The USA 2009 races that attracted nationwide interest

(Skip to conclusions.)

2009 was a non-election-year in most of the USA. Exactly three races attracted nationwide interest:

1. The New Jersey governor race.
2. The New York district-23 congressional race [special election to replace former Congressman, now Army Secretary, John McHugh(R)]
3. The Virginia governor race.

The last of these was uninteresting since it was a two-man race [McDonnell(R) vs Deeds(D)] with McDonnell holding a steady 10-to-18 point lead in all polls. The first two races both are excellent examples of what is wrong with the USA's plurality voting system and why score voting, also called range voting, would be an improvement.

The New Jersey Governor race

The contenders (pictured left-to-right):

• Jon Corzine (Democrat, governor seeking re-election and extremely wealthy from previous career as chair & co-CEO of the Wall Street firm Goldman-Sachs)
• Christopher J. Christie (Republican who'd served as US attorney for district of NJ 2002-2008);
• Chris Daggett (Independent, PhD in education and chief of staff for former gov. Tom Kean, also former regional administrator of EPA and commissioner of NJ Dept of Environmental Protection).

In my personal opinion (I live in neighboring New York) all three seemed unusually attractive and qualified compared to typical candidates, and I rather wish they were running in New York instead...

Corzine clearly had the greatest familiarity with the job (not surprisingly since he was in the job) but many felt he had not performed as he should, with, e.g, NJ forecast to have a $8 billion deficit in 2010; and Corzine's extreme wealth (and willingness to use it to, essentially, "buy votes" and support the "old boy network" of "pay to play" NJ politics, rather than striving to clean it up) left many disenchanted – especially in the economic post-crash climate in which Goldman-Sachs had been heavily implicated. E.g. 44, including three NJ mayors, two state assemblymen, and 5 rabbis, were charged in July as the culmination of a federal money-laundering and corruption 2-year investigation. (Corzine: "Unfortunately you don't give people lie-detector tests and then ask them for their vote in the Legislature.") Christie attacked him for "too many taxes" and his abandoned scheme to sell the New Jersey Turnpike to raise cash. However, Christie was not exactly in a great position to exploit anti-Wall-Street angst since his own wife and brother were themselves both wealthy Wall Streeters. Finally, Daggett attracted support as the apparently least-bought, most-independent-seeming candidate.

All three debated several times throughout October on radio and TV, with the final debate on the 22nd. Daggett appeared to have "won" the debates in the public mind (to a greater extent than the others, anyhow); he also benefited from Corzine and Christie each using their large financial and hence media resources to savage each other.

The polls were as follows. First, there were score-voting style polls with 0-100 score "temperature" range conducted by "Democracy Corps" (640 likely voters, Oct 20-21; 604 likely voters, Oct 27-28):

(Mean scores shown.) DailyKos/Research2000 also conducted their own score-voting-style polls, but now using an {0,1,2,3} scale instead of 0-to-100:

These polls constitute a near 3-way tie, most probably with Daggett the winner if any one is to be singled out.

Second, here are various approval-voting-style polls, conducted respectively by PPP (Oct 23-26 of 630 likely voters by telephone), Quinnipiac University (Oct 20-26; released Oct 28, 1267 LVs), Rasmussen Reports (telephone survey 26 Oct; 1000 LVs), and MonmouthUniversity/Gannett (722 LVs). Call these "P," "Q," "R," and "M."

Note that both Corzine and Christie have more disapproval than approval; only Daggett is, overall, (slightly) approved.

Third, the Quinnipiac poll also asked Daggett voters only whom they'd take as their second choice:

This also was asked by a later PPP poll (Oct 31-Nov 1, 994 LVs) which, note, got different results, perhaps because of voters leaving Daggett to support Christie during the intervening period. Keep in mind that these "second choice" polls had small sample sizes and hence large error bars of order ±10 points. For this reason they did not really contradict one another, and combined they suggest Daggett voters would prefer Christie (54%) over Corzine (46%) ignoring the "don't know"s.

