| Session Title: Improving and Applying Measurement Techniques to Identify and Account for Differences Across Social Groups |
| Multipaper Session 926 to be held in Mineral Hall Section B on Saturday, Nov 8, 4:00 PM to 5:30 PM |
| Sponsored by the Quantitative Methods: Theory and Design TIG |
| Chair(s): |
| Mary Kay Falconer, Ounce of Prevention Fund of Florida, mfalconer@ounce.org |
| Abstract: Recognizing the social diversity of populations served by many programs, evaluators should address how measurement can be improved and used to identify and account for differences across social groups. Adopting measurement techniques that account for differences across ethnicities or other social characteristics of program participants provides greater assurances that the measures are valid for the entire participant group. This session will cover at least two relevant approaches for improving and applying measurement techniques. In addition, information on recent efforts by product developers (testing resources) to improve several widely used measurement tools across different social groups will be shared. As a third component, presenters will be asked to list lessons learned and recommendations related to measurement when evaluating programs serving diverse populations. |
| Meeting the Challenge of Social Diversity in Measurement Tools |
| Mary Kay Falconer, Ounce of Prevention Fund of Florida, mfalconer@ounce.org |
| With ethnic diversity being important in target and participant populations for many programs that are being evaluated, the validity of widely used measurement tools must be addressed. Based on a survey of product and testing services, recent efforts to improve the validity of selected tools across multiple ethnic groups will be identified. The coverage of ethnic and age groups with these tools by product developers will also be indicated. The list of tools included in this survey or compilation of information will be reviewed and approved by the other presenters in the session. The objective in this presentation will be to provide one account of the 'status or progress" of measurement in meeting the challenges of evaluations of programs serving diverse ethnic populations. |
| Testing a Model of the "Mistreatment and Barriers to Help-Seeking by Elder Women Abused by an Intimate Other" |
| Frederick Newman, Florida International University, newmanf@fiu.edu |
| Richard Beaulauria, Florida International University, |
| Laura R Seff, Florida International University, emis2go@cs.com |
| An instrument was developed based on a qualitative analysis of 21 focus groups of women (N=134, ages 50 to 85) that represented Hispanics, White Non-Hispanics, and Black (Caribbean Haitian or Jamaicans). Using criteria of two or more persons in two or more groups having similar coded concepts to identify a "qualitatively supported factor," using Atlas.ti, we were able to identify three clusters of factors: Internal Barriers (e.g., self blame, helplessness/powerlessness, secrecy, protecting family), External Barriers (Family response, Clergy Response, Justice System Response, Community Response), and Abuser Behaviors (Isolation, Intimidation, Jealousy). We are also seeking a form of convergent validation with a standardized measure, the Conflict Tactics Scale (CTS), and comparing these three race-ethnicities on the CTS for different age groups. In addition to testing the model on 450 women (150 in each race-ethnicity) we intend to identify similarities and differences among the major ethnic and age groupings. |
| Using the Rasch Item Partition Model to Improve Theory-Based Evaluation |
| John Gargani, Gargani and Company Inc, john@gcoinc.com |
| I describe the Rasch item partition (RIP) model, a hierarchical generalized linear model that can be used to simultaneously estimate program impacts and gather evidence about the underlying mechanisms presumed to cause impacts. I explain how evaluators can use the RIP model to integrate theory-based evaluations with randomized trials, providing examples from randomized trials in which RIP models shed light on presumed mechanisms of change. |