Post-Election 2016 Part 1: Documenting the Polling Error

The level of shock attached to the 2016 election result surely had a lot to do with the once inconceivable and conventional wisdom-breaking phenomenon of Donald Trump becoming the 45th president of the United States. But tied into this notion of an unprecedented Trump presidency was what public opinion polling–a staple of the modern media’s election coverage–was saying in the run-up to last Tuesday.

Presidential vote intention polls oscillated plenty starting from the first months of 2016 through the week before the election. But even then, the signal concerning the nation’s vote preference mood was clear: Hillary Clinton maintained around a three to four percentage point lead over Donald Trump in terms of national popular support. Given national vote share’s typically strong relationship with Electoral College fortunes, electoral success seemed justifiably likely for Clinton. That’s why right before the election, the major forecasts–either largely or entirely derived from polls–pegged Clinton’s chances anywhere between 71 and 99 percent probability of winning the election.

As Tuesday night progressed, Trump emerged victorious in the early morning of Wednesday, and Americans looked back at what happened, much attention has been shifted to the polls. Clearly, it seemed polls measuring people’s vote preferences were off in some way, perhaps larger than they every have been, and that this error explained the unexpected. That’s certainly part of the story, but it’s a bit more nuanced than that.

Most importantly, at the national level, talk of a large polling error is exaggerated. Though Clinton’s popular vote lead only stands at about 0.5 percentage points right now, that edge will continue to expand as more votes are counted. It’s possible it reaches around one to two percentage points, which makes for about a two percentage point polling miss at the national level (compared to her three/four point lead in the polls). Not only would this level of deviation fall within the margin error for what most polls were saying before the election, but the polling error could also prove smaller than the one in the 2012 election. The big difference is direction: polls were biased against Obama in 2012, but biased against Trump in 2016.

Instead, polling error materialized differently across different states. Importantly, it wasn’t systematically in one direction, and rather biased against Trump and Clinton in different states. The election swung, however, on the direction of the bias where the state races were closest. In comparing the polls to the actual election results, the error was largely biased against Trump in swing states whose pre-election polls for the most part showed a Clinton victory on the horizon.

First, a few notes on how I’m going about this analysis:

  1. My estimate of “polling error” is based on comparing the Democratic (Clinton’s) percentage point margin of victory in pre-election polls to the actual Democratic margin of victory in the election for the main part of this analysis. For some parts of this analysis, I can clearly show where Clinton’s support was underestimated and where Trump’s was, such as in the below heat map of polling errors:huff_margin_error_map
  2. You’ll notice that for this map, the polling error pertains to one of the two main aggregators of election polls, HuffPost Pollster, the other one being RealClearPolitics. While similar for the most part, Pollster includes some online polls that RCP does not, RCP includes some landline-only polls that Pollster does not, and Pollster also uses a trend line estimate that’s close to but not the same as the averaging of polls RCP employs. In later parts of this analysis, I will take the average of these estimates of state-level races to give a general sense of “what the polls are saying.”
  3. I see four different ways to measure state polling errors: 1) Democratic margin (actual Democratic margin minus Democratic margin in polls), 2) absolute margin (the absolute value of the above equation and thus ignoring the direction in which direction the polling was biased), 3) Democratic vote share (actual Clinton vote share minus Clinton vote share in polls), and 4) Republican vote share (actual Trump vote share minus Trump vote share in polls).
  4. In parts of the analysis where there is an average of the Pollster and RCP reads of polls, Alabama, Hawaii, North Dakota, and Wyoming are excluded because RCP did not have poll averages for these states, and thus I could not average this aggregate polling picture with that of Pollster.
  5. Final note: this polling error data that includes actual election results is based on data as of about Thursday, November 10th. While a bit old a few days later, this amounts to very minor discrepancies in polling error calculations.

Here’s what gives the best idea of what happened with state-level polls in the election:

Data from HuffPost Pollster, RealClearPolitics, and U.S. Election Atlas.