I am unaware of any analogous "second choice" polling data for Corzine and Christie voters, but speculate that most would prefer Daggett over the other C. If so, then Daggett is the "Condorcet winner," i.e. would beat either opponent in a head-to-head race.

But unfortunately, the election was not conducted via score or approval (nor rank-order-ballot) voting; it was conducted via plurality voting. This meant the usual "must vote for one of the top two" strategic dynamic came into play to (massively) distort public opinion. The media played its role by widely urging voters not to "waste their vote" on Daggett while the New York Times ran a front page story (30 Oct) about their interviews with Christie and Corzine, but refusing to discuss or interview Daggett.

The PPP Oct 26-28 poll asked "Do you think it is possible for Chris Daggett to win?" Results: 14% yes, 72% no, 14% not sure. Obviously if 80% of voters think Daggett has no chance so that a plurality-vote for him is "wasted" no matter how objectively-good a candidate he is... then Daggett indeed does have no chance!

Real Clear Politics, by averaging many different party-unaffiliated pre-election polls conducted October 20-Nov 1 totaling over 6000 likely voters, found an almost exact (which certainly is within the statistical error) 2-way tie:

Note the tremendous distortion caused by plurality-style voting: the only candidate approved more than he is disapproved (who also is probably the Condorcet "beats all" winner) is with plurality voting far behind with no chance of victory! Note that IRV would have eliminated Daggett in the first round (same "center squeeze" thwarted-majority pathology as in Burlington) and hence would not have repaired that pathology. However, if Corzine had won the election, it quite plausibly would have been because Daggett played the role of a "spoiler" to deny votes to Christie, and IRV would repair that pathology.

The NY-23 congress race

The contenders were William L. Owens (Democrat, Air Force Captain, lawyer, and adjunct professor of business law; also endorsed by "Working Families Party"), Dierdre "Dede" Scozzafava (Republican Assemblywoman, also endorsed by "Independence Party"), and Douglas L. Hoffman (Conservative party; manages an accounting firm; no political experience but was controller of the organizing committee for the 1980 Winter Olympics in Lake Placid, where he lives, and was a staff sergeant in the army NY National Guard), pictured left-to-right. The district was the largest and northernmost in New York and with an amazing shape gerrymandered to resemble a diseased letter "M." It and its predecessors had elected a Republican every time since about 1870, but the "reliable Republican" era seemed perhaps at risk of ending, since, e.g. this district had just voted for Obama(D) by 52-47% in the 2008 US presidential election.

A radio and TV debate were held. Hoffman did not come to the North Country Public Radio debate and apologized later for his no-show. All three came to a TV debate held in Syracuse Oct 29. However, Syracuse lies well outside the district. A video recording of that debate was scheduled to be aired within the district on 1 Nov at 7pm on ABC 50. (Election day: 3 November.) Those who heard the debate did not regard it as either devastating or incredibly uplifting for any of the three.

(SUM and AVERAGE based on 3, 2, 1, 0 scores.) Based on these polls, the race is close, with Scozzafava being the SUM winner, but Hoffman would win based on AVERAGE. Based on both, however, Owens might win. (The reason for this discrepancy was that Scozzafava, the only candidate who previously served in elected office, was the best-known and hence got the fewest NO-OPINION scores; while Hoffman as the least-known got the most.) Approval-voting-style polls were also conducted:

Based on these results, again Scozzafava would have won based on total number of approvals (before she dropped out); but we include the poll afterwards or base it on the approve/disapprove ratio, Owens would win; albeit if we only use the later polls Hoffman is ahead of Owens. Again the reason for the total-vs-ratio discrepancy is that Scozzafava was the best-known candidate i.e. got the fewest NO-OPINIONs. (In a real election instead of a poll those discrepancies would shrink because few or no voters would remain undecided.) The Siena Mid-October poll indeed found that "Hoffman remains the least-known candidate; nearly half of those who say they are voting for Hoffman don't know enough about him to have a favorable or unfavorable opinion"(!).

