| Session Title: Using Latent Class Analysis to Target and Tailor Programs to Specific Populations |
| Demonstration Session 546 to be held in Lone Star E on Friday, Nov 12, 10:55 AM to 11:40 AM |
| Sponsored by the Quantitative Methods: Theory and Design TIG |
| Presenter(s): |
| Humphrey Costello, University of Wyoming, humphrey.costello@uwyo.edu |
| Reese Jenniges, University of Wyoming, jenniges@uwyo.edu |
| Abstract: In this demonstration, we present an application of Latent Class Analysis (LCA) to adult smokers in Wyoming. LCA is a probabilistic clustering method for identifying unmeasured class membership using both categorical and continuous variables (see Vermunt & Magidson, 2000). We apply LCA to derive a four-class typology of smoking behavior and intention to quit. We then use logistic regression to identify associations between smoking types and demographic variables. Programming and policies may then be tailored to target the needs of each type of smoker. The demonstration introduces the LCA method, discusses appropriate uses, details steps and diagnostics in developing LCA models, and describes how LCA may enhance both program design and program evaluation. |