What Happened in Michigan

This is the first in a multi-part series where I analyze the 2020 presidential election results. For those who have followed my coverage of the primaries, the conventions, and the anticipated swing states, this series is a long-time coming. For those who are new to my site, I hope you find this useful.

The first part will be looking at Michigan and what we can gather from the results there. If you haven’t already, I highly recommend you check out the pre-election analysis I did for Michigan, which identifies and breaks down the state’s solidly Democratic counties, its swing counties, and its Obama-Trump counties. I’ll reference key findings from this previous article as they become relevant, but I will not be doing a full recap.

In terms of structure, these articles will share some similarities to the pre-election articles, in that I’ll be outlining a big picture of the state as a whole and then breaking down more specific trends by subsection. Even so, there will be two main additions to this format. One is that I will be discussing the state exit poll results in close detail. Now I’m aware that exit polls are an imperfect measure of voting behavior, especially in this year’s election; however, I still think it’s interesting to observe how these trend lines change both across states and across time. And the other key addition is that I’ll be running more thorough statistical analysis, such as linear regressions.

Big Picture

Joe Biden- 2,804,040 (50.62 percent, up 3.35 percent from Clinton in 2016)

Donald Trump- 2,649,852 (47.84 percent, up 0.34 percent from 2016)

FiveThirtyEight Projection- 53.5 for Biden (up 2.9), 45.5 for Trump (down 2.3)

RCP Average- 50 for Biden (down 0.6), 45.8 for Trump (down 2)

OurProgress Projection- 52.4 for Biden (up 1.8), 44.6 for Trump (down 3.2)

In 2016, Donald Trump shocked the political world by flipping Michigan, a state that had gone blue in the previous six presidential elections and was widely expected to stay that way moving forward. In the four years since then, Michigan, along with the “Blue Wall”, became a central focus for Democrats’ 2020 strategy. Part of Joe Biden’s appeal during the primaries was his potential to win back the Rust Belt that his progressive opponents seemed to lack. This “electability” argument helped propel his way to the nomination. In the end, Biden lived up to these expectations (as well as my own), winning Michigan by almost 3 points.

As of Election Day, Michigan has had 200,103 COVID cases, ranking 14th in the nation. Despite this, its large population sets the incidence rate to only 2 percent, slightly below the national average and ranks 40th among the states. On the other hand, there have been 7,677 deaths, ranking 10th in the nation. And the fatality rate, 3.8 percent, is above the national average and ranks 8th among the states.

But the biggest story with the pandemic in Michigan is the economic impact. Going from a 4 percent unemployment rate in March to 23.6 percent the following month, Michigan sustained the third-largest increase to its unemployment rate in the nation, at 19.6 percent. Not to mention that among the top ten counties with the biggest unemployment hits in the nation, five of them are in Michigan. I discussed this more during the pandemic series, but a key reason why the pandemic took such a heavy economic toll on Michigan is the state’s reliance on manufacturing jobs that cannot be done remotely, resulting in widespread furloughs and layoffs during the early months of the pandemic. Luckily, unemployment did go down during the summer and early fall months; however, for Michigan, the economic crisis has been more pressing than the public health crisis.

Outside the pandemic, the other major election issue in 2020 is the nation’s reckoning with racial inequality, which erupted with the killing of George Floyd, a black man, during Memorial Day weekend. In the months since then, there have been thousands of protests organized by Black Lives Matter and its affiliates across all fifty states. According to the US Crisis Monitor managed by the Armed Conflict Location & Event Data Project (ACLED), there were 476 protests related to Black Lives Matter in Michigan between George Floyd’s killing in May and Election Day, the tenth highest of any state. Given the demographics of the state, this shouldn’t be too surprising. Michigan is the tenth most populous state. Michigan also has the 13th highest black population in raw numbers (17th highest by percentage of the total population), 13th highest amount of college graduates (but 33rd by percentage), and the 10th highest young adult population (16th by percentage). Considering that these demographics are all likely to participate in BLM protests, this activity makes sense.

