We’ll be releasing our analysis of the Seat 4 race later this week. In this posting, we are looking at the August 11th run-off special election between Loretta Smith and Dan Ryan.
In May, Smith placed slightly ahead of Ryan in the 16-candidate primary – winning 18.8% of the vote to his 16.6%. In the August runoff, Ryan squeaked out a victory. We looked into some of the results to try to understand the factors leading to this outcome.
But it’s clear from our analysis that this race was all about turnout, but what really mattered is differential turnout.
The most important factor in Ryan’s victory is his performance in high-turnout precincts.
Let’s start by simply averaging the candidate results across precincts. By this measure, Smith should have won the race. Smith’s share of the vote averaged 50.5% across Portland’s precincts.
But Dan Ryan showed a clear advantage in the higher turnout precincts. We’ve plotted below the turnout in each precinct against Ryan’s share of the vote (the dot size corresponds to the number of registered voters in each precinct). There’s a very strong relationship (r = .89) indicating that precincts with higher turnout favored Ryan.
Of course, this is not just a story of turnout, it’s also a story about what kinds of voters were more likely to cast a ballot in an August special election. Turnout (countywide) in the May primary countywide was 51.55% (209, 031 voted in the position 2 race) and turnout (city wide) in the August runoff was 39.76% (173, 882 ballots voted in the position 2 race). This is still impressive for an August runoff (yay vote by mail!), but a substantial drop off from May.
We know that there are a number of standard demographic factors that are correlated with turnout–income, age, and education, So if one candidate has a stronger appeal more to higher income, older, and better educated voters, they will perform better in precincts where these traits are common, and therefore their vote share will be correlated with higher turnout. On the other hand, a correlation between vote share and turnout could be the result of a candidate that energizes their voters and gets them to turn out at higher rates than an opponent’s.
Can we determine if one of these situations is driving Ryan’s performance in these high-turnout precincts?
It is hard to say precisely with available data (and avoid the ecological fallacy) but we can make some educated guesses. Using geocoded voter information and data from the American Community Survey, we have constructed estimates of quantities such as median income, education, age structure, and more (to learn about how this was done, see our methodology post.
The graphs below show that Ryan’s vote share was not strongly related to the average age in precincts, but was very strongly related to average levels of education and income. This suggests that Ryan was always going to do well in the high turnout precincts, but it does doesn’t to rule out the idea that he also energized his supports
Ryan’s victory also had a substantial geographic struture, as some have already noted. Ryan dominated the whiter, wealthier, more educated neighborhoods in inner Portland, particularly highly residential neighborhoods like Eastmoreland, Laurelhurst, and Ladd’s Addition. Smith performed better in North Portland and was the overwhelming favorite in the Easternmost precincts.
We can’t directly assess the cause of the relationship we observe between Ryan’s vote share and precinct turnout. However, we can look to past election to get a sense of how turnout typically varies between precincts – giving us some points of comparison.
Remember: the first explanation for Ryan’s turnout advantage is that he did better in the kinds of places that always have high turnout. If that’s true, we’d expect that Ryan’s August 2020 vote share has a similarly strong relationship to turnout in every recent election. If Ryan succeeded due to a particular ability to turn out his voters, then we’d expect to see that the association between turnout and support for Ryan is strongest in the August 2020 special election.
We can examine this question among a recent group of elections: the November 2016 General Election, a special local election in November 2019, and the May 2020 Primary for Commissioner Eudaly’s seat. What we find is straightforward. The relationship between turnout and support for Ryan is almost identical across elections. In other words, Ryan performed well in precincts that would likely have had high turnout no matter who was running.
This is all preliminary – we only chose three elections to compare to and there are many factors we have not attempted to measure. Nonetheless, it’s clear that in the August run-off, turnout really did matter. It’s true that Loretta Smith could have won had she been able to mobilize voters in the areas of Portland where most of her supporters lived. However, what this analysis makes clear is that the turnout patterns that propelled Ryan to victory were not unique to this election – they are impressively consistent.
It’s been an exciting year in Portland city politics. Three City Council seats were being on the ballot in May, and two of the most hotly contested were Seat 2, the replacement election for the late Nick Fish, and Seat 4, with current Commissioner Chloe Eudaly facing off with two high-profile challengers.