The DailyKos/Research2000 oct 19-21 poll also asked Hoffman-voters only: "Which candidate would be your second choice, Bill Owens, the Democrat, or Dede Scozzafava, the Republican?" Result:

I am unaware of any analogous "second choice" polling data for Owens and Scozzafava voters, but here is my approximate speculation:

In the event my Owens-2nd-choice guesses are correct, then we see that (at least, as of about Oct 28) Scozzafava was the Condorcet winner, i.e. would have defeated either rival in a head-to-head race. This also indicates that this election would have exhibited both "favorite-betrayal" and "thwarted majority" malfunctions if it had been conducted with instant runoff voting (IRV). Specifically (consult first-choice polls below), Scozzafava would have been eliminated immediately by IRV. if the election-winner is Owens, that means the Hoffman voters would have been better off betraying their favorite and instead voting for their honest second choice – Scozzafava. If the election winner is Hoffman, then the Owens voters would have been better off betraying their favorite to vote Scozzafava. Either way, with IRV voting for your true favorite was a strategically-stupid move for either the Owens or Hoffman voters, which caused their most-hated candidate to win! (This is yet-another refutation of the myth that with IRV you can vote for your true favorite "without fear of 'spoiler'-effects.")

IRV only works well in "two-and-a-half candidate" races, not in genuine 3-way races like this one.

However, unfortunately, the election was not conducted via score or approval (nor rank-order-ballot) voting; it was conducted via plurality voting. Here are the results of plurality-style pre-election polls, in reverse-chronological order, with "(R)" for polls conducted by groups affiliated with the Republican Party.

As you can see, Scozzafava initially held the lead in these polls, peaking in late September with a 7-point lead on her nearest rival. However, then Owens assumed the lead, and in the final polls it became an approximate tie between Owens and Hoffman, with Scozzafava 11-16 points behind.

What happened? In a historically-Republican district and with Scozzafava(R) reasonably popular and the only one of the three candidates who'd held elected office, initially it seemed she was going to win easily.

However, fringe ultra-right elements in the nationwide Republican Party such as (former Alaska Gov.) Sarah Palin, (former Pennsylvania Sen.) Rick Santorum, and the "Club for Growth" saw her as "too liberal" and swooped in with cash, endorsements, $300,000 in TV ads, and a publicity campaign to try to boost the campaign of the initially-little-known Hoffman and try to make this race a "referendum on the future of Republicanism" and to "purify" the Republican party. Scozzafava was indeed remarkably Democrat-like for a Republican, e.g. she was one of only four Republicans in the NY legislature to support "gay marriage," and she supports the legality of abortion, contrary to the Republican Party Platform. (Although actually, by removing Scozzafava from the NY legislature to the US congress, her stances on those two issues would become far less relevant.) Plus (yikes!) her husband held a position in a union!

Scozzafava's husband, Ron McDougall, was president of the Jefferson-Lewis-St. Lawrence Central Trades and Labor Council, and endorsed Owens after his wife withdrew (and before she endorsed Owens). She had the endorsement of a UAW local, and the NY AFL-CIO had stayed out of the race. Somewhat lost in the media frenzy about Scozzafava's Democrat-like stances on those particular two issues, was the fact that she had voted 95% with the Republicans while a NY assemblywoman.

It worked: Hoffman then grew in popularity and became better-known, although even near the end of the campaign almost half his own voters said they still did not know enough about him to rate him "favorably" or "unfavorably." Hoffman was also aided by the fact that Scozzafava, who perhaps had not perceived him as a threat until too late, failed to prepare attack ads against him. Such ads would have been easy: Hoffman actually did not even live in his district (albeit this is not quite as peculiar as it sounds, see map) and he (accompanied by former Republican leader, now chair of "FreedomWorks," Dick Armey) suffered an embarrassing interview with a local newspaper, the Watertown Daily Times, in which Hoffman's ignorance of local issues was made highly apparent, e.g.

When asked about the rooftop highway that could connect Watertown to Plattsburgh, Mr. Hoffman said he was "open to reviewing and studying it." When asked about winter navigation on the St. Lawrence Seaway, he took no position. When asked about widening and deepening the Seaway, he again said nothing.