Then there are exit polls, that point to a well-rounded improvement from 2016. Perhaps the most telling indicator of Biden’s improvement is that he retained 96 percent of Clinton voters here while winning 51 percent of those that didn’t vote in 2016, compared to Trump only retaining 94 percent of his voters from 2016. While this doesn’t sound like much improvement, such movements prove decisive in what was the closest state of the 2016 election. Even so, let’s break down movement for other types of voters.

For race, Trump carried over his win among white voters from 2016; however, Biden won 44 percent of these voters, an 8-point improvement. This makes Biden slightly more competitively here than he was nationally, where he only got 41 percent of white voters. Conversely, Trump made in-roads with Hispanic voters, going from 38 percent in 2016 to 44 percent this time around. Similar to Biden with white voters, Trump was more competitive here with Hispanic voters than he was nationally, only getting 32 percent of these voters. Meanwhile, Biden held Clinton’s margin with black voters, receiving 92 percent. This is slightly better than the 87 percent Biden received nationally.

Regarding educational attainment, trends from 2016 largely persisted. In Michigan, Biden won 58 percent of college graduates (compared to 55 percent nationally), an 8-point improvement from Clinton’s 2016 performance. Meanwhile, Trump won 52 percent of voters without a college degree (50 percent nationally), a 1-point improvement from his 2016 performance. And when broken down by race, Biden won 53 percent of white voters with a college degree (51 percent nationally; up 8 points from 2016), while Trump won 60 percent of white voters without a college degree (67 percent nationally; down 2 points from 2016).

As for age, Biden three of the four age brackets in 2020 (with the exception of the 30–44 bracket). He received 61 percent of the 20–29 bracket, a 4-point improvement from Clinton in 2016. He received 51 percent each from the 45–64 and 65 and over brackets, a 7-point and 4-point improvement respectively. And even in the 30–44 age bracket, Biden still received 48 percent of the vote among them, a 2-point improvement. Overall, these numbers are slightly more favorable for Biden than they are nationally, where Biden won among voters under 45 while Trump won among voters over 45.

Geographically, Michigan largely reflected the rest of the nation. Biden won 65 percent of urban voters in the state (60 percent nationally); however, this is actually 6 points less than Clinton’s performance among these voters. As for suburban voters, Biden narrowly lost this group with 48 percent (50 percent nationally), but this is a 6-point improvement from Clinton four years ago. And considering that suburban voters make up a much larger share of Michigan’s electorate than urban voters, this improvement more than makes up for his underperformance among urban voters. And for rural voters, President Trump won this group with 54 percent (57 percent nationally), only 2 percent less than in 2016.

Socioeconomically, Michigan is considered one of the hardest hit by the pandemic from an economic perspective. In addition to having one of the highest unemployment spikes in the nation from March to April, half of the top ten counties with the highest spikes are in Michigan. A lot of this is due to the state’s heavy reliance on manufacturing, which a disproportionately high number of jobs that cannot be done remotely. Interestingly, though, only 9 percent of Michigan voters said the pandemic has caused them severe financial hardship (compared to 16 percent nationally) and 55 percent said it has caused them no financial hardship (compared to only 44 percent nationally). This may have to do with the recovery that took hold in the state in the months since April, where many manufacturing plants reopened and the unemployment rate enjoyed a precipitous decline. Even so, Biden won 64 percent of voters that faced severe hardship (69 percent nationally) and 55 percent of those that faced moderate hardship (59 percent nationally), while Trump edged out Biden by winning 50 percent of voters that faced no hardship (60 percent nationally).

Regarding income brackets, Biden made up substantial ground across all levels. In Michigan, he won 61 percent of voters making under $50,000 (55 percent nationally), an 8-point improvement from Clinton four years ago. He also won 55 percent of voters in the $50,000-$99,999 bracket (57 percent nationally), a group that Trump won four years ago with 51 percent. And while Biden only got 47 percent of voters making over $100,000 (42 percent nationally), that is still a 4-point improvement from Clinton in 2016.