Here I’m using the Democratic margin of victory as a way to measure polling error. States that fall on the 45-degree line in the above chart are those whose pre-election polling estimate matched their actual voting results last Tuesday. If states fall above the line, the polling error was biased against Clinton and underestimated her margin. If they fall below the line, then the error was biased against Trump. States in red indicate where the margin of victory for either candidate was 10 percentage points or less–loosely defined swing states (since there are more here than usually qualify as “swing”). The x-axis variable, Democratic Vote Margin in Polls, is taken from the average of Pollster and RCP state poll trend lines/averages. As explained before, this is a composite indicator of what state polls were showing.

In several very blue states, such as California, New York, and Washington, the polling error was in fact biased against Clinton. In the more solidly red states, the polls missed much more so on Trump’s margin of victory. The problem for Democrats–and where the election was won–came in the middle: of the 17 states that were decided by 10 percentage points or less, the error was biased against Trump in 15 states. In 10 of these states, Trump won. In five especially crucial ones–North Carolina, Florida, Pennsylvania, Michigan, and Wisconsin–a projected Clinton win based on polls did not match the actual result of a Trump victory. During the last two elections for these five states, Democrats won nine times out of the total 10 (the only loss coming in 2012 in NC from this set). In 2016, all five states swung to Trump, and that proved decisive for his electoral victory and for cementing where the polling error mattered most.

There are a few other ways to look at the polling error and the general form it took. Here I come back to the four different approaches to evaluating error I identified before. Differentiating by poll aggregator–which as a whole combine to form my “position of pre-election polls” proxy in the prior graph–could also prove informative (I include a column for the average of these as well). The mean and standard deviation of polling errors also adds another layer of understanding–how far off polls were from the final result and how widely did their deviations from the actual result vary, respectively. That’s what I show in the below table:

Data from HuffPost Pollster, RealClearPolitics, and U.S. Election Atlas.

Polls were off by an average of -4.64 points in terms of Democratic margin by state, and thus 4.64 points biased against Trump on average. That error was larger for HuffPost Pollster’s trend line estimates. This makes sense given what these estimates attempt to do: smooth out the incoming stream of polls and not react as quickly to them as simple averaging would. With late voter deciders breaking toward Trump according to exit polls, this type of estimate would be less suited to pick up this kind of movement. Generally speaking, given the stability of pre-election polling and noisy swings, this would qualify as the more prudent approach. But this wasn’t as helpful an approach in this election. Reliance on polls meant depending on numbers that in hindsight now seem likely to have systematically underestimated Trump support for some time–and thus you couldn’t completely chalk all this up to a story of differential non-response.

But back to the above table. The absolute error regardless of the direction of the bias came out at 5.85 points, and HuffPost Pollster estimates once again had greater errors than RCP averages. Regardless of this comparison, a state-idea average of 5.85 points is a lot. The lack of a parallel comparison to 2012 and earlier elections precludes any definitive judgment, but that type of error remains fairly large. One caveat to keep in mind however: some states had little polling done before the election, and thus their estimates cannot represent as reliable ones as those in the heavily polled battleground states, for example.

For polling error in terms of estimates of Clinton/Democratic and Trump/Republican shares of the vote, greater errors occurred for estimating Trump support. The average polling error was 1.83 points for Democratic vote based on polls, and a much larger 6.47 points off the mark for Republican vote.

Standard deviations provide a picture of how widely the polling errors varied. If that value is small, then polls were off by similar amounts across all states. If the value is large, then polls deviated from the actual result to much more different degrees. For Democratic margin, absolute margin, and Republican share of the vote, the HuffPost Pollster errors in poll estimates varied less than those from RCP. At the average level, it’s once again difficult to claim much without historical context, but a 5.75 standard deviation for the Democratic margin error is still considerable.

Finally, just to conclude with evaluating the nature the errors took themselves, here’s the distribution of the absolute polling margin errors (inattentive to the bias’s direction):

Data from HuffPost Pollster, RealClearPolitics, and U.S. Election Atlas.

It’s definitely not the case that there were large polling errors across all states, as a lot of state-level errors are in single digits in terms of their absolute value. However, once you start to move further right on this histogram and away from 1-3 point errors, a margin of error explanation proves unable to solve why errors of these magnitudes occurred. Instead, other more serious (and actually problematic) causes might have driven the errors aside from the sampling error intrinsic to public opinion polling.

