Are the only reliable numbers out there in the Bible between Leviticus and Deuteronomy?
It seems that way to me sometimes — and even that Numbers takes lots and lots of context.
It hit me again this week when we dutifully reported that Oklahoma ranked No. 32 among the states in foreclosure rates last year, according to RealtyTrac in Irvine, Calif.
RealtyTrac said 12,016 homes, or 0.73 percent of Oklahoma housing units, had at least one foreclosure filing in 2011 — down 32.2 percent from 2010 and down 7.1 percent from 2009.
That's good news. Here's some context, which we also always report: RealtyTrac tracks foreclosures in 33 of Oklahoma's 77 counties, not all, and in more than 2,200 counties across the country, but not all 3,143 (or is it 3,141? Wikipedia, admittedly not the best source, is unsure).
Here's some more context, which sometimes gets lost in the blur of the news: RealtyTrac's main business is as an online marketplace for foreclosed houses. So it has a direct interest. I am not saying there is a conflict. Because we all know numbers don't lie — cough. But I can't help but wonder.
Not to shine the light on RealtyTrac unduly. The Standard & Poor's/Case-Shiller home price index is trumpeted every month all over the country as if it gauged home prices all over the country. It does not. The S&P/Case Shiller index does not include data from 13 states and it has incomplete coverage of 29 states, including Oklahoma.
But what really gets me is one main report from S&P/Case-Shiller actually tracks home price changes in just 10 metropolitan areas. Another one looks at 20 metro areas — but it's still limited data. Neither includes Oklahoma City. Keep that in mind the next time you see or read anyone making claims or drawing conclusions from S&P/Case-Shiller about Oklahoma.
Then there was this: Last summer, a poorly researched and methodologically unsound report by an online “financial news and opinion operation” zipped around the Internet claiming to call out “America's 10 Sickest Housing Markets.” It actually included Oklahoma City.
It was stunning in its absurdity — not because everything in housing here is all sunshine and roses, but because of the compilers' statistical malpractice in ignoring confidence intervals and what they mean for statistical probability, or what most of us would call the firmness of the numbers they spouted. And they were limp.
Last month, even the Federal Housing Finance Agency — which is usually the most reliable for tracking home values, in my book — stumbled with a report based on a study that correlated price appreciation with places with lots of energy jobs.
It concluded that in Oklahoma, home values in high-energy counties didn't rise as fast as in counties with fewer energy jobs, mainly because the way the number-crunchers defined “energy job” excluded Comanche County-Lawton and Tulsa County — two of the state's three metro areas.
Finally, late last year, the National Association of Realtors revised its home sales data for the past five years, saying sales were overestimated by 14 percent, which suggests that the bust was worse than people thought.
The Realtors cited a passel of reasons, including duplicate data, a big drop in for-sale-by-owner transactions and other statistical anomalies, none malicious or even that careless, as far I can tell. The Realtors said local stats from multiple listing services were sound.
To put all this in my native vernacular: Real estate stats should be taken with piles of grains of salt if they don't come with a side order of context. And if a housing economist or statistician says she loves you, check it out.