And when asked what was the most important issue in their vote, Michigan voters largely echoed the rest of the nation; however, the economy was the most cited answer, with 39 percent selecting that. Of those that said the economy was the most important issue, 86 percent voted for Trump (compared to 83 percent nationally). This is also reflected when voters were asked whether it’s more important to contain the virus or rebuild the economy. While 54 percent of Michigan voters said it was more important to contain the virus (and of those voters, 83 percent backed Biden), 42 percent prioritized the economy (and of those voters, 87 percent backed Trump). Conversely, Biden won 94 percent of voters that said the virus was the most important issue (81 percent nationally) and 93 percent of voters that prioritized racial inequality (92 percent nationally).

Now that we’ve established some general trends, let’s run through some OLS regressions to see how these factors affected both Biden’s vote share and his changes from Clinton’s 2016 performance. As it turns out, the single most powerful indicator of the 2020 election results by county for either candidate is the 2016 election results. In a single-variable linear regression, Clinton’s 2016 vote share (B=1.08) explains 97 percent of the variation for Biden’s vote share and Trump’s 2016 performance (B=1.07) is just as explanatory for his 2020 performance. As for the change in Democratic vote share (B=0.08), while Clinton’s 2016 vote share isn’t as all-encompassing, it’s still statistically significant in a single-variable regression.

These findings alone already tell us a lot about what changed between 2016 and 2020 (or more specifically, what didn’t change). But I want to go a little deeper than this. For each of the two dependent variables (Biden vote share and Democratic vote share change), I’ll be running four OLS regressions, each with different variable specifications. The purpose of this exercise is to test the effects of different variables while hopefully establishing robustness for different relationships. These specifications are clarified in the tables below, but there are two notes I wanted to make in advance. One is that none of the OLS regressions control for the 2016 Democratic vote share. This omission was made to reduce the risk of multicollinearity within the models (multiple independent variables, such as the share of black residents, are highly correlated with Democratic vote share). And the other is that Models 1 and 2 include the overall COVID incidence rate while Models 3 and 4 include just the early COVID incidence rate. This will address a key question of whether the entire span of the pandemic up until Election Day affected voting behavior or just a certain period. With that in mind, here are the regression results below.

For the most part, the variables are either not statistically significant nor that large. with regard to the Biden vote share. The pandemic metrics are very weak indicators, suggesting that the pandemic’s severity had little effect on voting behavior. Considering the severity of the economic hit for Michigan, it’s somewhat surprising how even those effects did not appear to have a strong effect on voting behavior.

On a similar note, the number of BLM protests also did not affect voting behavior in a statistically significant manner. With that said, the share of black residents is positively associated with Biden’s county vote share. This finding holds across all four models, indicating robustness. The share of Hispanic residents is also positively associated with Biden’s vote share; however, this does not hold in Models 2 or 4, which control for the share of college graduates and young adults, as well as the poverty rate. To that end, the relationship is not consistent.

Not surprisingly, Michigan has a strong and consistent education gap that holds across all four models. While Biden received more support in counties with a large share of college graduates (B=0.84; 0.81), he received less support in counties with a large share of adults without a high school diploma (B= -1.84; -1.65). This education gap is consistent with both the 2016 election results and the findings from the 2020 exit polls.

Finally, there are mixed findings with various socioeconomic indicators. There’s no statistically significant relationship with either median household income or the poverty rate; however, the Gini Index (B=0.91; 0.86) is positively associated with Biden’s vote share. This is noteworthy, considering that t-test results indicate that in counties that Biden won, median household income (u=$58,000) and the Gini Index (u=0.465) are higher than counties that Trump won (u=$50,624; 0.433), while the poverty rate is not significantly different between the two.

Next, let’s look at the regression models for the Democratic vote share change between 2016 and 2020.

Similar to the Biden vote share, there’s little indication that the pandemic or the BLM protest activity shifted voting behavior from 2016 in Michigan.

Regarding other variables, we still see the education gap persist, where counties with large shares of adults without a high school diploma shifted more into President Trump’s column (B= -0.33; -0.29) while counties with large shares of college graduates shifted more towards Biden (B=0.18; 0.18).

