In our last post, we looked at ballot return methods among Multnomah County, Oregon voters in the November 2020 election. As the most populous county in the first state to adopt vote by mail in 2000, Multnomah is an interesting case to explore voter use and adaptation over two decades of voting by mail. November 2020 is also interesting because the state began to pay for return postage as of August 2019, leading one well-known advocacy group to describe Oregon as a state where “every mail box is a drop box.” But is this actually how voters responded?
In the November 2020 general election, over half of county voters returned their ballots in person at drop boxes, and less than a third were returned through the mail. At the aggregate level, more voters used drop boxes in precincts that were physically closer to drop box locations. After a decline in drop box usage in the earlier part of 2020, county voters were using drop sites at historically high rates in November. They also returned their ballots almost a week earlier than in previous years, elections when half of drop box voters typically cast their votes in the last two days of voting.
Unsurprisingly, a global pandemic and hyper-polarized rhetoric about voting by mail may have changed how voters think about returning their ballots in 2020. In fact, in an October 2020 survey of Oregonians conducted by EVIC as part of the Oregon Election Integrity project, we discovered that 60% of our respondents said they planned to use a drop box in November. Convenience and security were the main reasons respondents gave for using a drop box.
The Unique November 2020 Environment
Elections in 2020 came with their own set of challenges, for both LEOs (as we’ve written about previously) and voters. This year, many individual voters had to adjust the way that they planned to vote. Even in Oregon, where we didn’t have to deal with administering COVID-safe physical polling places, the effects of the pandemic on many people’s daily habits impacted which ballot return options were convenient for voters. For example, somebody who previously worked near a ballot drop box might have lost their job or started working from home, so their daily life wouldn’t take them close to the site where they usually returned their ballot.
The November election also came with a unique level of national salience on the reliability of the postal service, and several pleas for voters to drop their ballot off in person rather than returning it through the mail. The President himself indicated that he was trying to withhold USPS funding in order to restrict voting by mail. In the wake of several operational changes implemented by Postmaster General Louis DeJoy, the Postal Service was experiencing slowdowns in first-class mail delivery across the country leading up to the election. On the positive side, however, Oregonians did not have to pay for postage to return ballots this year.
In previous years, the Multnomah County Library had a drop box at each location, though several of these were inside the building at the circulation desk. In March, however, library locations closed due to COVID-19 and later only reopened for appointment-only pickups of materials at the door. In response, Multnomah County Elections allowed ballot returns through library book drops and certified all library staff as election workers. This was particularly relevant in the May primary election, when libraries were still fully closed to the public and could not support any indoor drop boxes.
Is Multnomah County Special?
Of course, changes in mail balloting weren’t unique to Multnomah County. Jurisdictions across the country saw expanded absentee ballot usage, making Oregon one of the few places where mail ballot usage didn’t increase. Voters had to update their habits, and sought out alternative voting methods to deal with COVID safety and changes to their typical polling places.
To compare Multnomah County to Oregon and the nation, we rely on:
Administrative data from the Oregon Secretary of State, detailing when ballots were received by each county during the November 2020 election
Overall, we see below that Multnomah County is not that different from Oregon as a whole; there are only minor differences between the two on drop box usage and timing. It is interesting, however, that the slight difference has the entire state using drop boxes more than Multnomah County. You might think that voters in such an urban county, compared to the state at large, would be more likely to have a dropbox nearby and vote Democratic – two factors which are positively correlated with drop box usage in general, as we discussed last time.
However, the county and state are very different from the rest of the country in terms of how they return their by-mail ballot. Nationally, voters are much less likely to return their absentee/mail ballots in person, and much more likely to return their ballot earlier in the voting period.
There are a few reasons that this might be the case:
While Oregon’s laws about voting by mail have been relatively constant for the last few election cycles, other states were rapidly changing their laws and administration to adjust mail voting to the COVID-19 pandemic. Voters might have chosen to return their ballot early and in person for an added sense of security in ensuring their vote was counted.
Some states have earlier deadlines for mail ballot returns, where voters who wait until later in the voting period to return their ballot might have had to show up on election day and wait in line to give their mail ballot to a poll worker.
While Oregon has multiple dropboxes in every county, some states did not set up additional drop sites for voters to return their absentee ballots by hand.
