The Use of Rank Transformation and Multiple Regression Analysis in Estimating Residential Property Values With A Small Sample

Conventional multiple regression analysis which has been used in estimating residential property values typically relies upon cardinal data. This paper argues that appraisal theory requires the appraiser to rank the comparables from best to worst and use a regression technique which can be applied to ordinal data. The rank regression procedure illustrated here was successfully used on small sample sizes, and did not violate the critical assumptions underlying conventional multiple regression. The results indicate that the rank regression technique illustrated here is more theoretically correct than conventional multiple regression and produces a better model with more accurate price estimates.