Election 2016 Forecast / Turnout and Result: Whites with no College Degree
Many elections have key demographics and messages that define them. For 2016, each major party has targeted a key demographic with specific messages. The Democrats have targeted women with the underlying message that it is important to finally elect the first woman president. Buttressing this message are attacks on Trump attempting to show that he is a sexist and misogynist which disqualify him for the presidency. The Republicans have targeted the middle class with the message that illegal immigration and international trade practices have specifically hurt them. As African-Americans have so closely followed the Democrats and as many Hispanics might feel offended by Republican discussion over immigration, much of the Republican message directed at the middle class has gone towards the demographic of whites without a college degree.
This election has many moving pieces, so it is dangerous to over-simplify it to such a great extent — however, in a close race, the winner will likely be determined by these two demographics and their turnout. Again simplifying to a dangerous degree, if Democrats are successful in attracting higher numbers of women and getting them to turnout, Clinton has a strong chance of winning. If Republicans can motivate the demographic of whites without college degrees to turnout, Trump has an excellent chance of winning.
The generally accepted belief at present appears to be that women will in fact turnout at relatively high levels in order to prevent Trump from getting in the White House. As discussed elsewhere, however, women are showing lower levels of interest in the 2016 election as compared to men and the nation, to a somewhat surprising extent, appears to be viewing the election on a gender neutral basis. Additionally, female turnout is already significantly higher than male turnout, making a scenario of rapidly increasing female turnout difficult to backup. In other words, there are few indications that Clinton will get more than a nominal boost by the fact that she is the first female candidate from a major US party for the presidency. A Clinton win therefore would likely come from a broader range of demographic groups and not be so rooted in the female vote.
In contrast, the demographic group of whites with no college degree seems primed for a jump higher in turnout. Especially if the US votes in a gender neutral fashion, the whites with no college degree demographic group could be the determining factor.
Table 1: Net Support (Percent Supporting Clinton minus Percent Supporting Trump) of different demographic groups for 2016 US Presidential Election
|
Net Support for Clinton |
|
| Blacks |
83% |
| Hispanics |
24% |
| White, College + |
14% |
| White, Non-College |
-25% |
Source: PEW Research
In terms of apparent support from very large demographic groups, Trump’s core comes from whites without college degrees. Historical trends also show that Trump’s lead within this demographic is strong. The following chart is of non-college white males and their support for the Republican or Democratic candidate.
Chart 1: Non-College White Men Support for the Republican Candidate versus Democratic Candidate, 1980 to 2016e
Source: ABC News
ABC News, by comparing the results of their most recent poll and exit polls from previous elections, shows that support for Trump from non-college white males is at its highest pro-Republican spread since the Reagan election of 1984. The net difference between those supporting the Republican candidate and Democratic candidate is at 41 percentage points, which surpasses Reagan’s 37 point lead in 1984. The same metric for non-college white women is 13 points, much lower than that of men but still solidly positive. As a broad male-female group, whites with no college degrees are strongly pro-Trump.
In terms of relative size of the non-college educated white demographic, it is still the largest in the US of the following broad categories.
Chart 2: Percent Breakdown of US Electorate by Demographic Group, Number of Eligible Voters, 2016e
Source: Center for Immigration Studies, Coogan
This demographic has been mostly forgotten by politicians. It has been on the losing end against globalism, which until recently was seen as a panacea making it political suicide to attack it to any significant extent. Additionally, this group has been shrinking in absolute terms. For instance, whites without college degrees as a group declined by 3% from 2004 to 2012 in comparison to a 9% growth for the US in general. Other demographic groups such as Hispanics (45% increase over the same period) grew at very high rates. Political wisdom stated that the parties should align themselves with groups growing at the fastest pace in order to ensure future successes. These and other factors have made the whites with no college degrees group less appealing to politicians over the last few decades.
Trump, whether you like him or not, has provided the country with a great (and perhaps unrecognized) service by speaking directly to this disenfranchised demographic group and making them excited about politics again. Considering the fact that this group represents almost half of the eligible voters in the US makes its re-engagement with the political process an important event. Though many might disagree, his enthusiastic live rallies which appear to be heavily populated by this demographic seem to indicate that this is not just a temporary political blip but a movement with some real grass roots support behind it.
Let’s review before going further. So far, we’ve seen that the group whites with no college degrees disproportionately supports Trump. Using historical metrics, this support appears to be breaking records for either of the major parties. And, although this demographic group has been shrinking, it is still extremely large. Additionally, Trump is the first major ticket politician in many elections who speaks directly to this demographic and has stirred a following in them that appears more like a political movement. These observations lead to the main question of will turnout of this demographic be large enough to impact the race to any great extent?
The first part of this question can be answered by using reversion to the mean logic. Basically, voter turnout ratios shifted significantly with Obama and there is a strong argument that turnout ratios should revert to more normalized levels post-Obama. One of the topics covered in other posts was the demographic make-up of Obama’s ‘Coalition’ which helped him get elected in 2008 / 2012. Minority groups increased their turnout in some cases to record levels whereas turnout for whites actually declined. The marginal difference from the pre-Obama period to the 2012 election is shown in the following chart.
