atc: Think of it this way: A given house that cost $100,000 with lot might sell for $100,000, or it might sell for $120,000, or it might sell for $150,000, or it could even sell for $200,000. Believe it or not, this happens sometimes for various reasons. These sales are called outliners. The question is: Which price in this range of sale prices is most probable to take place. If we take about ten similar sales and the range is $90,000 to $110,000, then the data is telling us that a bell curve is forming with a peak at around $100,000. If we could post graphs I could show you some very interesting things. If you pick 30 similar sales and adjust them all using least sum of squares and graph the results, most of the time you will find outliners well above and well below the trend line. That is why I don't use only three random comparable sales. You could pick three outliners and really be high or low. In upper income property you can really screw up. When I do an appraisal, I pick the 20 best comparables from my data base, do a graph with GLA vs sale price, and cull out the outliners. Then do the same after the sales have been equalized.

This is why the definition of market value specifically says, "most probable price" because that is the only way to describe a random range of prices.

PS: If an appraiser uses a statistically significant number of the most comparable sales, by definition, he/she is never wrong. If the answer is what the data indictes and that is the answer to the question, then how can the appraiser be wrong. Unless they got some wrong data.