Logarithmic Transformation of Magnitude Estimation Data and Comparisons of Scaling Methods

Recent studies comparing magnitude estimation to other scaling methods have been criticized for failure to submit the data to logarithmic transformation before statistical analysis, a procedure which can improve the ability of magnitude estimation scaling to differentiate among products. Data from one of these studies was available for re‐analysis, which was conducted both with and without logarithmic transformation of magnitude estimation, category scales and line scales. The ability of magnitude estimation to differentiate among products was improved by log transformation, while the other methods were not. Improvements were associated in pan with reductions in positive skew and improved approximation of the normal distribution. In spite of this improvement, magnitude estimation remained slightly inferior to the other methods especially in the hands of an untrained consumer sample.