Of course, these are not exact comparisons because of the disparate data sources used for each jurisdiction. Notably, the Multnomah County and Oregon estimates rely on administrative data about when the ballot was received by an elections office, whereas the national survey instead asks respondents when they returned their ballot to an office. Delays in mail time and ballot processing mean that the numbers from the county and state are likely closer to a comparable national estimate, as (by definition) fewer ballots nationwide had been received by elections offices than had been sent by voters as of a specific date. Similarly, for survey questions measuring ballot return method there is a difference in response wording that we’ve attempted to handle appropriately, and these responses contain various sources of survey error that is not present in the administrative data from Multnomah County. We do not know of any reason to expect a response bias relating to method of returning mail ballots, though there’s not exactly plentiful research on this specific topic to turn to for guidance.
Are These Patterns New?
So, we saw that Multnomah County voters mostly returned their ballots in person this November. How does that compare to previous elections? Since 2016, Multnomah County has conducted 13 elections, and we included 12 of these in this analysis (the remaining one covered only 3,000 of the over 500,000 registered voters in the county). We’ve also labeled three relevant non-election dates: the opening of the Voting Center Express in Gresham, the implementation of paid postage for ballots, and the start of the statewide stay-at-home order (to denote an effective start of the COVID-19 pandemic).
Additionally, we split up the “Counter/Office” source here, re-assigning the “Counter/Office” subcategory to “All Other Methods” and the “Elections Bldg” subcategory to “Drop Box/Site”. This makes for a more consistent comparison across time, because the “Elections Bldg” subcategory (ballots cast in the 24-hour drop boxes outside of the County Elections Office in SE Portland) doesn’t appear until 2019 and these ballots were included with the “Drop Box/Site” source before this change. With this recode made to the default categorization scheme, we can compare the November 2020 election to the last four years and see if any patterns emerge.
Interestingly, November 2020 looks to be a return to the normal pattern of drop box usage, after a decrease in the earlier elections this year. From 2016 to 2019, drop box and mail ballots traded off between 40% and 60% of all ballots cast. In the May primary and August special election, ballot returns by mail were up and drop box usage was way down. In November, however, drop box use was back up even higher than historically “normal” rates while mail returns tanked.
What gave rise to these patterns? It’s hard to separate the competing causes. Two credible agents both came into effect in the same gap between November 2019 and May 2020: postage-paid ballots (which likely increased mail returns), and the COVID-19 pandemic’s impact on commuting and other travel (which likely decreased drop box usage). Would these phenomena have had the same impact by November 2020? It’s hard to tell, as the salience of mail ballots and the USPS likely contributed to an increase in voters using drop boxes for the general election.
As an aside for the true election geeks, before 2020 all other methods combined made up less than 4% of ballots in any election (and typically less than 2%). In 2020, however, this category increased to around 6% of all ballots cast – even without including the “Elections Bldg” subcategory. It’s not immediately clear what could have led to this small (but noticeable) increase, but one possibility might be that more people wanted to cast their ballot “in person” or receive direct help from elections staff. Looking more granularly at the data, we can see that the “Counter/Office” subcategory was consistently around 3% of all ballots cast in 2020, where in previous years it hadn’t risen above 2%.
Multnomah County voters heeded this advice. They returned ballots much earlier than in previous elections (and at “in-person” rates similar to non-COVID elections). We’ve typically seen a large spike of ballots cast on Election Day, almost all at drop boxes – in previous cycles, about 25% of all ballots were cast on the last day. This year, however, there was a huge spike of ballot returns on the first Monday after most voters had received their ballots (15 days before Election Day). Accordingly, there was less of a spike on Election Day this year because more voters had already returned their ballots – only about 10% of voters waited until the last day to vote. Mail ballots are typically cast earlier and trail off by Election Day, which is mainly due to the reception deadline as described earlier.
We can quantify these differences by looking at the median ballot return date for each method across elections, the date by which half of voters had already voted and the other half had not yet returned their ballots.
This data really captures how early ballots were cast this year. The median date of voting in November 2020 was two Fridays before Election Day, almost a week earlier than in the May primary (the Friday before Election Day). Across all methods, Multnomah County voters cast their ballots earlier in November 2020 than in recent elections. Historically, over half of all drop box ballots were returned in the last two days (if not on the last day). This year, half of drop box ballots were received over a week before Election Day. Mail ballots were similarly returned early (though it was a less drastic shift), diverging from the previous pattern of general and primary election mail returns occuring later than mail returns in special elections. Voters may have been extra concerned about their ballot arriving in time through the USPS, and chosen to drop their ballot off in person after an earlier perceived cutoff. The time difference between mail and drop box returns continued in 2020, though the gap between the two was partially closed due to how much earlier drop box ballots were returned.
