The AP has just reported on some interesting new poll results (“article”:http://hosted.ap.org/dynamic/stories/U/US_AP_POLL_UNLOVED_GOP?SITE=AP&SECTION=HOME&TEMPLATE=DEFAULT&CTIME=2010-09-24-10-05-24 ; “actual poll numbers”:http://www.ap-gfkpoll.com/pdf/AP-GfK%20Poll%20September%20Topline%2009.15.10%20final%20FULL.pdf). The bottom line: while President Obama (49% approve, 50% disapprove) and Congressional Democrats are far from popular (38% approve, 60% disapprove), Congressional Republicans are even less popular (31% approve, 68% disapprove). This led me to wonder – does anyone incorporate opposition favorability ratings into their election/forecasting models? Do we have any idea whether these really extreme negative ratings for the Republicans are likely to matter at all come November? And yes, I’m of course aware that that all the elections are ultimately district level affairs, but we routinely include national numbers for the incumbent party in these kinds of models (see “Andy’s recent post”:https://themonkeycage.org/2010/09/doug_hibbs_on_the_fundamentals.html for example).
From a more theoretical standpoint, Downs argues in _An Economic Theory of Democracy_ that there are essentially three ways a voter can calculate a party differential utility (essentially how much more one is likely to prefer what government does if one party/candidate wins as opposed to the other):
# Compare what both parties claim they will do if they win
# Compare what the incumbent party has done in office with what the opposition would have done had it been in office
# Decide whether or not one approves of what the incumbent did, and then vote for or against the incumbent without taking the opposition into account at all
Most formal models of elections eschew this third option, and essentially give voters some way to determine whether they prefer the claims/performance/likely actions of Party A to Party B. So here’s my question: if we don’t include anything *current* about the opposition in forecasting models, are we essentially assuming that voters are behaving in accordance with option 3? Or do things like past margin of victory negate the need to do so?