Heavy Lifting - thoughts and web finds by an economist
     I also contribute to Division of Labour Load HL's Front Page
Tuesday, August 10, 2004

Causation vs. Correlation

Time series econometricians have spent considerable effort trying to determine how to distinguish causation. Does X cause Y? Does Y cause X? Does Z cause X and Y? Does X cause Y and Y cause X? Seems complicated and it is.

Nevertheless, casual observations tend to confirm priors, which is not consistent with objective research but is fine for political rhetoric and aspersions.

Enter this picture that is coursing around the internet from JuliusBlog:

It seems to suggest that everytime Bush's approval rating starts to drop the terror alert level is increased, which is actually fairly difficult given that there are effectively only two terror alert levels (yellow and orange). It is not clear why Bush or the administration is overly concerned with dropping approval ratings when approval ratings are above 50% (which is the target for an incumbent president), but for conspiracy theorists this is all the correlation required to determine causation.

The terror alert levels have their own problems, far beyond any concerns for Bush's approval ratings. The major problem is the lack of variance in the possible levels. There will never be a red alert level (at least nationwide) because this would invite anarchy, looting, and general pandomonium. We will not move to blue or green because the terror threat is never really gone (we know from the wonderful Clinton years) and no government will take the chance that a major terrorist event would occur during the green/blue phase.

We are left with two levels of alert - high and higher - with no chance for differentiation. The public is thus at a loss: is the increase in the terror alert based on credible information? Does the terror alert have immediate implications? Does the increased alert imply pending doom or is it simply a "cover our ass" move by the administration?

A President Kerry would have no choice but to do the same if his Department of Homeland Security kept the same system - in which case I suppose we would have a similar picture of approval ratings and changes in terror alerts, only this time from the point of view of the Republican/right instead of the Democrat/left.

What is actually more impressive about this picture is the consistent wide disparity between the different polls of approval ratings. There is almost always close to a ten point difference between the lowest and highest approval rating - that is the more interesting statistical question.

Oh well...causation and correlation most likely confused again.

Comments: Post a Comment


Le Chai - galerie du vin



Posts that contain Craig Depken per day for the last 90 days.


Heavy Lifting's Main Page
Email Me
Atom Feed

Heavy Lifting

Great Links

Money I Found Today

Heavy Lifting - Firehose style (56k warning)

Recent Posts

- Causation vs. Correlation



Site Meter Blogroll Me!


Modified maystar design
powered by blogger