We also found that the relationship between distance and drop box usage was relatively unchanged from previous elections, even when those previous elections saw notably different political messaging around voting by mail, the lack of a global pandemic, and lower overall rates of drop box usage. Political science research conducted in Washington State has shown a significant relationship between the distance to a polling place, drop box, and turnout, but no one to our knowledge has looked at how this relationship changes over time and in response to an intervention like what happened in 2020.
The same precincts, however, were likely to have high or low residuals year after year – that is, a given precinct might have higher drop box usage across several elections then the purely distance-based model would predict. This further signals that something more than just distance is affecting drop box usage rates between precincts, but we were unable to isolate the effects of any known factor (such as partisanship or demographics) because these are so strongly related to where voters live.
Final Observations
Another question that interested us was whether the usage of library returns changed in 2020 as a result of the shift to returning ballots through the book drop instead of a dedicated ballot drop box inside. This could have seen a decrease due to changes in movement patterns among voters, or increased due to the increased availability of library drop sites – since book drops were used instead of interior boxes, these effectively became 24-hour drop sites. There was not a discernible pattern in library returns as a percentage of all drop box returns, as library drop site usage was down in May but up at relatively high levels in August and November this year.
One additional phenomenon is that military and overseas voters (covered under UOCAVA) were much more likely to return their ballots by email in November 2020. Over half a percent of all voters emailed their ballots back (2,611), far higher than rates in any other election since 2016. Perhaps this also stems from a lack of public confidence in the postal service delivering ballots on time, particularly given that international mail has extra delays in reaching domestic elections offices.
In Multnomah County, like in many places, the November 2020 election continues to surprise. As the COVID-19 pandemic continues and after it has passed, election science will have plenty of work to do in identifying which patterns will continue, be broken, or return to more “normal” expressions. In Multnomah County, the next election to watch for is on May 21, 2021. Even though Oregon is unlikely to significantly change its voting laws before then (as opposed to some other states), it’ll certainly be interesting to see if the pandemic and political messaging about voting by mail affect how Oregonians choose to return their vote-at-home ballots moving forward.
In our last post, we looked at ballot return methods among Multnomah County, Oregon voters in the November 2020 election. As the most populous county in the first state to adopt vote by mail in 2000, Multnomah is an interesting case to explore voter use and adaptation over two decades of voting by mail. November 2020 is also interesting because the state began to pay for return postage as of August 2019, leading one well-known advocacy group to describe Oregon as a state where “every mail box is a drop box.” But is this actually how voters responded?
In the November 2020 general election, over half of county voters returned their ballots in person at drop boxes, and less than a third were returned through the mail. At the aggregate level, more voters used drop boxes in precincts that were physically closer to drop box locations. After a decline in drop box usage in the earlier part of 2020, county voters were using drop sites at historically high rates in November. They also returned their ballots almost a week earlier than in previous years, elections when half of drop box voters typically cast their votes in the last two days of voting.
Unsurprisingly, a global pandemic and hyper-polarized rhetoric about voting by mail may have changed how voters think about returning their ballots in 2020. In fact, in an October 2020 survey of Oregonians conducted by EVIC as part of the Oregon Election Integrity project, we discovered that 60% of our respondents said they planned to use a drop box in November. Convenience and security were the main reasons respondents gave for using a drop box.
The Unique November 2020 Environment
Elections in 2020 came with their own set of challenges, for both LEOs (as we’ve written about previously) and voters. This year, many individual voters had to adjust the way that they planned to vote. Even in Oregon, where we didn’t have to deal with administering COVID-safe physical polling places, the effects of the pandemic on many people’s daily habits impacted which ballot return options were convenient for voters. For example, somebody who previously worked near a ballot drop box might have lost their job or started working from home, so their daily life wouldn’t take them close to the site where they usually returned their ballot.
The November election also came with a unique level of national salience on the reliability of the postal service, and several pleas for voters to drop their ballot off in person rather than returning it through the mail. The President himself indicated that he was trying to withhold USPS funding in order to restrict voting by mail. In the wake of several operational changes implemented by Postmaster General Louis DeJoy, the Postal Service was experiencing slowdowns in first-class mail delivery across the country leading up to the election. On the positive side, however, Oregonians did not have to pay for postage to return ballots this year.