(The newspaper was furious. I don't think I've ever seen such a negative reaction to a candidate by the newspaper interviewing him.) Also Mary Snow on CNN interviewed Hoffman. She asked him to name the three local issues most important to his district. He couldn't do it!

Fears then arose that thanks to a Scozzafava-Hoffman "split" of the rightist vote, the Democrat Owens might (unjustly) win. [And fueled by that, the Democrats swooped in with funding for Owens, causing Scozzafava(R) to be outspent by both sides, a highly unusual development in this sort of district.] That fear naively sounds plausible. However, the score- and approval-style poll data above do not greatly support that theory. Score and Approval voting are immune to "vote splitting" and they found high support, plausibly enough to win, for Owens. I conclude, therefore, that an Owens victory would not be "clearly unjust" and not clearly caused by a "vote split." This data instead supports the notion that the NY-23 voters were not so unsophisticated. They (or anyhow a lot of them) actually saw all three candidates as different and did not view Scozzafava & Hoffman as mere "clones" (or Scozzafava & Owens either).

The Grand Poobahs of the Republican Party, however, apparently were not interested in score- and approval-voting data and feared in their usual habitual unsophisticated manner that this race was going to be a Republican-Conservative "vote split" pathology as usual. They apparently did not consider that this race might be atypical in that respect and had to be viewed in a less-naive way.

Hence, top Republicans apparently made a strategic decision, in the final week of the campaign, to backstab their own candidate Scozzafava. Former NY governor G.Pataki(R) and future(?) NY governor R.Giuliani – the two most popular and best-known Republicans in the state – both suddenly came out with widely-publicized endorsements of Hoffman. Anonymous Republican "strategic consultants" were quoted in the press as saying they had no choice because Scozzafava had dropped in the polls and hence was clearly going to lose – votes for her would be "wasted." They felt they had to give Hoffman an extra boost to try to ensure he would defeat Owens.

It was clear that Giuliani's endorsement of Hoffman was a purely-cynical strategic calculation since, e.g. Giuliani: "I'm pro-choice. I'm pro-gay rights." And Pataki had also come out in favor of abortion rights, signed New York's gay rights bill into law, and appointed the openly-gay Mayor D.Stewart of Plattsburgh (which lies in district 23) into his administration – although Pataki does oppose same-sex marriage. (Incidentally, it is hard to believe these two luminaries were unconsulted when the NY Republican Party initially nominated Scozzafava; nor does it seem believeable the GOP would have done so if either had registered strong objections.)

The trouble with that strategy is this: No poll data was available about the "second choices" of Scozzafava voters.

It is conceiveable, but unlikely, the Republican Party had access to such data via their own private pollsters. I say "unlikely" because such private polls would normally be conducted by Scozzafava herself, not the nationwide party, and she would have no motivation to enquire about that question – and also no motivation to divulge this data, even if she had it and it pointed Hoffman-ward, to those likely to backstab her.

It is entirely possible that more Scozzafava voters preferred Owens than Hoffman. [Owens arguably was more similar than Hoffman to Scozzafava...] If so, then by backstabbing Scozzafava and telling her voters not to "waste their vote" on her, the result would actually be to help Owens win. If so, then, those Republicans would have been better off (which anyway was the honorable thing) supporting their own candidate. Sometimes the obvious move actually is the right one! (And since Scozzafava had led the polls only 1 month before, it would seem at least conceivable that a sudden infusion of support and money from the national party might be enough to allow her to come back and win.)

But that did not happen. The backstab fired up on about Oct 28; Scozzafava duly collapsed in the polls, and the usual "must vote for one of the 'top two'" dynamic of the flawed plurality voting system slammed into gear. The collapse in Scozzafava's poll numbers did not seem clearly to be benefitting either Hoffman or Owens more than the other in the pre-election polls, so it remained unclear (to me, anyhow) whether this backstab actually worked for the strategic purpose the stabbers had in mind, or the opposite!

Scozzafava then, amazingly, suddenly saw that it was in her, her political future's, and/or her party's best interest for her to withdraw from the contest!

Dede Scozzafava, the Republican and Independence parties candidate, announced Saturday [Oct 31] that she is suspending her campaign for the 23rd Congressional District and releasing all her supporters.
The state Assemblywoman has not thrown her support to either Doug Hoffman, the Conservative Party candidate, or Bill Owens, the Democratic candidate.
"Today, I again seek to act for the good of our community," Ms. Scozzafava wrote in a letter to friends and supporters. "It is increasingly clear that pressure is mounting on many of my supporters to shift their support. Consequently, I hereby release those individuals who have endorsed and supported my campaign to transfer their support as they see fit to do so. I am and have always been a proud Republican. It is my hope that with my actions today, my party will emerge stronger and our district and our nation can take an important step towards restoring the enduring strength and economic prosperity that has defined us for generations."
Ms. Scozzafava told the Watertown Daily Times that Siena Research Institute poll numbers show her too far behind to catch up – and she lacks enough money to spend on advertising in the last three days to make a difference. [Quoted from Watertown Daily Times 31 October story by Jude Seymour.]

This didn't make sense to me. The TV debate between the three had not yet been broadcast at the moment Scozzafava withdrew. If I'd been her, I would have stayed in the race at least until after the broadcast! In view of this it seemed likely Scozzafava must have been placed under great pressure by her Party.

If we disregard strategy and just view things from the standpoint of party loyalty, the backstab was reprehensible. It somewhat resembled the previous Republican backstab of Alan Schlesinger(R) in the 3-way CT senator race of 2004, where the Republicans switched their support to Joe Lieberman (Independent, former Democrat) who (sure enough) won. That move had been understandable since Schlesinger had been involved in scandals involving unpaid gambling debts and fake names and was polling 8% versus Lieberman (50%) and Ned Lamont (38%). But Scozzafava was not involved in any scandal and while trailing in the polls, was not anywhere near as tremendously behind as Schlesinger had been; indeed she'd been leading only 1 month before. Further, while the Schlesinger-backstab worked to ensure Lieberman's election, it was far less clear the Scozzafava backstab would "work." It is even possible some fraction of the voters, because of their disgust at this backstab, decided to vote anti-Republican.

Bruce Gilson comments: This election is approximately the mirror image of the Burlington mayoral election. Instead of (as in Burlington) having a relatively liberal constituency with a right-of-center, left-of-center, and far-left candidate, there is a relatively conservative constituency (Obama may have won it, but it hasn't sent a Democrat to Congress since 1870) with a left-of-center, a right-of-center, and a far-right candidate. So Owens is the equivalent of Wright, Scozzafava of Montroll, and Hoffman of Kiss. And if Burlington had used plurality, Wright would have won. So the analogy would point to Owens winning.

Scozzafava was leading in the early polls, because it's a Republican district and third parties are usually discounted. As people got to realize that the split in the right wing would elect Owens, they consolidated. (A split GOP elected Woodrow Wilson in 1912.) To avoid this outcome, they're all moving to Hoffman. It's doing the equivalent of the Burlington IRV process, but before the election, instead of during the counting.

Late-breaking update: The nationwide GOP, including

Minority Leader John Boehner of Ohio, Former House Speaker Newt Gingrich, Republican National Committee Chairman Michael Steele (these three had previously endorsed Scozzafava), Republican Whip Eric Cantor of Virginia, and Texas Representative Pete Sessions, chairman of the National Republican Congressional Committee

all, immediately after Scozzafava's withdrawal statement above, announced that it/they now endorsed Hoffman. But then Scozzafava herself stunned the GOP by announcing she'd decided to endorse Owens! (During the intervening period, she'd "reflected," consulted some prominent Democrats including NY Senator Charles Schumer, and also been basically insulted by Hoffman while praised by Owens.) Hoffman then denounced her as a "turncoat," while the Watertown Daily Times, which had been endorsing Scozzafava, switched its endorsement to Owens, claiming Hoffman was running as an "ideologue" while Owens would "serve [the district's] interests first and foremost." The Ogdensburg Journal felt similarly.

