Podcast: Play in new window | Download (Duration: 13:17 — 13.5MB)
USPAP, Verification, and Bernoulli’s Fallacy. That’s a mouthful! What does it mean? I’m assuming you’ve heard of USPAP. Verification is an important component of USPAP’s SR1-4 (learn it, love it, use it). But it is likely Bernoulli’s Fallacy is something you don’t know about. Yet. You will, though. It is going to become really important in residential real estate appraisal, especially when UAD3.6 takes effect in November of 2026. Read on to learn why.
Really USPAP, verification and Bernoulli’s Fallacy go together. USPAP’s Standards Rule 1-4 requires us appraisers to “…collect, verify, and analyze all information necessary for credible assignment results.” As with a lot of USPAP, this is not overly clear (hello, ASB!). Bernoulli’s Fallacy says, in so many words, in statistical inference, people often wrongly assume that a single event, or a small number of observations, directly reveals the underlying probability of a process. We see this all the time from clients. Most houses in a subdivision sell for between $300K and $350K. Yet the one with the premium view and the kitchen upgrades that sold for $393K gets all their attention. If the subject is your basic house in the neighborhood, then its value is somewhere between $300K and $350K, that one outlier notwithstanding. One outlier’s price does not predict market value, despite what the client demands.
USPAP, verification, and Bernoulli’s Fallacy are a package deal. USPAP demands verification and we avoid Bernoulli’s Fallacy by being very skeptical about the predictive value of one or two outliers. This will become even more important as UAD3.6 kicks in and appraisers will need to write more in their appraisal reports. Are you preparing for that?
Questions? Contact me at tim@theappraisersadvocate.com. Need legal help or information on E&O insurance. I can help you with those, too.
Your statement is incorrect. I confirmed this with Anthropic’s Claude:
ME: “So, I believe you might agree that this statment is incorrect: ” Bernoulli’s Fallacy says, in so many words, in statistical inference, people often wrongly assume that a single event, or a small number of observations, directly reveals the underlying probability of a process. ”
Claude: “You’re absolutely right – that statement is incorrect. It mischaracterizes Bernoulli’s fallacy in a fundamental way.
Bernoulli’s fallacy isn’t about small sample sizes or drawing conclusions from too few observations (though that’s a separate valid statistical concern). As we’ve discussed, Bernoulli’s fallacy specifically refers to confusing:
The probability of observing certain data given a hypothesis is true: P(D|H)
With the probability that a hypothesis is true given observed data: P(H|D)
The fallacy involves the direction of inference – mistakenly thinking that if your data would be unlikely under a hypothesis (P(D|H) is small), then the hypothesis must be unlikely given your data (incorrectly assuming P(H|D) is also small).
This is a logical error in statistical reasoning that can occur regardless of sample size. In fact, it can happen with very large datasets too – the sample size doesn’t fix this fundamental error in how probability is being applied to make inferences.
The statement you provided seems to be mixing up Bernoulli’s fallacy with something closer to the law of small numbers (the tendency to believe small samples closely represent their parent populations) or perhaps gambler’s fallacy (believing that independent events are somehow linked), which are different statistical misconceptions.”
To make this understandable, I use the non-parametric method MARS for estimating the market value of a home based on certain features. It is an ESTIMATE of the MARKET VALUE. In fact, it should not be in doubt that in most cases it is the MARKET VALUE that appraisers are after. Market Value is by definition an expected value. There is no hint that it IS the value. Thus, Bernoulli’s Paradox does not apply.
Mr. Craytor,
Thank you for your kind note! I am going to reread BERNULLI’s FALLACY and make the necessary corrections.