Monday, March 26, 2007

Dan Green understands...

Dan Green at the Mortgage Reports blog makes a good point in this post. Essentially this is why I put little weight in any given release. Ever.

3 comments:

Vivek Vish said...

The article made a good point, but he also muddles the issue on another point.

No matter what the margin of error, the expected value is still the expected value.

And the margin of error is dependent of the degree of confidence one requires. So although I cannot be 95% certain that 9% construction is different from zero, I can probably be 90% certain (depending obviously on the stdev numbers). That's not too bad.

Accrued Interest said...

Yeah. A commenter on his blog made that point so I didn't want to pile on.

I look at it less from the "margin of error," because that number is based on an assumption of standard deviation. But what's the proper standard deviation? Is it 3 years? All time?

I look more at the revisions. That is the "real" error if you will. If the revisions are consistently large, then you know that a given number is indicative, but not a trend.

Anonymous said...

Thanks for going on easy on me.

Yeah, the revisions are where the truth lies, but I don't think any of the federal agencies or NAR wants to embargo data for 12 months while they figure out the "real" results.

Heh. "Truth" and "lies" next to each other looks funny...