interest in using the measure across primary drug types. Conclusions: Having available psychometrically sound, full and short-formmeasures of self-help involvement for use across drugs and treatment types strengthens our ability to discern treatment effects. Financial support: SAMHSA/CSAT Contract No. 270-12-0397. http://dx.doi.org/10.1016/j.drugalcdep.2014.02.123