Lies, Damn Lies, Statistics…. and Headlines

Sometimes our favorite statistics mislead us.  I was most recently reminded of that when I was reading one of Ardell’s North King County Stats post – Ardell does a great job on these stats, and sales volume and median price tell a lot of the story.  But there is another dimension we need to keep an eye on, and that is the change in mix over time – particularly the change in the ratio of number of higher price homes sold to number of lower-price homes sold.  That change in mix can make the same set of statistics generate a variety of very exciting, or depressing, headlines.

Here’s an example: if over a year or so the number of sales of high-priced homes drops a lot, and the number of sales of low-priced homes doesn’t drop as much, then both the median and the average prices are affected dramatically.  If we are concerned that our home prices are dropping, we watch that median number like a hawk – we’ve been trained that ‘median’ is better than ‘average’ for telling what’s really going on.

Suppose we build an example set of data where the mix of home sales has changed dramatically over the course of a year or so, and the total number of homes sold has dropped in half, but the value of individual homes has dropped only 10%.  As it turns out, there are lots of ways to cut the data, and some of them yield more exciting statistics than others.  We all know the old saying “Bad news sells newspapers

14 thoughts on “Lies, Damn Lies, Statistics…. and Headlines

  1. The case shiller index looks at same homes sales over time to reduce the impact of mix changes. Its results have fairly closely mirrored the NWMLS median price changes over the last couple of years.

  2. I think you mean “median is better than the arithmetic mean” since average is a general term referring to a measure of the “middle” of a set.

    Anyways, the changing mix of houses selling has been discussed at length many times here and elsewhere and is why Case-Shiller is generally regarded as the most accurate measure of house prices changes. There was even an attempt to view the breakdown of sold houses and approximate Case-Shiller numbers in advance here: http://seattlebubble.com/blog/2007/08/14/median-price-not-telling-the-whole-truth/

  3. I appreciate this topic being addressed in this manner on the site as the headlines of papers do usually miss the mark of what’s going on in local markets. Sometimes a specific market area can be extremely micro in size and not as impacted as other neighborhoods, but buyers are reading broad data.

  4. Good post. When posting the median graphs on my blog that is why I say “This is the median or average price of ‘homes that are selling” – not the median price of all homes. The fact that sales are so heavily weighted in the $200K – $400K range (at least in Snohomish County) and nearly non existant in the above $500K we can no longer rely on the median or average as a true reflection of home values as a whole. It is only telling us what price ranges that homes are selling in.

  5. I tackled the same question on my blog about 1Q numbers in Leavenworth.

    http://iciclecreekrealestate.com/2009/04/05/leavenworth-real-estate-first-quarter-2009-market-update/

    Here are the options in my market:

    Leavenworth Home Sales Drop 60%!
    Leavenworth Home Prices Plateau!
    Leavenworth Home Prices Up 7%!
    Leavenworth Real Estate Investments Trounce Wall Street!

    Lots of choices and ways to spin it. I guess the truth is always a bit more complicated than a headline.

  6. I tackled the same question on my blog about 1Q numbers in Leavenworth.

    http://iciclecreekrealestate.com/2009/04/05/leavenworth-real-estate-first-quarter-2009-market-update/

    Here are the options in my market:

    Leavenworth Home Sales Drop 60%!
    Leavenworth Home Prices Plateau!
    Leavenworth Home Prices Up 7%!
    Leavenworth Real Estate Investments Trounce Wall Street!

    Lots of choices and ways to spin it. I guess the truth is always a bit more complicated than a headline.

  7. Thanks, Reba. We have had a pretty dramatic drop in sales volume and number of transactions over the past two years, and an even great drop in sales of high-end homes. However, Dr. Short is right that the Case-Shiller Seattle region index has tracked our King County NWMLS median home price pretty closely. I just did a quick check of the period August 2007 to February 2009 (most recent C-S data). If you scale up the C-S index by a factor of 2,400, you get a number that tracks our median home price within a range of +/-5%. Joel, your reference to Tim’s August ’07 article is a good one. Perhaps we can get The Tim to do a reprise 🙂

  8. I plucked this sentence from today’s MSN Money, Market dispatches, 5/12/2009, first paragraph:

    “The median price of a home fell 14% in the first quarter, the National Association of Realtors says.”

    OK, I’m thinking that’s a pretty big drop in one quarter. Better check it out. 14% is about what I am mentally framing for the entire year.

    Well, it turns out that it is the entire year.

    So, it’s not just the headlines. The impresicion of language continues to cloud the picture, and our inability to understand the meaning of numbers worsens it.

    Thanks for the post.

Leave a Reply