In previous years, the Multnomah County Library had a drop box at each location, though several of these were inside the building at the circulation desk. In March, however, library locations closed due to COVID-19 and later only reopened for appointment-only pickups of materials at the door. In response, Multnomah County Elections allowed ballot returns through library book drops and certified all library staff as election workers. This was particularly relevant in the May primary election, when libraries were still fully closed to the public and could not support any indoor drop boxes.
Is Multnomah County Special?
Of course, changes in mail balloting weren’t unique to Multnomah County. Jurisdictions across the country saw expanded absentee ballot usage, making Oregon one of the few places where mail ballot usage didn’t increase. Voters had to update their habits, and sought out alternative voting methods to deal with COVID safety and changes to their typical polling places.
To compare Multnomah County to Oregon and the nation, we rely on:
Overall, we see below that Multnomah County is not that different from Oregon as a whole; there are only minor differences between the two on drop box usage and timing. It is interesting, however, that the slight difference has the entire state using drop boxes more than Multnomah County. You might think that voters in such an urban county, compared to the state at large, would be more likely to have a dropbox nearby and vote Democratic – two factors which are positively correlated with drop box usage in general, as we discussed last time.
However, the county and state are very different from the rest of the country in terms of how they return their by-mail ballot. Nationally, voters are much less likely to return their absentee/mail ballots in person, and much more likely to return their ballot earlier in the voting period.
There are a few reasons that this might be the case:
Of course, these are not exact comparisons because of the disparate data sources used for each jurisdiction. Notably, the Multnomah County and Oregon estimates rely on administrative data about when the ballot was received by an elections office, whereas the national survey instead asks respondents when they returned their ballot to an office. Delays in mail time and ballot processing mean that the numbers from the county and state are likely closer to a comparable national estimate, as (by definition) fewer ballots nationwide had been received by elections offices than had been sent by voters as of a specific date. Similarly, for survey questions measuring ballot return method there is a difference in response wording that we’ve attempted to handle appropriately, and these responses contain various sources of survey error that is not present in the administrative data from Multnomah County. We do not know of any reason to expect a response bias relating to method of returning mail ballots, though there’s not exactly plentiful research on this specific topic to turn to for guidance.
Are These Patterns New?
So, we saw that Multnomah County voters mostly returned their ballots in person this November. How does that compare to previous elections? Since 2016, Multnomah County has conducted 13 elections, and we included 12 of these in this analysis (the remaining one covered only 3,000 of the over 500,000 registered voters in the county). We’ve also labeled three relevant non-election dates: the opening of the Voting Center Express in Gresham, the implementation of paid postage for ballots, and the start of the statewide stay-at-home order (to denote an effective start of the COVID-19 pandemic).
Additionally, we split up the “Counter/Office” source here, re-assigning the “Counter/Office” subcategory to “All Other Methods” and the “Elections Bldg” subcategory to “Drop Box/Site”. This makes for a more consistent comparison across time, because the “Elections Bldg” subcategory (ballots cast in the 24-hour drop boxes outside of the County Elections Office in SE Portland) doesn’t appear until 2019 and these ballots were included with the “Drop Box/Site” source before this change. With this recode made to the default categorization scheme, we can compare the November 2020 election to the last four years and see if any patterns emerge.
Interestingly, November 2020 looks to be a return to the normal pattern of drop box usage, after a decrease in the earlier elections this year. From 2016 to 2019, drop box and mail ballots traded off between 40% and 60% of all ballots cast. In the May primary and August special election, ballot returns by mail were up and drop box usage was way down. In November, however, drop box use was back up even higher than historically “normal” rates while mail returns tanked.
What gave rise to these patterns? It’s hard to separate the competing causes. Two credible agents both came into effect in the same gap between November 2019 and May 2020: postage-paid ballots (which likely increased mail returns), and the COVID-19 pandemic’s impact on commuting and other travel (which likely decreased drop box usage). Would these phenomena have had the same impact by November 2020? It’s hard to tell, as the salience of mail ballots and the USPS likely contributed to an increase in voters using drop boxes for the general election.