We’ll be releasing our analysis of the Seat 4 race later this week. In this posting, we are looking at the August 11th run-off special election between Loretta Smith and Dan Ryan.
In May, Smith placed slightly ahead of Ryan in the 16-candidate primary – winning 18.8% of the vote to his 16.6%. In the August runoff, Ryan squeaked out a victory. We looked into some of the results to try to understand the factors leading to this outcome.
One of the most overused cliches in elections is “it’s all about turnout.”
But it’s clear from our analysis that this race was all about turnout, but what really mattered is differential turnout.
The most important factor in Ryan’s victory is his performance in high-turnout precincts.
Let’s start by simply averaging the candidate results across precincts. By this measure, Smith should have won the race. Smith’s share of the vote averaged 50.5% across Portland’s precincts.
But Dan Ryan showed a clear advantage in the higher turnout precincts. We’ve plotted below the turnout in each precinct against Ryan’s share of the vote (the dot size corresponds to the number of registered voters in each precinct). There’s a very strong relationship (r = .89) indicating that precincts with higher turnout favored Ryan.
Of course, this is not just a story of turnout, it’s also a story about what kinds of voters were more likely to cast a ballot in an August special election. Turnout (countywide) in the May primary countywide was 51.55% (209, 031 voted in the position 2 race) and turnout (city wide) in the August runoff was 39.76% (173, 882 ballots voted in the position 2 race). This is still impressive for an August runoff (yay vote by mail!), but a substantial drop off from May.
We know that there are a number of standard demographic factors that are correlated with turnout–income, age, and education, So if one candidate has a stronger appeal more to higher income, older, and better educated voters, they will perform better in precincts where these traits are common, and therefore their vote share will be correlated with higher turnout. On the other hand, a correlation between vote share and turnout could be the result of a candidate that energizes their voters and gets them to turn out at higher rates than an opponent’s.
Can we determine if one of these situations is driving Ryan’s performance in these high-turnout precincts?
It is hard to say precisely with available data (and avoid the ecological fallacy) but we can make some educated guesses. Using geocoded voter information and data from the American Community Survey, we have constructed estimates of quantities such as median income, education, age structure, and more (to learn about how this was done, see our methodology post.
The graphs below show that Ryan’s vote share was not strongly related to the average age in precincts, but was very strongly related to average levels of education and income. This suggests that Ryan was always going to do well in the high turnout precincts, but it does doesn’t to rule out the idea that he also energized his supports
Ryan’s victory also had a substantial geographic struture, as some have already noted. Ryan dominated the whiter, wealthier, more educated neighborhoods in inner Portland, particularly highly residential neighborhoods like Eastmoreland, Laurelhurst, and Ladd’s Addition. Smith performed better in North Portland and was the overwhelming favorite in the Easternmost precincts.
We can’t directly assess the cause of the relationship we observe between Ryan’s vote share and precinct turnout. However, we can look to past election to get a sense of how turnout typically varies between precincts – giving us some points of comparison.
Remember: the first explanation for Ryan’s turnout advantage is that he did better in the kinds of places that always have high turnout. If that’s true, we’d expect that Ryan’s August 2020 vote share has a similarly strong relationship to turnout in every recent election. If Ryan succeeded due to a particular ability to turn out his voters, then we’d expect to see that the association between turnout and support for Ryan is strongest in the August 2020 special election.
We can examine this question among a recent group of elections: the November 2016 General Election, a special local election in November 2019, and the May 2020 Primary for Commissioner Eudaly’s seat. What we find is straightforward. The relationship between turnout and support for Ryan is almost identical across elections. In other words, Ryan performed well in precincts that would likely have had high turnout no matter who was running.
This is all preliminary – we only chose three elections to compare to and there are many factors we have not attempted to measure. Nonetheless, it’s clear that in the August run-off, turnout really did matter. It’s true that Loretta Smith could have won had she been able to mobilize voters in the areas of Portland where most of her supporters lived. However, what this analysis makes clear is that the turnout patterns that propelled Ryan to victory were not unique to this election – they are impressively consistent.