Unit selection and the sales comparison approach

Reports the results of an empirical investigation conducted for the purposes of exploring the issue of unit selection and the sales comparison approach. The proximate motivation of this study was in determining how non‐reinforcing appraisal estimates may be addressed. The investigation proceeds by exploring two possible criteria through which the reliability of appraisal estimates may be measured. The first involves the percentage error made in price per unit of comparison (UOC), while the second concerns the total valuation error in the appraisal of real property. Results involve the utilization of the coefficient of variation and the Markov inequality, and may assist appraisers when different units of comparison yield non‐reinforcing estimates of value. It is shown that maximum confidence in guaranteeing that the percentage error between the estimated and actual price per UOC lies within a tolerance level chosen ex‐ante obtains through choosing the UOC with the minimum coefficient of variation. Total valuation error is minimized as a function of the standard deviation for the price per UOC, the sample size, and the UOC’s value for the real property being appraised. While minimum per unit percentage error may be obtained utilizing a particular UOC, the minimization of total valuation error may imply the utilization of an alternative unit. It is shown for the empirical analysis conducted, that when two common units of comparison are considered ‐ acreage and front footage ‐ both percentage and total valuation error were minimized through the use of price per acre rather than price per front foot.