Poll data

This one landed in my inbox this morning, so of course I felt the need to blog.

First, some minor notes. I might have preferred the Y-axis to go from 0-100, since I feel somewhat manipulated by the exaggeration; on the other hand, the one who's making the graph is clearly trying to make a point. Secondly, most polls tend to be bunched around an average. That's good news, and can lend well to predictive analysis.

Assuming that the questions asked were more or less the same--in this case probably something like "do you approve of George W. Bush"--predictability measures would contain some data incorporating response volume. For lack of statistical training, I'm thinking something along the lines of Bollinger Bands and MACD, etc. These would help in determining whether those previously "no" responders actually switched over their vote to "yes" or if they chose not to respond following a dramatic event like 9-11 or the beginning of the Iraq war. To extend the analogy: did the shorts go long or hold, and by extension, once the crisis is deemed over or almost over, did they go back to shorting or stay with hold until enough time had passed to which they were confident in their short position again (bad analogy, I know, but I think one can get the point). These longer-term analyses could allow a short-term predictive value in assessing how--all thing remaining equal--the mood of the nation, or at least a fair subset thereof, is heading.

I'm sure there's those measures out there, and I just haven't found them. I suppose that if volume were the same, one could make the assertion that foreign policy-related matters tend to evoke lower highs and rebound to slightly (though maybe not significant) lower lows. Which is why the tone flopped on the N. Korea issue and Israel-Palestine has been relegated to damage control.

Another phenomenon this graph illustrates is the "rally-around-[the President]" reaction. I vividly remember two polls in particular, one immediately preceding the Iraq war and one during the first week of combat. The before question asked was whether you feel the President should attack Saddam without UN approval (somewhere around 15% of respondants stated "yes"), and the after question whether you approve of the President going to war, implicit was "without UN approval" (somewhere around 60% favored the idea). That's a huge jump from one week to another. Maybe in part due to the Vietnam scars and numerous other factors, it seems that this is a now-traditional American pattern; that as soon as the bombs start falling the majority of those opposed flock to the other side. But do their feelings about the course of action itself actually change? Seems to me that it'd be reasonable that some people's opinions might have changed after, say, the statue of Saddam was toppled, or some other similar event. But how can it be that responders are strongly enough opposed to a particular course of action to respond "no" on a poll, and then the very next week that exact course of action is undertaken, and all of a sudden a significant proportion responds "yes?" In my eyes, this is not a matter of support for the troops. Luckily Americans are instilled with this deeply rooted virtue, and also luckily, we have the right to interpret how to express our support for the troops. What this is simply a matter of: Action A or Action B, which one do you prefer?

Either way, interesting to ponder.

Thanks Joe