Chart 3: Net Change in Voter Turnout in Percentage Points from US Presidential Election 2012 from 2004 for Major Demographic Groups
Source: Center for Immigration Studies, Coogan
As you can see the net swing in voter turnout from the pre-Obama period (2004) to Obama’s last election (2012) was substantial. Minority groups dramatically increased their levels of voter turnout while whites turned out at much lower levels. As covered in other posts, 2016 data shows that much of Obama’s Coalition (from 2008 and 2012) are showing slightly lower indications of voting in 2016. Such data confirms the basic observation that with many demographics having exceeded or approached voter turnout records during the Obama period, it would make sense for there to be a reversion to the (previous) mean once that catalyst (Obama) is no longer present. On the other hand, the white demographic will likely increase voter turnout, likewise reverting towards pre-Obama levels.
As a side note, this topic is very central to the 2016 race and is being sorely under analyzed and reported. Voter turnout clearly had a significant impact on the race in 2008 and 2012, so this should be an obvious topic. Likewise, it is no secret that Trump is attracting +10,000 people at regular rallies who appear to be heavily weighted towards the ‘whites with no college degree’ demographic. It is extremely frustrating from an analytical perspective that more time is not spent gathering data on this demographic and analyzing how its turnout can impact the current election. Unfortunately, from an analytical perspective, almost all of the quantitative commentary is conducted around the latest poll releases. Somewhat ironically, if voter turnout levels shift dramatically from the previous elections, which they are expected to, such polls are biased anyway making much of the focused commentary on polls misleading.
Returning to the reversion to the mean analysis, if we assume that voter turnout levels for each of the main demographic groups covered here return to 2004 levels, the marginal impact will greatly benefit Trump. In other words, voter turnout as compared to 2012 levels will greatly increase for the white demographic and decrease for the minority groups. As minority demographics have a high propensity to vote for Democrats, any decrease in voter turnout will be a major blow to Clinton. Conversely, as we have seen, the whites with no college degree group has a very high propensity to support Trump so any increase in its turnout will greatly benefit him.
At the very low end of estimates, Trump should pick up 2 percentage points of popular vote just by assuming a reversion to the mean for voter turnout. This is an extremely conservative estimate and one that does not appear to be reflected in current polls. Again, polls use turnout models based mostly on recent elections which implies that polls are heavily weighted towards the assumption that Obama’s Coalition will reform to the same extent. If there is reversion to the mean post-Obama to pre-Obama voter turnout levels, then Trump’s actual support is much higher than current polls show. A swing of 2 percentage points might not seem like a lot but in states that are too close to call, such a swing could determine the winner.
From this point, we can start to build scenarios which is the second part of the answer. We can assume that this election is very different from 2004, or the base year we are using for the reversion analysis. From other posts, we have seen that voter interest as measured by traditional and on-line metrics is much higher in 2016 than 2004 which implies a much higher voter turnout rate. We have also seen that interest in the current election is lower in minority groups and higher with whites. Additionally, we have seen in this post that net support for the current Republican candidate has never been higher among the whites with no college degree demographic.
Some very basic scenarios based on the aforementioned observations are included in the following table.
Table 2: Net Popular Vote Swing for US Presidential Race based on Scenarios for Turnout and % Vote of Whites without College Degrees Demographic Group
|
Turnout |
% Vote of Marginal Change of Turnout |
Popular Vote Swing in Trump’s Favor |
|
| Scenario 1 |
Reversion to 2004 |
No Change |
2.0% |
| Scenario 2 |
Reversion to 2004 |
High Whites without College Degree for Trump |
4.0% |
| Scenario 3 |
High Turnout |
High Whites without College Degree for Trump |
6.2% |
| Scenario 4 |
High White Turnout, Minority 2004 Turnout |
High Whites without College Degree for Trump |
7.2% |
Source: Center for Immigration Studies, Coogan
These scenarios show just how much the popular vote can change by modifying the expected turnout and the percent vote of the marginal increase of the whites with no college degree group going to Trump. In the rather conservative Scenario 1, which includes reversion to 2004 turnout levels, Trump gains approximately 2 percentage points — or enough to determine the outcome for a number of swing states.
Scenario 2 keeps the same turnout assumptions, but also assumes that 100% of the turnout increase from the whites with no college degree will go to Trump. This might seem excessive at first glance, but once you take into consideration the size and consistency of Trump’s rallies making such an assumption is logical. The rationale is that turnout for this demographic has been declining and would likely not increase without a specific reason. For this demographic, Trump is that reason. Put another way, if this group’s voter turnout remains flat it will likely be because Trump failed to motivate the marginal voters enough, but if voter turnout increases it will likely be a direct result of Trump’s unique dialogue with them. To be clear, the 100% refers only to the marginal increase in votes, not to the percent breakdown of votes for the entire demographic.
Scenario 3 assumes high turnout across the board. For each major demographic group, voter turnout is forecast above respective 2004 levels. In this scenario, national turnout spikes higher with an emphasis on whites more than making up for turnout declines from 2004 to 2012.