One other trend worth noting is that the poverty rate is associated with a shift towards President Trump (B= -0.23; -0.21). This suggests that compared to 2016, President Trump did a better job at appealing to voters on the lower rungs of the socioeconomic ladder, where white voters without a college degree are well-represented. This could also provide some evidence that President Trump’s framing of the pandemic as a predominantly economic issue resonated more in areas with more economically vulnerable voters (i.e. voters that saw the cost of public health restrictions more so than the proposed benefits).

Solidly Democratic Counties

Ingham- Biden Hold

Biden- 94,212 (65.18 percent, up 4.85), Trump-47,639 (32.96 percent, down 0.49)

OurProgress Projection- 65.7 for Biden (up 0.52 from actual), 34.3 for Trump (up 1.34)

Washtenaw- Biden Hold

Biden- 157,136 (72.61 percent, up 4.48), Trump- 56,241 (25.99 percent, down 0.86)

OurProgress Projection- 74.53 for Biden (up 1.92), 25.47 for Trump (down 0.52)

Wayne- Biden Hold

Biden- 597,170 (68.45 percent, up 1.67), Trump- 264,553 (30.32 percent, up 0.78)

OurProgress Projection- 73.59 for Biden (up 5.15), 26.41 for Trump (down 3.91)

First, there are the three solidly Democratic counties. As expected, Joe Biden performed well here. In this subsection, Biden received 68.8 percent of the vote, a 2.5-point improvement from Hillary Clinton in 2016, but still short of the 71 percent President Obama received in 2012. Even so, this translates to a net gain of 121,481 votes.

Biden saw his biggest improvements in Ingham and Washtenaw. This makes sense, considering that as home to major universities, these counties skew fairly young; over 20 percent of their populations are in the 20–29 age bracket, well above the national average of 13.7 percent. They are also fairly affluent and among the most highly educated counties in the state. It’s not too surprising that the OurProgress Projection came fairly close in forecasting Biden’s performance in these two counties.

As for Wayne County, Biden made a modest 1.6-point improvement; however, he still fell short of President Obama’s 2012 performance. OurProgress overestimated Biden’s performance by 5 points, noticeably more than Ingham or Washtenaw. Some of this has to do with Wayne’s status as a large, fairly diverse county with pockets of whites without a college degree, a group that has stuck with the President even in 2020. Even so, his strong improvement in the other solidly Democratic counties make up for this.

The pandemic took its toll in this subsection. As of Election Day, this subsection had 54,436 COVID cases, an incidence rate of 2.3 percent, slightly below the national average; however, with 3,256 deaths, its fatality rate of 6 percent is over double the national average. Most of these cases and deaths come from Wayne County, where the fatality rate was a staggering 7.1 percent. Regarding timing, the subsection took its largest hit in the early months of the pandemic, before slowing down in the summer and early fall months. Again, most of this trajectory is drawn by Wayne County. And for unemployment, the subsection took an 18-point hit, going from 4.6 percent in March to 22.8 percent in April. Again, of these three counties, Wayne took the biggest hit of 20.4 percent. In the months since then, the subsection has mostly recovered; however as of September, its unemployment rate of 10.6 percent was still above the national average.

Finally, as of Election Day, there have been 188 BLM protests in this subsection, most of which have been in Wayne County. This makes sense, given that the subsection is 31.5 percent black (more than double the Michigan average), 16 percent young adult (above the Michigan average), and have some of the most highly educated counties in the state. But, as established earlier, these protests had little effect on Biden’s vote or the change from 2016. This finding indicates that a major reason why protest activity didn’t affect Biden’s vote share is that such activity was disproportionately concentrated in parts of the state that are already heavily Democratic. Rather than reaching out to voters that could be persuaded to change their minds, many of these protests were simply preaching to the choir. As a result, the effect here is minimized.

