June 26, 2006

In Defense of Darrell West, or the Theory of the Surly New England Independent, Part 1

Carroll Andrew Morse

In anticipation of the release of June’s Brown University(*) poll on the upcoming November elections, I compared past Gubernatorial and Senate election results to the corresponding Brown poll taken closest to the election date. I was curious to see if evidence of the current conventional wisdom -- that in races involving incumbents, undecided voters tend to “break” in favor of the challenger -- existed.

Three Rhode Island Gubernatorial or Senate races since 1994 have involved full-term incumbents seeking re-election. In none of these cases did undecideds (as measured by the Brown results) break in favor of the challenger...

  • During the 1994 U.S. Senate race, the September poll showed incumbent John Chafee leading challenger Linda Kushner 55%-24%. Undecided voters broke relatively evenly, with Chafee winning the election 65%-35%.
  • During the 1998 Governor's race, the September poll showed incumbent Lincoln Almond leading challenger Myrth York 41%-35% (with only 13% undecided, due mostly to the presence of Robert Healey). In the actual election, undecideds broke slightly in favor of Almond, giving him a 51%-42% victory over York.
  • During the 2002 U.S. Senate race, the October poll showed incumbent Jack Reed leading challenger Robert Tingle 61%-14%. Undecideds broke in favor of Reed, giving him a 78%-22% victory.
A fourth race involving an incumbent who had received a late-term appointment to his seat did exhibit a break in undecided voters towards the challenger...
  • During the 2000 U.S. Senate race, the October poll showed incumbent Lincoln Chafee leading challenger Robert Weygand 52%-28%. Undecideds broke in favor of Weygand, though Chafee still won a comfortable 57%-41% victory.
There have also been three Gubernatorial or Senate races involving no incumbent since 1994...
  • In the 1994 Governor’s race, the September Brown Unversity poll showed Lincoln Almond in a dead heat with Myrth York 38%-37%, with 21% undecided. Undecideds broke slightly in favor of Almond, but enough to give him the 47%-43% victory (Robert Healey picking up 9%).
  • In the 1996 U.S. Senate race, the September poll showed Jack Reed with a 49%-32% lead over Nancy Mayer. In the actual election, undecideds broke in favor of Reed, giving him a 63%-35% victory.
  • And, of course (in the only case where the Brown poll got the final result wrong) the October poll showed Myrth York leading Don Carcieri by a 41%-34% margin. In the election, undecideds broke for Carcieri by a large enough margin to give him a 54%-45% victory.
There’s at least one technical issue affecting poll accuracy that goes beyond the scope of the data presented: making accurate “likely voter” adjustments. Despite this large X-factor, it's still possible to draw a few conclusions.

First, there's no obvious ideological bias in the Brown University result set. In the three cases where undecideds split evenly, (John Chafee, both Almond/York races), everyone’s support was understated about equally. In three cases (Reed in ’96 and ‘02, Weygand in ‘00) eventual Democratic support was understated and in one case (Carcieri in ’02) eventual Republican support was understated.

Second, the results suggest a necessary refinement to the idea that undecideds not supporting an incumbent are likely to vote for a challenger. In Almond/York II, for instance, the controlling dynamic was probably as much "if I didn't vote for her before, why would I vote for her now?" as it was a referrendum on the incumbent. The number of undecideds that a challenger picks up is likely inversely proportional to how well-known the challenger is going into the election.

Third, most interestingly, there may be a factor called the phenomena of the surly New England independent in play here...

(*)The Brown University poll is conducted by Brown University Political Science Professor Darrell West, Director of the Taubman Center for Public Policy and the John Hazen White Sr. Public Opinion Laboratory. I mention this in full detail because 1) I believe in giving full credit where credit is due and 2) because I would never think of slighting the fine contributions the Whites have made to the civic culture in Rhode Island, especially when John Hazen White Jr. might be looking in innovate directions to find for moderators for a series of political debates.

Comments, although monitored, are not necessarily representative of the views Anchor Rising's contributors or approved by them. We reserve the right to delete or modify comments for any reason.


Very interesting and insightful piece. To me the big issue in the Nov elections with the governors race is the casino. If the people get the right to vote on the referendum then many undecides will stick with Carcieri. Poeple have been angry with what they perceive as the governor blocking their ability to vote on a casino. If they get that right many will forgive Carcieri. On the other hand if the governor goes to court and this question is again blocked from the voters then Carcieri is all done. The casino and the people's perception that the governor has blocked their ability to vote has been the big issue which has eroded the governor's numbers.
Letting the people finally get the casino question will be a good thing for the governor in November.

Posted by: Tim at June 26, 2006 4:18 PM

I'm not sure there's any hard and fast theory about how the late undecideds break - it's at the mercy of events and the candidates' tactics.
In '96, Reed got the late break because of the heavy-handed ad campaign the RNC launched against him (much to Nancy Meyer's displeasure). In '02, York made an attack on events overseas at one of Carcieri's companies which just didn't resonate with R.I. voters - she overreached. Reed and Carcieri capitalized on the opposition's mistakes.

Posted by: Rhody at June 29, 2006 11:35 AM

The October Chafee/Weygand October 2000 poll may have been influenced very heavily by the fact that John Chafee passed away around that time. I don't know if it was before or after the poll, but it may be a reason for why that poll was different.

Either way, your analysis does lend credibility to West's polls. Generally, he appears to have been accurate in the past.

Posted by: Anthony at July 5, 2006 6:49 AM