As reported on several other blogs (notably Political Wire and, I believe, Atrios?) Julius Civitatus over at
JuliusBlog has juxtaposed terror alerts on a graph of the Presidents approval ratings. (
Here) From this, he makes four claims.
I am unconvinced that Julius Civitatus' claims are, in fact, true. I have a background in science, so I approach any claims with that in mind. As much as possible, I try to examine without bias and withour prior assumption. While some cases can be identified where the 4 contentions are true, it is equally possible to find counterexamples.
As reported on several other blogs (notably Political Wire and, I believe, Atrios?) Julius Civitatus over at
JuliusBlog has juxtaposed terror alerts on a graph of the Presidents approval ratings. (
Here) From this, he makes four claims:
There are few things that are quite evident from the chart:
- Whenever his ratings dip, there's a new terror alert.
- Every terror alert is followed by a slight uptick of Bush approval ratings.
- Whenever there are many unfavorable headlines, there's another alert or announcement (distraction effect).
- As we approach the 2004 elections, the number and frequency of terror alerts keeps growing, to the point that they collapse in the graphic. At the same time, Bush ratings are lower than ever.
I am unconvinced that Julius Civitatus' claims are, in fact, true. I have a background in science, so I approach any claims with that in mind. As much as possible, I try to examine without bias and withour prior assumption. The evidence presented has to make the case for any claims. I'm using only the chart and
the post below the chart as evidence to consider the four claims - according to JuliusBlog those alone led to the conclusions. While some cases can be identified where the 4 contentions are true, it is equally possible to find counterexamples.
Claim 1: But what of the drops in approval rating that were not followed by a terror alert? For example, in the period between 8/02 and the start of the Iraq War (3/03) there are 5 notable, significant dips in approval rating (based on median rating) but only 2 terror warnings. And what of the terror alerts that did not follow a drop in approval rating? For example, the Dec. 21st alert, around the time that Saddam was captured.
Further, the chart lacks sufficient detail to really determine if a terror alert fell during a downward approval trend. A technicality perhaps, but it wouldn't hurt us to get more detailed.
Claim 2: Perhaps the most notable of several counterexamples: Sept. 02. Many of these appear to be an example of "correlation without causation" Many factors, among them "regression to the mean" as discussed above, impact approval ratings. Also, the "upticks" are inconsistent in size and shape. Some of the "upticks" in fact last several months and are very gradual or even delayed; others are more temporary, very sudden and very small. One facet of a good theory is its ability to consistently and accurately predict repeatable results. Instead the nature of the "uptick" is very unpredictable, likely indicating multiple causes.
Claim 3: This administration is continually (and rightly) faced with "bad" news. So tying a terror alert to bad news should not be all that difficult. (Nor, I suspect, would it be difficult to tie, say, Cubs loses to bad news - which, sadly, are quite plentiful.) The bad news examples in JuliusBlog are, again, inconsistently "bad." Sometimes the terror alerts "motivated by bad news" come at the end of the bad news' "15 minutes of fame/News coverage," other times they come before a story has fully played out. Ex.: JuliusBlog cites the May 12th replacement of top generals in Iraq as motivation for the alert on May 20th? I don't remember that getting 8 days of horrible press on the level of the Clarke extravaganza, for example.
Claim 4: I agree, the reports are more frequent the lower the President's ratings. Fortunately, this would seem to belie the second claim, that the alerts give President Bush a boost. The overall approval-rating trend would seem to indicate that terror alerts have only minimal immediate effect, and very little long-term effect. (Admittedly, this overlooks the possibility that the continual terror alerts are holding off what would otherwise be a large downward trend for the president. I don't see evidence for that in these charts or elsewhere - correct me if I'm wrong. The internals of polls, for example, have only recently - in the past couple months - looked negative for Bush.) And the oscillations in the approval rating indicate that if indeed claim 2 is true, the effect is getting smaller, the "crying wolf" effect. The president's political team is not dumb - if they were purposefully using the terror alerts to sway voters preferences, they wouldn't want to wear out their "power," at least not now, three months ahead of the election.
While this chart may not prove that the terror alerts are politically motivated, it does not disprove anything. I believe there is a case to be made, but this chart does not make it.
Let me be clear, I do not support this President and I am working to ensure that he is not reelected. I simply want to avoid hypocrisy. If we claim, for example, that evidence for the Iraq war was "found" where none really existed, then we cannot let ourselves "find" evidence where there is none, simply because we want it to be found. We must hold ourselves to even higher standards than we hold our opposition to. (Grammatically awkward, but true.)
Ultimately, the validity of these claims can only be proven by the holy grail-property of theories (in the scientific world at least:) their ability to repeatedly predict future events. The same of this critique - if there are errors, correct them. The beauty of science is that "disproving" something is as valuable as "proving" something! Either way, we learn something.