Operationalizing Halo: Problems with the Computation of a Standard Deviation Across Dimensions within Ratees.
Abstract
The use of a standard deviation across rating dimensions for each rate as a measure of halo is criticized for those cases in which the average of the ratings on each dimension is not equal. Standardizing the data within each dimension prior to computing the standard deviation across dimensions corrects this problem. Further, an example is presented in which this standardized standard deviation is shown to be correlated nearly 1.00 with the average intercorrelation among rating dimensions and the average difference between the true and observed dimension intercorrelations. Correlations between the commonly used unstandardized standard deviation and the other operationalizations of halo were approximately .80.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1984
- Accession Number
- ADA138393
Entities
People
- E. D. Pulakos
- N. Schmitt
Organizations
- Michigan State University