Post-Election 2016 Part 1: Documenting the Polling Error

How 2016 Battleground Polls Look Relative To What Happened in ’08 and ’12

Here’s a quick overview of two different perspectives on the current 2016 polling data and how it relates to the past election/s. If Hillary Clinton defeats Donald Trump on Tuesday night, it will mark only the second time in 64 years that one political party will control the presidency for more than two terms. The graph below shows how Clinton is faring in the most important races–those in the battleground states that have been close in the last decade–as she tries to achieve this feat of topping off a third consecutive term in power for the Democratic Party.

Data from HuffPost Pollster and U.S. Election Atlas.

The race in 2008 was a (relative) landslide for Barack Obama. Even then, according to trend line averages from the HuffPost Pollster poll aggregation, Clinton’s current polling strength in North Carolina exceeds Obama’s small margin of victory in the state in 2008. Otherwise, the 2008 margin for Democrats is far above that in the two latter elections.

Comparing Obama’s actual margin of victory in 2012 to Clinton’s current polling position, the greatest disparities come from her relative underperformance in Iowa, Nevada, and Ohio. Virginia marks the only other state than North Carolina where Clinton has built on her predecessor’s success. The race is much tighter in a large majority of key battleground states this year for Clinton relative to earlier elections. That also lends credence to an important point Nate Silver made two days ago in explaining why his forecast model has proved so much less bullish on Clinton’s chances of winning the presidency.

Similar state-level results come up when comparing the final polling position of Obama in 2012 and now Clinton in 2016 (in battleground states). Note: the “final polling position” comes from the Democratic candidate margin based on Pollster’s trend line estimate/average.

Data from HuffPost Pollster.

The most meaningful examination will come when comparing these final polls to the actual 2016 vote, but still there are some interesting results here. Clinton is slated to do much better in Colorado than Obama was, even though Obama’s actual vote margin in the state–as seen in the first graph–came much closer to Clinton’s current polling position in 2016. A similar phenomenon seems to be occurring for Florida: Pollster’s trend line margin left Obama and Romney even (a margin of zero, explaining why no bar graph appears there), though Obama won by 0.88 percentage points–moving closer to Clinton’s current 1.7 point lead in Florida.

Unsurprisingly given the earlier analysis, Ohio, Iowa, and North Carolina show the biggest differences in the final state of the polls in 2012 and 2016. For the 2012 races in these states, the final poll trend line approximated the actual results fairly well. Perhaps that foreshadows Trump victories in Ohio and Iowa and a Clinton triumph in North Carolina. That’s at odds with my suspicion that Trump could retain North Carolina. While Pollster gives Clinton a 1.9 trend line lead, Trump and Clinton led two polls apiece and tied in another among the five most recent polls to be released. In the seven polls before that, Trump led three times, Clinton three as well, and the two candidates tied in another. Going back further gives you a more Clinton-friendly state of the race, but it’s much more of a mixed bag and up in the air at the moment than the 1.9 Clinton lead from Pollster would suggest.

How 2016 Battleground Polls Look Relative To What Happened in ’08 and ’12

Electoral Map 2016 Prediction and Thoughts

Just to have it on the record, and based on a rough mix of my interpretation of the course of this race, the current state of polls, demographic trends, historical state results, and early voting information*, here’s what I’m envisioning as the most likely electoral map to emerge on Tuesday night:

Created on

* A few of the key points of what went into the decision-making:

  1. Hispanics not only look like they will assume a greater share of the electorate based on early voting, but it seems they’re siding even more greatly with the Democratic Party and Hillary based on the highest quality polling data about their behavior. Considering where Hispanics make up the greatest shares of state electorates, that helps particularly in crucial states like Nevada and Florida, makes New Mexico fairly safe, though doesn’t quite tip Arizona into Clinton’s batch of victories–leading in a few stray polls doesn’t outweigh Trump’s much broader strength across the last months of polling.
  2. Again, based on early voting numbers, black turnout seems to be down in North Carolina, and some other states as well. That hurts most in NC however, and while the state is trending toward the Democratic column over the long run, the marked downturn in North Carolinian black turnout should throw the back-and-forth polling in the state to Trump.
  3. I think it’s reasonable to say that Trump’s relative strength and potential for more in the Upper/industrial Midwest of the United States has made this election competitive (see Nate Cohn’s great work and argument on this general topic). Here are the states I’d roughly consider part of this dynamic and where Trump stands in each right before the election (+/- indicating he’s up/down based on HuffPost Pollster numbers): Pennsylvania (-5.2), Wisconsin (-6.5), Michigan (-6.2), Minnesota (-6.7), Ohio (+0.9), and Iowa (+1.2). Given his Trump’s strengths with lower educated white voters and those who’ve had economic troubles perhaps tied to rising globalism (it’s a vague and not totally founded story, but that’s the general gist), his inability to perform better here–and the now very unlikely possibility of him stealing one of these states he’s losing in–is all but a death knell for his electoral fortunes.

Happy Election Day!

Electoral Map 2016 Prediction and Thoughts

Trends in Demographic Group Voting and How They’re Shaping Up in 2016

While demographics as a force in American politics are often overstated, the historical trajectories of how different subgroups vote merits a close look. Americans of certain races, educational level, age, and gender have differing roles in the coalitions of the two major parties in this country. When underlying demographic compositions start to shift, era-defining political realignments can take place (e.g., Southern whites flocking to the Republican Party in the 1960s). Given that demographics are still important in understanding the electorate, here’s a quick overview of how four key subgroups have voted since 1964, and based on polls in 2016, how they are primed to vote in one day (the 2016 poll numbers are in the yellow highlighted portions of the graphs).

1. Race 


White voters have predominantly favored the Republican candidate during this span of elections, except for in 1964 as part of Lyndon Johnson’s landslide victory, and in the two years Bill Clinton won the presidency, 1992 and 1996. Otherwise, whites have always been the best racial group in elections for Republicans. In 2016, there seems to be a small downturn in vote intention for the two major party candidates and a corresponding uptick in support for “Other” candidates. That makes sense given that the strongest third party candidate, Libertarian Party nominee Gary Johnson, does best among white voters.

African-Americans have always overwhelmingly gone for the Democratic candidate on the ballot, and that unsurprisingly will continue in 2016 given recent polls. Hispanic vote intention and its historical context is harder to suss out. For one, I couldn’t include Hispanic vote intention for election years before 2008 in this graph because of a sample size issue, as only in 2008 and 2012 did the ANES include an oversample of this group to overcome this issue. Secondly, this shows a downturn in Hispanic vote share going to Clinton and rise in that going to Trump. That stands at odds with perhaps the best polling data on Hispanics at the moment, which comes from Latino Decisions and most recently has Clinton at 74 percent and Trump at 14 percent among Hispanics–a good amount off from what you see in the graph. It’s probably best to trust Latino Decisions for the closest read on Hispanic voting behavior, and they’ve been insistent that Trump will not overperform Romney’s share of the Hispanic vote–as the above graph would indicate. There are several problems most other polling firms don’t navigate as well but Latino Decisions is likely better equipped to handle, such as survey interview language, callbacks on surveys, and assumptions about the share Hispanics will make up in the electorate. For these reasons and more, getting an accurate read on Hispanic voter preferences has been plagued by ambiguity both in the general and primary election seasons.

2. Age


Drawing meaning from over time trends for vote choice by age group gets complicated because of the changing age cohorts that make up these groups. Given that disclaimer, the above graph seems to hint at a possibly growing split in voter preferences by age. Clinton, for all the worries about her connection to and Democratic primary deficiency with young voters, appears to be building on Democratic Party strength with them–the 18-29 age group–in 2016 relative to earlier years. With Trump at the helm of his party, there’s an even larger difference relative to 2012, but in the negative direction, as his candidacy has pushed younger voters even further away from the GOP. Conversely, in the oldest age group (65+), Trump has seen an uptick in his party’s support from the oldest voters, while Clinton has seen a slight drop among this group.