Swing Counties

Genesee- Biden Hold

Biden- 119,390 (53.93 percent, up 1.59), Trump- 98,714 (44.59 percent, up 1.71)

OurProgress Projection- 60.57 for Biden (up 6.64), 39.43 for Trump (down 5.16)

Isabella- Trump Hold

Biden- 14,072 (47.83 percent, up 2.92), Trump- 14,815 (50.36 percent, up 1.77)

OurProgress Projection- 51.6 for Biden (up 3.77), 48.4 for Trump (down 1.96)

Kalamazoo- Biden Hold

Biden- 83,686 (58.35 percent, up 5.19), Trump- 56,823 (39.62 percent, down 0.79)

OurProgress Projection- 58.57 for Biden (up 0.22), 41.43 for Trump (up 1.81)

Kent- Biden Flip

Biden- 187,915 (52.05 percent, up 7.05), Trump- 165,741 (45.91 percent, down 2.17)

OurProgress Projection- 50.97 for Biden (down 1.08), 49.03 for Trump (up 3.12)

Leelanau- Biden Flip

Biden- 8,795 (52.04 percent, up 6.14), Trump- 7,916 (46.84 percent, down 2.21)

OurProgress Projection- 50.38 for Biden (down 1.66), 49.62 for Trump (up 2.78)

Marquette- Biden Hold

Biden- 20,465 (54.63 percent, up 5.98), Trump- 16,286 (43.47 percent, down 0.94)

OurProgress Projection- 55.07 for Biden (up 0.44), 44.93 for Trump (up 1.46)

Muskegon- Biden Hold

Biden- 45,643 (49.37 percent, up 1.6), Trump- 45,133 (48.82 percent, up 2.56)

OurProgress Projection- 55.29 for Biden (up 5.92), 44.71 for Trump (down 4.11)

Oakland- Biden Hold

Biden- 434,148 (56.36 percent, up 4.74), Trump- 325,971 (42.31 percent, down 1.2)

OurProgress Projection- 56.33 for Biden (down 0.03), 43.67 for Trump (up 1.36)

Saginaw- Biden Flip

Biden- 51,088 (49.43 percent, up 2.36), Trump- 50,785 (49.14 percent, up 0.93)

OurProgress Projection- 53.64 for Biden (up 4.21), 46.36 for Trump (down 2.78)

Biden also improved upon Clinton’s performance in this subsection. Whereas Clinton only won five of the nine counties in 2016, Biden won eight of them this time with an average per-county vote share increase of 3.8 percent. Biden received 54.4 percent of the vote here, a 4.6-point improvement from Clinton in 2016 as well as a 0.5-point improvement from President Obama in 2012. This translates to a vote total of 965,202 votes, a net gain of 197,630 votes from Clinton four years ago.

As discussed in the pre-election article, there are several urban/suburban counties in this subsection: Genesee, Kent, and Oakland. Genesee, which Clinton won despite seeing an 11-point drop from Obama’s 2012 vote share, only saw a 1.6 percent improvement for Biden this time around. Much of that progress was stymied by the high poverty rate and low share of college graduates. In Oakland, which had already gone Democratic in the previous six elections, gave Biden improved Clinton’s performance by 4.7 percent. In addition to its legacy as a Democratic county, Biden was helped here by the relatively large share of racial and ethnic minorities in this county, along with the high concentration of affluent and college-educated voters. But the biggest gain was in Kent County which, with a 7-point boost from 2016, was enough to flip it blue. Overall, Biden received the suburban boost in these three counties he needed to elevate his vote share.

Outside the suburbs, one key flip for Biden came in Leelanau County, which has only gone blue four times since 1884 (most recently in 2008). By most measures, this is not an area that a Democrat like Joe Biden should win. It’s a popular retirement destination, where 32 percent of its population is over 65 (and 60 percent is over 45). It’s also 93 percent white, 42.8 percent college educated, and above the state average for most socioeconomic indicators, and. On the other hand, while it seems that Leelanau’s population wouldn’t be the biggest fans of the Democrats nor the progressive movement’s policy proposals, it also isn’t supportive of President Trump’s brand of conservatism. And while this county has been mostly spared from the pandemic through Election Day, these older voters would be more responsive to the public health side of the crisis as they are at greater risk of serious complications or death from the virus. To that end, Leelanau is an interesting story from this year’s election.

As of Election Day, there have been 68,139 COVID cases in this subsection, setting the incidence rate at 2.2 percent (slightly below the national average). On the other hand, there have been 2,182 deaths, setting the fatality rate at 3.2 percent (above the national average). Regarding timing, there has been a steady escalation in cases as the months have gone on. And for unemployment, this subsection took a 18.6-point hit, going from 3.4 percent in March to 22 percent by April. In the months since then, though, this figure has mostly recovered; however, unemployment was still at 7.6 percent as of September.

Finally, as of Election Day, there have been 151 BLM protests in this subsection, with half of them occurring in Kent and Oakland. This makes sense, given that these are the two largest counties in this subsection as well. While they aren’t the most racially diverse or the youngest counties, they are both highly educated and fairly affluent, making them reasonable hubs for protest activity. Kalamazoo County also had a lot of activity, hosting 26 protests, despite not being especially diverse or young; however, it did have a lot of college graduates. And Genesee County, despite having the largest black population in the subsection, only had 18 protests. Not nothing, of course, but also indicative of how Genesee isn’t as beholden to progressive movements as its suburban counterparts. Overall, it appears that while this subsection has plenty of counties with substantial college graduate populations, most of the protest activity comes more from its sheer size than any indicative demographics; the subsection is 13.6 percent black (slightly below the Michigan average) and 14.3 percent young adult (slightly above the Michigan average). And while Biden made substantial improvements within this subsection, it seems that more of that comes from natural shifts in how its demographics voted than a catalyst such as increased protest activity.

Obama-Trump Counties

Bay- Trump Hold

Biden- 26,151 (43.4 percent, up 2.55), Trump- 33,125 (54.97 percent, up 1.5)

OurProgress Projection- 50.02 for Biden (up 6.62), 49.98 for Trump (down 4.99)

Calhoun- Trump Hold

Biden- 28,877 (43.57 percent, up 2.56), Trump- 36,221 (54.65 percent, up 1.18)

OurProgress Projection- 49.75 for Biden (up 6.18), 50.25 for Trump (down 4.4)

Eaton- Trump Hold

Biden- 31,299 (48.66 percent, up 4.3), Trump- 31,798 (49.43 percent, up 0.32)

OurProgress Projection- 51.1 for Biden (up 2.44), 48.9 for Trump (down 0.53)

Gogebic- Trump Hold

Biden- 3,570 (43.19 percent, up 3.28), Trump- 4,600 (55.65 percent, up 0.83)

OurProgress Projection- 50.48 for Biden (up 7.29), 49.52 for Trump (down 6.13)

Isabella- Trump Hold

Biden- 14,072 (47.83 percent, up 2.92), Trump- 14,815 (50.36 percent, up 1.77)

OurProgress Projection- 51.6 for Biden (up 3.77), 48.4 for Trump (down 1.96)

Lake- Trump Hold

Biden- 2,288 (36.19 percent, down 0.2), Trump- 3,946 (62.41 percent, up 3.12)

OurProgress Projection- 46.24 for Biden (up 10.05), 53.76 for Trump (down 8.65)

Macomb- Trump Hold

Biden- 223,952 (45.31 percent, up 3.26), Trump- 263,863 (53.39 percent, down 0.19)

OurProgress Projection- 50.13 for Biden (up 4.82), 49.87 for Trump (down 3.52)

Manistee- Trump Hold

Biden- 6,107 (41.67 percent, up 2.15), Trump- 8,321 (56.78 percent, up 1.91)

OurProgress Projection- 50.83 for Biden (up 9.16), 49.17 for Trump (down 7.61)

Monroe- Trump Hold

Biden- 32,975 (37.84 percent, up 1.64), Trump- 52,710 (60.48 percent, up 2.19)

OurProgress Projection- 44.18 for Biden (up 6.34), 55.82 for Trump (down 4.66)

Saginaw- Biden Flip

Biden- 51,088 (49.43 percent, up 2.36), Trump- 50,785 (49.14 percent, up 0.93)

OurProgress Projection- 53.64 for Biden (up 4.21), 46.36 for Trump (down 2.78)

Shiawassee- Trump Hold

Biden- 15,347 (39.13 percent, up 2.35), Trump- 23,149 (59.02 percent, up 2.65)

