| Session Title: Using Rasch Measurement to Strengthen Evaluation Designs and Outcomes |
| Demonstration Session 786 to be held in Lone Star E on Saturday, Nov 13, 10:55 AM to 12:25 PM |
| Sponsored by the Quantitative Methods: Theory and Design TIG |
| Presenter(s): |
| Christine Fox, University of Toledo, chris.fox@utoledo.edu |
| Svetlana Beltyukova, University of Toledo, svetlana.beltyukova@utoledo.edu |
| Abstract: While most evaluation plans use raw ordinal survey data to assess attitudes, beliefs, or behaviors, the Rasch model can be used to determine the extent to which these data yield meaningful linear measurements from these survey responses. This model also allows for obtaining evidence of the functioning the rating scales, suitability of the instrument for specific populations, and facilitates a better understanding of how precisely different samples can be evaluated. Using a survey of high school teachers as an example of one evaluative instrument in a comprehensive evaluation for the Department of Education, we will demonstrate how a variety of Rasch diagnostics can aid in using typical rating scales to construct scientifically defensible measures. Specific focus will be on determining the appropriate number of rating scale categories, designing questions to target the appropriate sample, exploration of dimensionality, and construct interpretation. By the end of the demonstration, participants will be able to identify properties of measures, understand the limitations of working with ordinal data, and be familiar with the way in which the Rasch model overcomes these limitations to provide empirically defensible ways to build and test the quality of measures from survey data. |