One random piece of interest here: the graph also shows that the youngest age group at the time fueled the support behind the most successful third party candidate in recent decades, Ross Perot in 1992. It serves as a reminder of the interesting tidbit that younger voters have traditionally been most likely to support third party candidates, even going back to George Wallace’s candidacy in 1968.

3. Gender


Around 1988, vote choice among male voters became a lot more divided between Democratic and Republican candidates while women began to decidedly vouch for Democrats. According to recent 2016 polling data, women don’t seem any more likely than in recent elections to vote for the Democratic candidate. It should noted that a Fox Poll included in the average (see below section for details) underestimated the female vote for Hillary given what other polls showed. This poll is making the trend more stable than what could otherwise easily be a rise (at the very least) in female support for the Democratic candidate.

The same set of polls examined, however, shows that male voters are breaking a lot more for Trump in 2016 than in recent elections. The Republican candidate could reach a recent high point in 2004 for male support behind the GOP. Given both the trends and current states of vote choice among men and women, gender is likely to prove an equally great if not greater factor in this election than in past ones.

4. Education (whites only)


Due to such a strong degree of Democratic Party support from blacks and to a lesser extent Hispanics, parsing through vote choice numbers broken up by education becomes difficult. (The same issue comes about when using income as well.) It is thus important to distinguish by race when looking at socioeconomic variables. Accordingly, many poll releases include data for vote choice by education only among white voters in their crosstabs. For this reason–i.e., they don’t show vote choice by education for blacks and Hispanics because of small samples for these groups–as well as sample size considerations working with minority groups in ANES data, I only show the vote choice split between college-educated and non-college-educated voters among whites.

Electoral vote choice in the election is becoming more divided on several demographic frontiers, but none more so than on educational level among white voters. At essentially every point in this 2016 election cycle, Trump fared much better among voters without a college degree than those with one. This determinant first was acknowledged during the GOP primary, then held true near the start of the general election campaign, and still remains sharply evident one day from Election Day. Non-college-educated whites are moving in large numbers away from the Democratic Party and Clinton and in large numbers to the Republican Party and Trump. In some sense, this movement explains why Trump has remained competitive in this race. At the same time, Trump is also poised to become the first Republican candidate to ever lose among college-educated whites (at least in the history of the ANES).

In sum, difference in race, gender, age, and now increasingly education have all combined to drive an even greater wedge in the American electorate. Perhaps that contributes to the common feeling that the 2016 election represents some sort of inflection point in this country’s political history. In less than two days, the political landscape will become much clearer, but will assuredly remain rife with division.

Notes on the data: Percentages for 1964-2012 were weighted averages taken from the American National Election Studies. For the 2016 data, there were generally three steps I took to get a sense of how things were shaping up in 2016 based on public opinion polls.

1) I looked at about the 5-10 most recent polls that had publicly available data to inform the four subgroups I examined. The following polls were included in one way or another (i.e., not all were included in every subgroup vote calculation): NBC/WSJ (11/3-11/5), ABC/Post (11/1-11/4), McClatchy/Marist (11/1-11/3), FOX (11/1-11/3), YouGov/Economist (10/30-11/1), and Politico/Morning Consult (10/29-10/30).

2) For the data relevant to ascertaining vote share for the four subgroups, I took percentage vote intention for Hillary Clinton, Donald Trump, and “Other”–Gary Johnson, Jill Stein, and “Other”–and recalculated vote share percentages to exclude the following option percentages from the denominator of the vote share calculation: “Undecided,” “Neither,” and “Not sure.” See here for my reasoning behind this decision. That should explain why vote shares for many of these subgroups will appear higher than if you just look at the given total percentages in polls.

3) Finally, after step 2, I averaged out the resulting vote shares between the two-way and four-way questions (because there are different percentages depending on whether two or four candidates were named when poll respondents were asked about their vote choice). In cases where only four-way percentages were offered, I made an average subtraction from their denominator based on the difference between two-way and four-way questions in other polls for the same question topic.

