Wednesday, September 19, 2007

August numbers, 2+2=5?

Looking at the DataQuick numbers for August a total of 1470 homes were sold in Riverside county. DataQuick calculates they sold on average 3.9% less than last year. But upon closer inspection of their data I spotted an obvious error in the way they calculate the change. They list they change by zip codes. There were 75 zip codes listed. Nearly all the zip codes were negative with the average negative amount of -12.4% (for the negative zips). There were a few zip codes with increases. Most of these were very low quantity zips with only a few sales but BIG increases. Take Thermal for instance, it had ONE sale but there was a 67% increase over last year.

So, they way they figure the average is to simply take the changes for the 75 zip codes add them up and divide by the 75. So a city with one sale and a 67% increase has the same weight as a city with 40 sales and an average decrease of 25%. Totally bogus (and lazy) way of calculating the change.

If you remove the cities with less than 5 sales the number jumps from -3.9% to -9.8%.

If you calculate it out properly with all the sales and all the zip codes the result is -9%

So DataQuick reports an average decline of 3.9% but in reality the decline is closer to 10%

No comments: