Demographics

Maine’s Second Congressional District is the largest district east of the Mississippi River, and encompasses about 80% of the state’s total land, occupying the northern area. While this district does not include Maine’s largest city of Portland, it does include Bangor, Lewiston, and Auburn. It is the second-most rural district in the U.S. and the district with the second highest proportion of non-Hispanic White residents, as 72% of the population within the district live in rural areas and 94% of residents identify as non-Hispanic Whites. As of the 2020 Census, this district represents about 680,000 individuals, with a population that is 49.3% male and 50.7% female. The medium household income in this district is $51,585, making it poorer than most other New England districts. While the high school graduation rate is relatively high at 92.1%, the college graduation rate in the district falls on the lower end at around 25.3%. Another interesting characteristic about Maine’s 2nd district is that it is the only district in all of New England that voted for Donald Trump in 2020, and only one of eight districts that voted for Trump in 2020 while being held by a Democrat. Ballotpedia also recognizes it as a battleground district in the 2016 election when incumbent Republican Bruce Poliquin defeated former State Senator Emily Ann Cain. New England is known as a rather liberal area, often going blue in most elections, but the makeup of this district in terms of its lack of diversity and rural population set it apart from other New England areas.

In Presidential elections prior to 1992, Maine as a state voted for Republican candidates. From 1992 until Trump in 2016, Maine voted for Democratic candidates every presidential election. In terms of the District’s Congressional history, its representatives have shifted back and forth between Republican and Democratic for much of recent history. Jared Golden is the current representative, after having first been elected to the position in 2019. His Republican opponent in 2022’s race, Bruce Poliquin, was the district’s representative from 2015 until 2019, and prior to him, a Democratic representative held the position from 2003-2015. The district does not seem to have a strong voting pattern across Congresses.

It is worth noting that in the November 8th, 2016 election, Maine voted to approve ranked-choice voting. Since then, they have conducted their statewide elections for governor, state legislature, and Congress in this manner.

Forecasts

At the beginning of this class in September, the forecasts for Maine’s 2nd Congressional District varied widely across sources. On September 9th, FiveThirtyEight forecasted Jared Golden (D) winning with a 63% chance and predicted a vote share of 49.6% Democrat to 46.9% Republican. Despite still being 2 months out from the election, this predicted vote share was fairly accurate, overestimating the Democratic share by 1.4% and overestimating the Republican share by 2.0%. One month prior to the election, on October 7th, FiveThirtyEight still had Jared Golden with a 63% chance of winning, although their predicted vote shares changed a small amount, with their models predicting 49.4% Democrats and 46.6% Republican, getting slightly closer to actual vote shares seen in the election. Right before the election, on November 6th, FiveThirtyEight’s models still had Jared Golden favored with a 62% chance of winning. Again their vote shares shifted slightly, to 49.3% Democrat and 47.2% Republican. While the Democratic vote prediction got 0.1% closer to the actual vote share, the Republican vote share just before the election was farther off than it had been a month previously. Some of this likely has to do with the share of third party votes being higher than expected. Tiffany Bond, the Independent running against Golden and Poliquin, received 6.9% of the vote, which is higher than the 3.5% predicted by FiveThirtyEight in the days before the election. From September up until the election, FiveThirtyEight’s model had the district coded as “Lean Democrat”, which coincides with the results that we ended up seeing. Overall, FiveThrityEight’s model was a solid predictor for the district, acknowledging the closeness of the race by having it coded as a “Lean” state, and coming pretty close with the winning party’s vote share. The one place where they deviated from observed results was in the vote share of the third party, which may be underestimated in polling given the small population that do vote for this candidate.

Now turning to the other major source for House Election Forecasts: The Economist. As of September 12th, 2022, The Economist’s forecast listed Maine’s 2nd Congressional District as “Very Likely Republican”, reporting Bruce Poliquin’s chance of winning at 90%. At this time they also predicted both the Democrats and Republicans of winning 48-52% of the vote share nationwide, which is not very meaningful and much less precise than FiveThrityEight’s predictions. For the district’s vote share, The Economist’s model predicted 54.6% Republican and 45.4% Democrat. This is a two-party vote share so we cannot directly compare it to the results that we saw, especially since nearly 7% of votes went to the third party. However, this Republican vote share estimate is off by nearly 10%. Even for this point in the election – 2 months before it – this is a large margin to be off by. By one month prior to the election, on October 7th, The Economist changed their prediction to “Likely” Republican for this district. At this time their model predicted Bruce Poliquin had a 66% chance to win - similar to the 62% likelihood predicted by FiveThirtyEight, but with the opposite candidate. This decline in certainty of Poliquin winning over such a short period of time, a 24 percentage-point decline, is very large. I am interested in any changes in Poliquin’s campaign that led to this sharp decline in chance of winning by The Economist’s model, given that a month prior that had it as a “very likely” seat for Republicans. Looking at their predictions here, their model’s expectations about the vote share predicted 51.9% Republican and 48.1% Democrat. For the Democrats, this is only off by 0.1 percentage points of the actual vote share, however, again, we must acknowledge that this is the two-party vote share so this estimate is not as accurate as it may initially seem. They also still have the prediction showing Republicans winning the majority of the vote share, which is incorrect, and way overestimates their vote share. Right before the election on November 7th, The Economist again changed the classification of the district, changing it to “uncertain” and predicting that Jared Golden had a 56% chance of winning, changing the candidate in addition to the likelihood classification. While we saw stability in FiveThirtyEight’s model over the months leading up the election, The Economist’s forecasting model changed a lot over the same time frame. I would be interested to see how their models differed and hopefully identify where this drastic change in predictions stems from. The day before the election, they predict the two-party vote share to be 50.4% Democrat and 49.6% Republican. While here the direction of the winning party is better represented, these percentages are again fairly far off. It wasn’t until November 3rd that the model showed the Democratic two-party vote share overtaking the Republicans. While we know that the Democratic candidate did end up winning, the race was never represented as being this much of a toss-up by FiveThrityEight. They had predicted the Democratic candidate winning for months, making it interesting to me that this was not the prediction by the Economist until days before the election. Again, I would love to see how their models differ, as I cannot imagine things like polling between the two would vary so dramatically, and in the case of The Economist, change so quickly.

Election Facts

According to Politico, because neither Golden nor Poliquin received 50% of the vote, the race went to an “instant run-off”. Given Maine’s ranked-choice voting, residents do not need to go back to the ballots in a run-off – rather, their second choice vote will be counted as the winner. In the first round of voting, Jared Golden received 151,570 votes, Bruce Poliquin received 141,119 votes, and Tiffany Bond received 21,565 votes. After the instant run-off, Jared Golden received 53.1% of the votes and Bruce Poliquin received 46.9%. Given this run-off scenario, this election was definitely much more of a toss-up than The Economist had predicted it to be in September.

While this election was a big year for redistricting in some states, with attempted Gerrymandering by parties in certain areas, this was not much of an issue with this district. According to Politico, the makeup of the map in terms of who the population supports remains the same.

This election was predicted to be a “Red Wave”, with Republicans highly favored to take over the House of Representatives and do so easily. As the election drew closer, however, the expected gap between Republican seats and Democratic seats narrowed, and the election ended up being much closer than experts and followers had expected a few months before. These predictions – speficially The Economist forecast – reflect this nationwide trend, to a slight extreme however. The 2018 election between Poliquin and Golden also entered a rank choice run-off in which Golden came out victorious. However, Poliquin tried his hand at contesting the ranked-choice tabulation in a federal court case (Maine Public).

Campaigning

Between the two candidates, over $25 million was spent on campaigning. According to the Sun Journal, this budget was spent on “more than 25,000 television commercials along with reams of mailers, radio spots and social media advertising”. It was also the eighth most expensive House race in the entire country according to non-profit campaign finance tracking organization OpenSecrets, with around $7 million in total candidate spending along with $21.8 million in total outside spending.

Poliquin’s campaign focused on issues such as crime, border security, and inflation. By using inflation as an argument, he was running what Lynn Vaverick (2009) would call a clarifying campaign: he is the challenging party and the U.S. is in a relatively bad economy overall. According to Vaverick, this is the correct type of campaign that he should be running in that it will “prevent him or her from increasing voters’ uncertainty about your relationship to current economic situation” (Vaverick 2009). On his personal website, Poliquin has a page dedicated to the issues that he stresses. The first thing that comes up on this page is a large quote from him stating, “We must lower the cost of home heating, gas, and groceries. I will bring Maine common sense back to Washington to get inflation under control,” followed by “Fighting Inflation” at the top of his list of important issues. In addition to inflation, he also has “Cutting Taxes” as one of his campaign’s eight main issues, which would also fall under the umbrella of economic issues.

Poliquin’s website also stresses the fact that he is a working-class man and a single-father after his wife tragically passed away. In terms of his background that makes him a strong candidate, he worked at Bath Iron Works “creating hundreds of jobs, and [managing] pensions for workers” before tackling the role as Maine’s Treasurer. His website also shares his success in cutting taxes in Maine through his “honesty, integrity, and fiscal responsibility”. After his role as Treasurer, he served in Congress after the 2014 election where he “became known for his tenacity when fighting for Maine”. He is also a strong supporter of the 2nd Amendment, which his website lists as another one of his core campaign issues.

On the other side of the ballot, we have incumbent Democrat Jared Golden. On his campagin website, he lists his top issues as campaign finance reform, expanding healthcare, and supporting seniors, among others. As a democrat in a difficult economy, he should be running what Vaverick would call an “insurgent campaign” in which the campaign does not look at the economy, but instead focuses on an issue that benefits from public opinion. This issue for Golden might be the campaign finance reform - an issue that does not seem like a thing most voters highly value but can cast his opponent in an unfavorable light. He wants to get “big money out of politics and clean up government corruption” by focusing on getting rid of lobbyists and protecting/expanding voting rights. His campaign manages to avoid discussion of the current economy, and the issues that he does focus on align with Vaverick’s insurgent campaign in that she stresses the importance of the candidate’s main issue being one in which their opponent “is committed to or constrained by previously taken unpopular opinion” (Vaverick 2009). According to nonpartisan policy institute and advocacy organization Center for American Progress Action Fund, Bruce Poliquin received donations to his campaign from many lobbyists and corporate PACs, both of which Golden has spoken against outwardly and refused donations from. By making this issue the forefront of his campaign, Golden likely aimed to cast Poliquin as corrupted.

Golden’s website also casts him as a hometown hero-esq candidate. Born and raised in Maine’s second district, Golden enlisted in the Marine Corps following the September 11th attacks and completed combat tours in Iraq and Afghanistan. His website also casts him as a hard-working man, saying that when he returned home he had three jobs, working nights and weekends. This is similar to Poliquin’s biography describing him as a working-class man.

Overall, it seems that both candidates ran campaigns that followed Vaverick’s “right” campaign strategy. This election was hard fought on both sides and was very much a toss-up, having to go into a run-off election via ranked-choice voting.

Speculations

In looking at various site’s predictions for this election, it seems that many had predicted Golden would win by a small margin, but many did still list the district as a toss-up. Politico, FiveThirtyEight, FairVote, and others made these similar forecasts for the election, and acknowledged the historical voting data citing the 2018 election between the same two candidates. In 2018, these two candidates faced off in a toss-up election that went into a ranked-choice voting runoff in which Golden came out on top, very similar to the results of this year. In terms of where The Economist’s predictions got this election wrong, forecasting a “Very Likely Republican” win in September, it is hard to understand where the data they were using came from. If they had looked at the historical data alone, it should have been noted that this election would likely mirror that of 2018 and be very close. Jared Golden is also the incumbent, and unless something with his campaign or his tenure in Congress went very awry (which it did not), it would be hard to imagine such a strong challenger likelihood. Initially when I saw these discrepancies, I though that it could be possible for their model to not include ranked-choice voting, and thus many votes would be unaccounted for in the case of a runoff. However, this cannot explain the deviation because ranked-choice voting is only enacted in a runoff, when no candidate reaches 50% of the vote share but which would thus acknowledge the election being very close to begin with. Looking at the timeline of their predictions, it appears there were sharp changes in the forecasted vote shares and likelihoods of winning on September 29th, 2022 and again on October 21st, 2022. I tried to find some news that might explain these sharp changes in predictions – as on each day the likelihood of winning changed by 20% for each party – but could not find any news that would cause this. The University of New Hampshire ran the Pine Tree State Poll with results coning out on September 28th, however these polls showed Golden with 44% support, Poliquin with 33% support, and 13% of voters undecided. These results do not support the 70% likelihood of winning for Bruce Poliquin that The Economist had at this same time.

After further research, The Economist has a page about their models and forecasts that details issues they found when the probability of winning did not change with the updated news - similar to what we saw in Maine’s second district. On here they list Pennsylvania as an example, where John Fetterman’s likelihood of winning was listed at 91% despite his lead in the polls being only about 5 points ahead of his competitor. This is very similar to what I saw with Maine, and therefore it was an issue with the model rather than an exact issue or shock that caused these estimates. The Economist paused their forecasting on October 16th, 2022 and related multiple issues they found with their models that caused these inaccurate predication. They started off detailing how their “polling averages turned out to be insufficiently prepared for this year’s polling landscape” and blamed two pollsters in particular for the inaccuracies, Center Street PAC and Echelon Insights. In addition to these pollsters’ inaccuracies, The Economist also blamed a “decline in the number of surveys of individual districts in the House of Representatives”. Finally, they adjusted the design of their model to adjust for other possible inaccuracies and had all their forecasts adjust – which is probably what we saw around October 21st when the predictions of the race became much closer and forecasted the toss-up that we saw on all other websites.

Aside from The Economist, most other websites were very accurate in their predictions. They forecasted toss-up races with a very slight favor towards Democrats, which is exactly what we ended up seeing. Maine’s ranked-choice voting allowed for the run-off election to be instantaneous, and did not require voters to go back to the polls which could have changed the expected results. Overall, most expert predictions and models were right on target when it came to Maine’s 2nd Congressional District, as this election closely mirrored that of 2018.

Sources: https://www.sunjournal.com/2022/11/08/waiting-for-results-in-maines-closely-fought-2nd-district-race/ https://ballotpedia.org/Maine%27s_2nd_Congressional_District_election,_2022 https://ballotpedia.org/United_States_House_of_Representatives_elections_in_Maine,_2022 https://www.economist.com/interactive/us-midterms-2022/forecast/house/how-this-works https://fairvote.org/results-and-analysis-from-maines-2nd-district/ https://en.wikipedia.org/wiki/Maine%27s_2nd_congressional_district https://www.mainepublic.org/politics/2022-11-09/jared-golden-leads-bruce-poliquin-in-2nd-district-race-likely-headed-to-ranked-choice-runoff https://www.mainepublic.org/politics/2022-11-16/jared-golden-prevails-over-bruce-poliquin-after-ranked-choice-runoff