the proverbial kinds of falsehoods, lies, damned lies, and statistics (Mrs. Andrew Crosse, 1892)
Every Housing Report, whether by NAR, appraisal firms or AVMs, touts the magic word “median” to imply that the results are a true reflection of market reality. Read the articles in any newspaper or online source and this magic word is sure to appear. For those who have not studied the magical and sublime craft of statistics, Median, Average’s fraternal twin, is simply the middle number in a range of numbers. There are as many results above the median as below it. An average, on the other hand, is the sum of the data divided by total number of results.
An example: 100,000 200,000 300,000 301,000 1,000,000 2,000,000 5,000,000. The median is 301,000 while the average is approximately 1,270,000. The numbers can represent actual sales, AVM unsold values or AVM error rates. Can you tell me which better reflects reality? Perhaps neither. The raw data is the reality, the statistical computation an attempt to pull meaning from it. Does it?
James Surowiecki (author of the Wisdom of Crowds), in his recent article in The New Yorker entitled “Safe as Houses”, argues that the median may not be a useful statistic when deciding whether to invest in real estate.
Unfortunately, the numbers upon which these comforting conclusions depend—namely, median home prices for the country—are unreliable and misleading…. if you’re trying to figure out what kind of investment housing is—what rewards you can expect and what risks you’ll run—median prices become a lot less useful. In the first place, the data don’t adjust for improvements in quality….And the impact of quality adjustments isn’t trivial; a study of home prices between 1977 and 2003 found that adjusting for quality reduced the return to homeowners by forty per cent….What makes the problem worse is that sellers have recently been offering buyers huge incentives, ranging from granite counters to free cars and, in some cases, large rebates. These are, in fact, price cuts, but they never make it into the data.
Then, there’s the problem of sample bias….(which can) lead to the curious phenomenon of median prices rising even as the number of sales is plummeting and the backlog of houses on the market is soaring. Because nominal median prices compare completely different groups of homes (all those sold in August, 2005, say, and all those sold in August, 2006), they can overstate how much prices go up during booms and understate how much they go down during busts.
For those who want to know how the magician statistician performs his trick, read the article, otherwise remain entertained by the show.