Scenario 4 is fairly aggressive in that it assumes the strong white turnout from Scenario 3 but the reversion to the 2004 turnout of Scenario 1 and 2 for minority groups. This scenario can be defended using poll responses showing relatively lower excitement and interest in the 2016 election from minority groups as compared with the white demographic and/or the aggregate voter base which are showing record-high interest levels.
Creating scenarios such as these should be second nature to political analysts but they are actually fairly rare. This in and of itself should make independent thinkers hesitate the next time they listen to political pundits on TV. Not to get on a high horse but if you don’t have the time to create some simple turnout scenarios for an election that should see large swings in turnout it does not seem like you should have a place conducting public analysis.
Having said that, Brookings Institute did conduct a scenario analysis that played around with some turnout levels – with specific focuses on white non-college educated men and white college educated women, so hats off to them. Their conclusion basically states that Trump could receive something like 4.86 million additional votes by increasing the turnout for white males without college degrees. They deem this to not be enough to change the outcome of the election. However, it is important to note that Obama won in 2012 by 4.98 million votes. In other words, according to Brookings Institute, the same increase in turnout of the white non-college male vote would have produced an election too close to call in 2012.
A key factor is your starting point. If you use the 3.86% margin of popular vote victory from 2012 then the swings presented in the previous table will make or break the 2016 election. If you use the current RealClearPolitics poll average showing Clinton with a 5% margin lead, then this group can still change the outcome of the race, but only under one of the more aggressive scenarios. Independent of your starting point, it certainly seems like this race is much less in-the-bag than most believe. Using the swing estimates from the previous table to adjust national polls, we are very likely within the margin of error with Clinton up slightly, or if you prefer the aggressive scenario Trump up slightly.
Taking this analysis on to the state level, we can see that the surprises in 2016 will also likely break the mold from the past. Much state-level focus has been on Florida, Virginia, North Carolina, and Ohio as these states were won by unusually small margins in the past few elections. However, if you believe that the demographic group of whites with no college degrees will be the major surprise factor in 2016, then a new set of states should take center stage — such as Iowa, Wisconsin, Minnesota, New Hampshire, Michigan, and Ohio (which is on both lists). These states, which all voted Democrat in 2012, can be seen in the following chart which depicts the relative size of the whites with no college degree demographic in comparison to the US average within states that had relatively small margins of victory in the 2012 US Presidential race.
Chart 4: Relative Size of State-Level Demographic Group Whites with No College, as measured by state’s whites with no college as a percent of state population minus the national average of same, states shown had the tightest races in 2012 US Presidential Election
Source: Cook Political, Coogan
The previous chart shows which states might show the largest electoral surprises due to turnout of whites with no college spiking higher. Many of the states on the left hand side of the chart have been solidly democratic for generations. For instance, Wisconsin has not voted for a Republican presidential candidate since Reagan and Minnesota since Nixon. Other states as New Hampshire, Iowa, and Ohio however have had more of a tendency to swing between parties — certainly these states should be pushed Republican in 2016 from Democrat in 2012 due to a spike higher in turnout. These three states flipping into the Republican column in 2016 will not be enough to change the course of the election starting from the 2012 Electoral map.
However, with such a starting point, the viable electoral map looks much improved for Trump, much more so than almost any political analyst is forecasting. For instance, given the addition of Ohio, Iowa, and New Hampshire, if Trump is able to win Florida, which is currently a very tight race not likely to be highly influenced by whites without college degrees, the Republicans would win the White House.
Summarizing, the key demographic group of whites with no college degree appears to have the power to swing the 2016 race depending on its relative turnout in comparison to other major groups in the US. Polls have discounted this group as it has posted low and declining turnout levels during the last few elections. As polls normally re-weight their response results based on assumptions derived from the most recent elections, a significant spike higher in turnout from this group would in fact throw almost every poll off significantly. The degree will depend on turnout levels. From information currently available, this group looks more enthusiastic about Trump than it has for any politician in a generation. In contrast, other demographic groups which heavily favored Obama do not appear as enthusiastic to vote in 2016 which will likely result in a relative decline of their respective turnout ratios. The end result should amount to a 2 to 7 percentage point swing in popular vote in favor of Trump that is simply not getting picked up by polls at present. This is not the result of any nefarious preference for Clinton but just how polls are calculated — they inherently will not do well under scenarios where turnout radically shifts from one election to another. The electoral map will likely produce some surprises based on higher turnout from this demographic — namely states that have a relatively high level of whites with no college degrees will be the most apt to produce ‘unexpected’ support for Trump. Even under more conservative assumptions, states like Ohio, Iowa, and New Hampshire which voted Democrat in 2012 will likely flip to Republican in 2016. If other state polls are too tight to call going into the election, such as perhaps Florida, Colorado, or Pennsylvania, a surge in this demographic’s turnout could declare the winner. In short, the ‘whites with no college degree’ group appears to have the power to determine the outcome of the 2016 election but only if they turnout in unusually high numbers breaking its cycle of disengagement from the political process.