OurProgress Projection- 46.94 for Biden (up 7.81), 53.06 for Trump (down 5.96)

Van Buren- Trump Hold

Biden- 16,803 (43.01 percent, up 3.17), Trump- 21,591 (55.27 percent, up 1.5)

OurProgress Projection- 49.36 for Biden (up 6.35), 50.64 for Trump (down 4.63)

Finally, there’s the twelve Obama-Trump counties, which present a different story from the swing counties. Of these counties, Biden only flipped one (Saginaw) that overlapped with the swing counties. Not that Biden didn’t make improvements here. On average, he increased his vote share in each county by about 2.5 percent. Overall, Biden received 44.7 percent of the vote, up from the 41.8 percent Clinton got in 2016, and netted an additional 87,165 votes. Even so, this remains a far cry from the 51.7 percent President Obama received in 2012. And it’s also a far cry from the OurProgress projections, which expected Biden to flip seven of these counties.

Some of these misses were larger than others. For example, the model overestimated Biden’s performance in Lake County by 10 points, a place it didn’t even expect him to win in the first place. But the main loss I want to focus on is Macomb County, a suburban region neighboring Oakland County. In my pre-election article, I highlighted how these two counties have diverged in their voting behavior during presidential elections, 2016 included. Even so, there was an expectation that of the Obama-Trump counties, Macomb would be one that Biden would be highly competitive in, given its suburban nature, making it a microcosm of Biden’s increased support among such voters nationally. Even the OurProgress projection saw Biden winning this county, albeit narrowly. But in the end, it didn’t produce that result.

Biden did slightly better than Clinton in 2016, enough to shift the county into the “swing” category, but President Trump still won by a decent amount. Meanwhile, its neighbor Oakland more closely matched the national trend of Biden winning over suburban voters, exacerbating the divergence. Demographic differences can account for this. Macomb has significantly fewer college-educated voters than Oakland. It also isn’t as affluent as Oakland, although it’s still above the state average for median household income. And Macomb has more residents that were born in Michigan than in Oakland, indicating that its population holds deeper roots in their communities while having fewer transplants moving in strictly for economic opportunity. All these factors contributed to Macomb sticking closer to the Republican candidate than Oakland County, despite both of them being considered suburban.

Credit should be given to President Trump for retaining much of the ground he had taken in Michigan four years ago. Looking at these counties, it’s clear that high turnout would aid Trump in retaining his 2016 base. Many of these are below the state average for median household income, six of the counties he retained had a poverty rate above 15 percent, and none of them are more than 30 percent college educated, reflecting the working class nature of his base. Not to mention that each of these counties are majority white; the one county he couldn’t keep, Saginaw, was the only one in this subsection to be less than 80 percent white. The fact that President Trump kept these voters in his corner is impressive; however, this advantage wasn’t large enough to overcome the surge in turnout across the other counties. While it was helpful to keep his base intact, he wasn’t able to expand beyond this, which proved costly statewide.

As of Election Day, there have been 38,280 COVID cases in this subsection, setting the incidence rate at 2.1 percent (below the national average). On the other hand, there have been 1,542 deaths, setting the fatality rate at 4 percent (above the national average). Regarding timing, there has also been a steady escalation in cases as the months have gone on, meaning that this subsection never had a break from the pandemic. And for unemployment, this subsection took a 21.6-point hit, going from 4.2 percent in March to 25.8 percent by April. In the months since, it has mostly recovered; however, unemployment was still at 8.3 percent as of September.

Finally, as of Election Day, there have only been 63 BLM protests in this subsection, with more than half of them occurring in Macomb and Saginaw, the two largest counties. Although it’s more conservative than its neighbor in Oakland, Macomb is still a populous county in the Detroit metropolitan area, meaning there are simply more people that are able and willing to organize. And as mentioned above, Saginaw is the most racially diverse county in the subsection, where 19.3 percent of its population is black and 8.7 percent is Hispanic. While this smaller number is partially due to the fact that this is the smallest subsection by population, it’s also indicative of the more conservative bent of the area, where BLM has lower support. The subsection is only 10 percent black (below the Michigan average), 13.3 percent young adult (roughly the Michigan average), and, as mentioned earlier, has a pretty low share of college graduates.

