An experimental study using Rasch analysis to compare absolute magnitude estimation and categorical rating scaling as applied in survey research.

Limited research has applied a measurement model to compare the rating scale functioning of categorical rating scaling (CRS) and absolute magnitude estimation scaling (MES) when rating subjective stimuli. We used an experimental design and applied the Rasch model to the survey data, with each respondent rating items using MES and one of four commonly used agreement-disagreement rating scales. The results indicated that the CRS and MES data were comparable in person and item separation and reliability when the respondents' scales were known. MES had lower standard error for people and items; however MES had disordered step calibrations. Finally, the respondents reported preference of CRS to MES.