This job is exhausting lately.

[LordofTexas]

Based upon my limited experience, IMO the minimum number of comps, of 3, is entirely arbitrary. Why not 2 or 4 or 10? I heard it's because 3 could be easily gridded across the 8.5 inch width of a page. Statistical principles reveal that the degree of accuracy increases with the sample size up to a certain number that's much larger than 3. Does anybody know the original rationale for 3?[/quote]

I'm sure it has to do with: One sale being similar but superior adjusted downward to the Subject; one sale being similar but inferior adjusted upward to the Subject; and (like the three bears) one that's juussssst right! Funny thing is, most lenders I've known freak when you have 3 or 4 model matches on the appraisal without any adjustments. They ALWAYS seem like they need just 1 more Comp, lol.

Lordy

 

[glenn walker]

In this market I flood them with comparables and pendings and listings, the underwriters and reviewers love having a big fat file to tell their boss that they did due diligence, also for the younger appraisers that old thing called "bracketing" is back so give um big homes, little homes and middle sized homes , then tell them you bracketed your sales , old reviewers , get all happy just thinking about being on the old 1980 FHA fee panel with the rotating chums number assignments , and hearing those old FHA instructors talking about how the sales need to be bracketed.... yeah baby it was bracketed...)

 

[Rice Brewer]

Does anybody know the original rationale for 3?

If I remember right, you need a minimum of 3 data points to calculate Standard Deviation.

Either that or the previous post was right - 3 and the SP will fit on a 8.5 wide piece of paper.

My money is on Reason #2

 

[xmtraterbcra]

Based upon my limited experience, IMO the minimum number of comps, of 3, is entirely arbitrary. Why not 2 or 4 or 10? I heard it's because 3 could be easily gridded across the 8.5 inch width of a page. Statistical principles reveal that the degree of accuracy increases with the sample size up to a certain number that's much larger than 3. Does anybody know the original rationale for 3?

Statistical significance relates to proving something is true with a certain degree of certainty. The basic assumpition is that you have a nearly infinitely sized population of samples to choose from - but need to keep expenses minimized by choosing the smallest sample which will give you the degree of certainty you need. For example, collecting "samples" may mean randomly calling people on the phone to ask them their opinion - and that takes time and time is money.

This doesn't completely apply to sales comparables for several reasons:

1. If you are asked to appraise a property and the available evidence is meager to poor - you are still expected to come up with a value based on the "best evidence" you have. That is to say, we don't "reject" a statistical hypothesis (e.g. that 3 comps is sufficient for an estimate) because the reliability of the derived average adjusted price falls below a certain degree of significance. That option is just not available, if 3 "reasonable" comps is all we have.

2. Our samples are almost never perfect matches for what we want. We have to make adjustments - many of which involve subjective judgement.

3. The samples and adjustments have to be reviewed and this takes time which costs money. Thus, too many samples could be counterproductive.

4. There are other time and resource constraints.

5. Other ...

Deciding on the optimal number of comps for a particular assignment, depends on a number of factors - the central being the judgement of a knowledgeable and experienced appraiser, followed by applicable guidelines or SOW.

Bert Craytor, SRA