Friday, June 25, 2010

Reporting Presidential Approval

Anyone who believes that Barack Obama's approval ratings have been damaged by the oil spill should head right now to Brendan Nyhan's comprehensive examination of the evidence.  Shorter Nyhan: we don't know!  And reporters or pundits should, really, leave it at that.  See also a nice story by Greg Marx, to basically the same effect.

There's nothing new here.  Just to recap what people have been saying for years...

1.  Use all the polls, not just one poll (even if you're working for the news organization that commissioned the poll).  Always set any single poll in the context of polling averages, such as the ones at the terrific Pollster.com.

2.  Very small changes are often just random noise, not real changes.

3.  Causation is extremely difficult to prove in most cases.  Hey, we can be pretty sure that George W. Bush's approval rating record spike had something to do with the September 11 attacks -- but knowing the effect of the oil spill on minor changes (if they exist) on Barack Obama's approval ratings over the last two months is a lot harder.

4. Also, things don't necessarily move in the same direction.  For all we know, the oil spill has boosted Obama a couple of points, but the economy (or Afghanistan, or something else) has pulled him down by the same amount or a bit more. 

5.  On approval ratings, even when it's pretty clear what general event may be behind shifts in approval ratings, it's still not going to be clear what specific part of that event is doing the work (and, not all parts of it are necessarily pushing in the same direction).  So even if we know that Obama's approval ratings have gone down over the last couple months (probable, not totally certain) and even if we knew that the oil spill was responsible (we don't), that would still leave a lot more questions than answers.  Perhaps people are happy with Obama's demeanor, but not with the results.  Perhaps most Americans aren't affected by it one way or another, but there's a small segment of oil company enthusiasts who previously liked Obama but for whom Joe Barton's accusations ring true.  Perhaps the fact of the oil spill puts everyone in a negative mood and would have cost Obama 10 points, but his well-regarded performance meant he only lost two or three points (or the other way around -- perhaps the fact of the oil spill would have produced a rally effect that Obama ruined by performing badly).  Finding out that the president's approval rating changed over some period basically does nothing at all to help answer those questions. 

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.

Who links to my website?