And why are statistics and “data-centric” thinking suddenly a big deal?

First, we need to know what data-centric means, and what statistics are needed.  Here, we talk about what data-analysis tools and methods are needed to know, not clever academic mumbo-jumbo stuff.

As many of you may know, I have spent the last 20 years attempting to bring basic, relevant, simple, and understandable “statistical” tools and visuals to the profession.  Not magical buttons.  And not clever statistical tests of no real relevance.  No magic.  No slick.  No stupid.

This is knowledge to sell, not software to buy.

Valuation and appraisal applications require only a very small part of the academic study of statistics.  So what parts are there?  And what do we need to know?  Is this overwhelming?

First, we need to remember what we are analyzing:  the supply and demand for a particular type of property.  A property with a few competing available properties.  And a few potential buyers.  This is not a traditional “equilibrium supply-demand” analysis.  (It’s called a game theory problem.)

Next, let’s see what “data-centric” means.  It means a focus on drawing the most out of the data available.  It means exploiting the power of the computer.  And it means using the best of modern hardware and software technology.  It means beyond “credible – worthy of belief.”

Data-centric means logic and facts and systematic selection of data (“comparables”).  It means merging the best of trained judgment and critical thinking, to that technology.

A small part of that technology, that science of data, is data analytics – inclusive of descriptive statistics.

So, what statistics apply to real estate competitive market segments?  What best describes a few comps, a few buyers, limited exposure times in a “multivariate differentiated asset”?  Unless you are taking a random sample to pick your sales — you do not need inferential random-sample statistics.  All you really need is mean, median, maximum, minimum, and perhaps percentile groups.  You don’t even really need standard deviation, COV, nor even z-score.

What you also need is visuals.  Graphs to see markets.  We measure markets, not compare comps.

These simple describer ‘statistics’ are parameters of an actual population, not of a random sample.  These simple descriptive numbers are simple.  The graphs can be done in seconds.  (Mostly histograms and scatterplots.)  The ‘new’ methods are new – but not difficult.  They present the truth in an understandable visual manner.

These new ways are important because of things called URAR, UAD, USPAP, and new GSE guidelines (requirements) for market conditions analysis.  “Appraisal is market analysis!”

NO – appraisal statistics are not hard.

As we go from the old ‘3 comps’ method to the complete market method, we can use the data descriptives: mean, median, quartiles, range, and standard deviation.  We find and deal with outliers.  We see graphs.  That’s it!

For residential, the new URAR ‘forms’ require the display of the larger data set.  For non-residential, current standards also require the use of all available information, as well as all available comparable sales, cost, and income data.  Not just some.  Not just “carefully-picked” data.

Good judgment must be combined with inclusive market analysis.  Appraisal is market analysis!