As an aside for the true election geeks, before 2020 all other methods combined made up less than 4% of ballots in any election (and typically less than 2%). In 2020, however, this category increased to around 6% of all ballots cast – even without including the “Elections Bldg” subcategory. It’s not immediately clear what could have led to this small (but noticeable) increase, but one possibility might be that more people wanted to cast their ballot “in person” or receive direct help from elections staff. Looking more granularly at the data, we can see that the “Counter/Office” subcategory was consistently around 3% of all ballots cast in 2020, where in previous years it hadn’t risen above 2%.
A Changing Timeline for Ballot Returns
In response to the challenges of administering elections during the pandemics, multiple commissions and advisory boards urged states and local jurisdictions to make it easier to cast a ballot prior to Election Day, and urged voters to cast ballots as soon as they felt comfortable doing in order to avoid problems in curing, validating, and counting ballots.
Multnomah County voters heeded this advice. They returned ballots much earlier than in previous elections (and at “in-person” rates similar to non-COVID elections). We’ve typically seen a large spike of ballots cast on Election Day, almost all at drop boxes – in previous cycles, about 25% of all ballots were cast on the last day. This year, however, there was a huge spike of ballot returns on the first Monday after most voters had received their ballots (15 days before Election Day). Accordingly, there was less of a spike on Election Day this year because more voters had already returned their ballots – only about 10% of voters waited until the last day to vote. Mail ballots are typically cast earlier and trail off by Election Day, which is mainly due to the reception deadline as described earlier.
We can quantify these differences by looking at the median ballot return date for each method across elections, the date by which half of voters had already voted and the other half had not yet returned their ballots.
This data really captures how early ballots were cast this year. The median date of voting in November 2020 was two Fridays before Election Day, almost a week earlier than in the May primary (the Friday before Election Day). Across all methods, Multnomah County voters cast their ballots earlier in November 2020 than in recent elections. Historically, over half of all drop box ballots were returned in the last two days (if not on the last day). This year, half of drop box ballots were received over a week before Election Day. Mail ballots were similarly returned early (though it was a less drastic shift), diverging from the previous pattern of general and primary election mail returns occuring later than mail returns in special elections. Voters may have been extra concerned about their ballot arriving in time through the USPS, and chosen to drop their ballot off in person after an earlier perceived cutoff. The time difference between mail and drop box returns continued in 2020, though the gap between the two was partially closed due to how much earlier drop box ballots were returned.
We also found that the relationship between distance and drop box usage was relatively unchanged from previous elections, even when those previous elections saw notably different political messaging around voting by mail, the lack of a global pandemic, and lower overall rates of drop box usage. Political science research conducted in Washington State has shown a significant relationship between the distance to a polling place, drop box, and turnout, but no one to our knowledge has looked at how this relationship changes over time and in response to an intervention like what happened in 2020.
The same precincts, however, were likely to have high or low residuals year after year – that is, a given precinct might have higher drop box usage across several elections then the purely distance-based model would predict. This further signals that something more than just distance is affecting drop box usage rates between precincts, but we were unable to isolate the effects of any known factor (such as partisanship or demographics) because these are so strongly related to where voters live.
Final Observations
Another question that interested us was whether the usage of library returns changed in 2020 as a result of the shift to returning ballots through the book drop instead of a dedicated ballot drop box inside. This could have seen a decrease due to changes in movement patterns among voters, or increased due to the increased availability of library drop sites – since book drops were used instead of interior boxes, these effectively became 24-hour drop sites. There was not a discernible pattern in library returns as a percentage of all drop box returns, as library drop site usage was down in May but up at relatively high levels in August and November this year.
One additional phenomenon is that military and overseas voters (covered under UOCAVA) were much more likely to return their ballots by email in November 2020. Over half a percent of all voters emailed their ballots back (2,611), far higher than rates in any other election since 2016. Perhaps this also stems from a lack of public confidence in the postal service delivering ballots on time, particularly given that international mail has extra delays in reaching domestic elections offices.
In Multnomah County, like in many places, the November 2020 election continues to surprise. As the COVID-19 pandemic continues and after it has passed, election science will have plenty of work to do in identifying which patterns will continue, be broken, or return to more “normal” expressions. In Multnomah County, the next election to watch for is on May 21, 2021. Even though Oregon is unlikely to significantly change its voting laws before then (as opposed to some other states), it’ll certainly be interesting to see if the pandemic and political messaging about voting by mail affect how Oregonians choose to return their vote-at-home ballots moving forward.
This is part two of a two-part series. Part one is available here.