You are right, pretty much. If you want to conclude prices have either headed up, down or stayed flat, you need the best evidence in each case. This appraiser is biased in terms of no price change. He calls this "stability" or "stable" perhaps. But, in fact, we know that the markets are not stable when there is an imbalance between supply and demand exacerbated by fears of a worsening pandemic that makes buyers act hastily rather than take their time searching the market - even if the the market provided more selection.
What you are seeing here is the impact of statistical courses based on Gaussian or parametric statistics taught by the Appraisal Institute and other appraisal education providers. This form of statistics isn't even suitable for Real Estate. They should be teaching non-parametric statistics. Parametric statistics makes the assumption that the data you are using is a sample from a population with known parameters such as the Normal Distribution. This is almost never the case in typical complex markets that are the backbone of most high density markets composed of mixtures of very old, old, medium aged and newer homes of all styles and design, including condos, townhouses and subdivisions, with and without HOA fees. Our data does not fit parametric distributions. When we lack sales we have to data mine for sales data wherever we can find it, spanning property types, dates and locations. We have to deal with chunky data. And we get so little, we deal with the entire population, not a sample.
So, we don't need to talk about "small" data sets providing "inconclusive" results. Small data sets are all we have and we need to extract every bit of information we can from them and use it to provide the "best" conclusion of value regardless of its reliability. The client wants the best indication of value. "Best" certainly does not mean perfect. It just mean best. So, we engage in data mining techniques using tools like MARS.
The statistics courses for appraisal need to be fundamentally changed.
The Appraisal Institute, TAF and others have been guided for years by Real Estate Statisticians's who have Ph.D.'s from third rate universities and simply don't know what they are talking about. They are writing books for a virtual audience.
Caveat: There are a few simple markets around, where things are sufficiently simple enough to use simple averages and means and multi-linear regression. But don't try that in the SF Bay Area.
And in this day and age, ..., well I was just talking to a realtor in Mill Valley who said that buyers who are able to buy are coming in with enough cash to pay 30% - 40% down on $1M - $2.5M homes. They get snapped up fast; - and he said, if buyers have that much cash, how important are interest rates, really? And, if the appraiser wants to be able to deal with these issues in many areas, he needs to be adept at dealing with non-parametric statistics just to stay afloat in a sea of erratic complex decision making often by highly intelligent individuals who need certain functionality specific to their situation in terms of family (children?), relatives and guests, business considerations, first and foremost, followed by quality, condition and other less important considerations.
Another Caveat: Many will decry that appraisers are simply neither smart enough nor equipped with the tools to handle non-parametric statistics. Deal with it.