This move enhanced and partially-confirmed our "failed-backstab" hypothesis above – that actually Scozzafava voters might move toward Owens more than Hoffman. The two final pre-election polls by Siena and PPP (not shown) both happened during the unfolding of this soap-opera, so it was highly unclear exactly what the voters were going to make of it. Siena saw a doubling of the number of "undecided" voters and argued that whomever those undecideds eventually supported ought to win. PPP thought Hoffman was clearly now going to win and indeed that he appeared (based on "A versus B" comparative polling) now to be the "Condorcet winner," but looking at their poll data both before and after the Scozzafava endorsement of Owens (they could do this because they'd kept track of the time; mostly this poll occurred pre-endorsement) it appeared that this endorsement was causing a 2% drop in Hoffman support and a 7% rise in Owens support (these percentages are in terms of the full voter pool) so that wasn't really clear. Further, all these polls mostly or entirely occurred before the TV debate broadcast! So the race was still developing in an unclear direction on the eve of election day.

The election results (from CNN)

Virginia Governor:

McDonnell wins with 59% versus Deeds 41%. [CNN based on 99% of the vote.]

New Jersey Governor:

Christie wins with 49% (1132698) vs Corzine 45% (1026899) and Daggett 6% (132181) [CNN based on 99% of the vote.]

New York district-23 congress seat:

CNN projection: "The three-way race in New York's 23rd congressional district ended Tuesday night with a surprise Democratic win – the first for the party in the reliably-Republican district since the 19th century. Democrat Bill Owens defeated Conservative Party candidate Doug Hoffman to claim victory in a race where an internal GOP fight drew national attention – and forced the party's candidate, Dede Scozzafava, out of the campaign."
The results based on counting 92% of the vote were Owens 49%, Hoffman 45%, Scozzafava 5%. Consequently Hoffman conceded on 3 Nov, and Owens was sworn in. Scozzafava later remarked on national TV that she thought that if such Republican internecine "purges" continued, it would risk the downfall of her party, and (further) she believed that a diversity of views are required in government to solve problems, and this sort of "purification" of Republicans could abolish that diversity.
The only problem: counting and recounting continued, and Owens' margin narrowed, making it still seem possible as of 12 November, that Hoffman actually may have won!
Still-later note: Owens did eventually officially win.

Conclusions

In both these races, it is clear that harm was caused by not having score voting, and this injury to democracy would not have been healed by instant runoff voting (IRV).

In the NJ governor race, the truth revealed by the score-voting style poll was that all three candidates enjoyed approximately equal popular support. The plurality voting system's "must vote for one of the top two" strategic dynamic, though (interacting with media advice not to "waste your vote"), unjustly deprived Daggett of any chance and undemocratically left him far behind. However, the score-voting and approval-style polls both indicated Daggett was the most-probable true-winner (as well as probably the "Condorcet winner" who would have defeated either rival head-to-head). This same huge undemocratic distortion also would have happened with instant (or non-instant) runoff voting.

In the NY-23 congress race, the same perceived strategic dynamic led the Republican Party cynically to backstab its own handpicked candidate Scozzafava, who (despite probably being the Condorcet "beats all" winner) then plunged to an unjustly-vast defeat. The situation was so insane that she actually felt it best to drop out of the race to "aid" her party and/or career. The truth, as revealed by the score-style poll, is that all three should have been close. Ironically, it might have been the case that this strategic perception by Republican leaders actually was wrong and by pressuring Scozzafava to withdraw they actually accomplished the opposite of their goal. With score voting, there is no "vote splitting" and there would have been no perceived need (wrong or not) for the backstab; e.g. those Republicans could simply have supported their own candidate Scozzafava while recommending Hoffman as second choice. Note also that Scozzafava would have been eliminated with "instant runoff voting" despite being Condorcet beats-all winner, illustrating IRV's dreaded center squeeze pathology which causes IRV to favor extremists and act against centrists.

Return to main page