Comparable? That is the question. In a rural area that can get a little squishy. For example when I put in the search critieria I try to put in parameters that I would do a search on. And lenders don't want to use sale price criteria. And I understand that, but then I get these ridiculous ranges in values that have properties that aren't the least comparable due to differences in quality or other features like view. When I go through and eliminate those not comparable then it ends up with a definite price range. So typically when I know a house is going to be in that 200,000 range, I prefer to search by specific criteria such as GLA, acreage, age, and then by price range. But that's not what I put in the 1004MC, because they don't want you to search by a price range. But if I don't then I really don't get what I consider comparables. The 1004MC is all smoke and mirrors. Time to dump it.
I'm going to add on a bit to what JGrant said (although she may not agree with what I have below).
I don't see the same challenge as you do. In fact, Sale price is not what defines a property's similarity, comparability, or its market-acceptance as a substitute for the subject. But what price does reflect (in nearly all cases) is the market's perception of the sum of the benefits of a particular property.
Properties that are superior to my subject are expected to sell for more than what my subject would sell for (let's take the component of moving markets out of the equation, because that can be identified, analyzed, and adjusted for).
Properties that are inferior to my subject are expected to sell for less than what my subject would sell for.
Properties that are similar to my subject are expected to sell in a range that likely reflects what my subject will sell for.
The price a property sold for reflects its perceived benefits. What you describe is when you put in a search parameter that doesn't include price, you are getting a wide range of prices because, despite your best efforts in refining your search, it isn't possible to exclude things like "view" or "quality". Once you review the reason for the outliers, it is clear to you why they are selling for more or less and they are excluded. I do the same thing... I state what my search parameters are in my 1004mc and add to that "outliers excluded". In almost all cases, the feature that caused that outlier to be excluded is a significant superior or inferior component vs. my subject. And, in almost all cases, the that aspect of the property impacts its price. So price isn't what creates the outlier but is a result of what creates the outlier.
I think we all need to use some common sense in analyzing a market.
I work in San Francisco where there can be a huge variance of price within a well defined neighborhood for what otherwise may appear to be similar properties with similar physical features (size, bed/bath count, etc.).
It doesn't take long to see that the homes that sold for $600k to $700k are fixers, and the homes that are selling for $2,500,000+ are recently renovated/remodeled, or have that view of the Bay.
My home is neither a fixer nor recently renovated/remodeled; my view is the row of homes across the street. Homes similar to mine are clustered around $950k to $1,250k. At this point, I don't see the issue with refining the search using a price range of say $800k to $1,600k, and run the final search. And, if there just happens to be a similar home in all other regards that sold for $750k or $1,650,000k, who would consider that a reasonable comparable to use in light of all the other evidence? Most would consider that sale to be an outlier. And that is what I'm excluding; outliers.
The danger is presuming that a house is worth $950k to $1,250k before all else, using that as the filter, and then just analyzing the data that falls within that range. That would be incorrect (IMNSHO). But running a market search, looking at the results, reviewing the extreme ends of the range, identifying them as so dissimilar to the subject that the consequence is it becomes an outlier in the price range, and then re-running the analysis based using price as a filter to exclude those outliers is not the same as using price as a search parameter.
Those properties at the extreme end of the price range (in our discussion) are not comparable to the subject. One would exclude them from the 1004mc not because of their price but because of their dissimilarity. If after a review of the data, the properties that are significantly dissimilar to the subject consistently fall within the extreme ends of the price range, I don't need to be make a leap of faith to exclude those properties and use price as a means to do so. But I cannot make a leap of faith in predetermining where that cut-off (in price) may exist without evaluating at what range homes similar to my subject are selling at and why there are homes significantly outside that range are selling for significantly more or less.
I supplement the 1004MC with a 4-6 page market analysis of the defined neighborhood, and in it I do consider the full range of prices with maybe the exception of 1 or 2 outliers (there can be a $8,000k sale in San Francisco where the next highest sale is $2,000k within a neighborhood). So I'm not concerned about the making a material error (I'm not perfect, but my research and analysis is thorough) that might result in a mismatch in selecting the right comparables for my subject.
I'm just say, according to "da rulez" - you're wrong, even though you make more sense. 