Reclassifications

The subsections I drew out in the pre-election articles are based on the 2016 election results. With each new election cycle, this begs the question of how these subsections would be redrawn based on the parties’ more recent performances. For the most part, 2020 didn’t offer much change to the composition of these subsections. This will be a recurring trend we’ll see across the different swing states and a testament to the strength of party polarization. Even with President Trump’s low approval ratings and unfavorable conditions, he still managed to retain a sizable portion of his base intact. Even so, there was still some movement involved. Here are all the counties that changed subsections:

Berrien (Solidly Republican to Swing)

Clinton (Solidly Republican to Swing)

Eaton (Solidly Republican to Swing)

Grand Traverse (Solidly Republican to Swing)

Kalamazoo (Swing to Solidly Democratic)

Kent (Trump to Biden)

Leelanau (Trump to Biden)

Macomb (Solidly Republican to Swing)

Oakland (Swing to Solidly Democratic)

Saginaw (Obama-Trump to Trump-Biden)

As seen above, all of Michigan’s reclassifications are in favor of Joe Biden. This reflects how, in general, Biden improved his standing in Michigan compared to Hillary Clinton in 2016. This will not be the case with every state, however.

Conclusion

I predicted in my pre-election analysis and my predictions that of the states Trump won in 2016, Michigan would be one of the first to flip back to the Democrats. I said this based on Michigan’s history as a traditionally Democratic state at the presidential level and the fact that it was one of the closest states in 2016, only being decided by 10,704 votes. And ultimately, my prediction was right as Biden won the state by 154,188 votes, more decisive than a lot of other swing states on the board.

With that said, though, President Trump didn’t lose as badly as he could have, which is a trend we’ll see with other swing states. OurProgress projected Biden to win the state by 7.8 percentage points with a margin of almost 400,000 votes. There were multiple counties where the model overestimated Biden’s performance, including those that Biden ended up winning, such as Wayne, Genesee, and Saginaw. And as mentioned above, President Trump was able to retain almost all of the counties he had flipped back in 2016. This speaks to President Trump’s ability to command high voter turnout and loyalty, particularly among voters without a college degree and those in the middle and lower rungs of the socioeconomic ladder.

What’s particularly noteworthy is the lack of stark differences in the pandemic’s severity across these subsections. While the COVID fatality rate is the highest in the solidly Democratic counties, it’s also fairly high in the swing counties and the Obama-Trump counties. The incidence rate is roughly the same in all three subsections. And while the unemployment spike was the largest in the Obama-Trump counties, it was still considerably high in the other subsections and it enjoyed a fairly substantial recovery in the summer months. One might conclude from these observations that in Michigan, places that took larger economic hits from the pandemic are the places that supported President Trump in higher numbers; however, the regression model doesn’t establish a strong statistically significant relationship with this finding. Rather, it seems that the places that took the biggest hits economically are those that are less well-off to begin with and already supportive of President Trump (such as counties with fewer college graduates). To that end, pre-existing demographic and socioeconomic conditions help explain both the pandemic’s severity and voting behavior.

And finally, while the regression models didn’t find a statistically significant relationship, there is a sharp difference in protest activity across the subsections. The solidly Democratic counties and the swing counties were notable hubs for BLM protests in the months leading up to the election. This is due to their proximity to cities such as Detroit and Grand Rapids as well as their population’s notable representation of black residents, young adults, and college graduates (three demographics that are most likely to organize and participate in such events). On the other hand, there were relatively few protests in the Obama-Trump counties, even after accounting for their population. This makes sense, given that progressive ideals, which have become more vocal within the Democratic Party, garner less support in these areas in favor of backing the President.

Overall, Michigan was one place where Joe Biden delivered on his promise to rebuild the “Blue Wall”. While credit must be given to President Trump for remaining more competitive than expected here, Biden’s victory here was crucial to his successful presidential campaign.

So that will do it for the first part in this series. If you enjoyed this, please like and follow the Book Aisle. Also share this article on Facebook, LinkedIn, and other social media platforms.