Trends in Demographic Group Voting and How They’re Shaping Up in 2016

The Strength of Major Party Vote Intention by State

In an era of an increasingly strong attachment to parties and increasing numbers of people who feel this way, vote share for the two major political parties in U.S. elections has understandably risen. As an example, major party vote–defined as the sum of the Democratic and Republican candidate vote shares (or vote intention shares in polls)–in states during the 2012 election reinforced this idea. Four states saw their population vote for major party candidates a combined 96 percent of the time, nine states were at 97 percent, 22 states at 98 percent, 14 states at 99 percent, and one state (Oklahoma) saw its citizens give 100 percent of their to either Obama or Romney.

At least at this point in the 2016 election given the state of the polls, that same level of major party vote has dropped. This development can be attributed to two main sources: 1) the unusually higher number of undecided voters at this point in the race and 2) the unusually higher support for third party candidates. Intended votes for Gary Johnson and Jill Stein have declined over the last few months, but the two remained poised to exceed their success from 2012. Independent candidate Evan McMullin’s rise in Utah has contributed an even bigger wrinkle in this dynamic. The undecided and third party vote percentages vary across different states,  but overall represent valuable opportunities for Hillary Clinton and Donald Trump to collect more support in a tightening race just days removed from Election Day. The states where the major parties have the weakest consolidated vote also have the greatest possibility of late-term shifts that could alter the outcome.

The below two maps speak to this issue, with the first here showing where the Democratic and Republican Parties have most locked down the vote. The more purple a state is, the greater the major party vote is; yellow indicates where this vote is smallest.

Data from HuffPost Pollster.

Unsurprisingly, the state in which a third party candidate has the highest support in the nation stands out vividly. With McMullin currently polling at 22 percent in his home state, Utah currently stands far and away as the state with the lowest major party vote with just 62 percent going to Clinton and Trump. Utah represents more of an outlier–but no less significant in what is shows because of that status–and thus changes the color distribution where it’s hard to clearly see what other states have low major party support. The below map shows the same dynamic but excludes Utah from the color gradient calculation and map.

Data from HuffPost Pollster.

Here we see a lot more variation in major party support in the U.S. at this moment in the election. The five states with the highest major party intended vote are as follows:

  1. Kentucky (96 percent)
  2. Alabama (94 percent)
  3. Mississippi (93 percent)
  4. North Dakota (93 percent)
  5. Florida (92 percent)

Many of these states are in the South, where Republican Party attachment is strong with the majority population (white voters) and where Democratic Party attachemnt is even stronger but with a minority population (African-American voters).

Interestingly, the fifth ranked state here in Florida also claims the status of one of the foremost battleground states in the country, currently holding the highest chance of being the “tipping point” state in the election (the state whose electoral votes puts the winner past the threshold). Much less indecision remains in The Sunshine State, and so turnout rates will matter more than the few minds that are yet to have been made up.

The five states with the lowest major party intended vote are the following ones:

  1. Vermont (74 percent)
  2. Idaho (76 percent)
  3. New Mexico (76 percent)
  4. Alaska (79 percent)
  5. Washington (79 percent)

As I mentioned in an earlier post, the remaining effect and animosity generated from the primary season may be affecting matters in the general election. There’s a chance that may be playing out here as well. In the two traditionally Republican-leaning states listed, Idaho and Alaska, Trump lost his primary battles. Similarly, in two of the three Demcoratic-heavy states, Vermont and Washington, Clinton faltered during the primary season and suffered her most resounding defeat in Vermont–the state now showing the lowest signs of major party support. Among the two sets of states, the major party vote seems to be low due to the dominant partisan consituencies not (yet) falling back on their party nominees.

The national average of major party support in all states currently stands at 85 percent, a far cry from the 98 percent average in the actual results from the 2012 election. The comparison between polls three days out from the election and the election results themselves may exaggerate this difference a bit, but it’s fairly clear we’re going to see some of the highest non-major party support in recent decades come November 8th. At the same time, it also make for an interesting test for the well-documented partisanship strength among Americans these days–and whether the most important tie in American politics can reel these voters in to their respective parties by Election Day.

The Strength of Major Party Vote Intention by State

Changes in the Electoral College Map 5 Days Before Election Day

Over the last decade or so, the electoral college map has become increasingly fixed. States are less likely to flip their support from a Democratic candidate to a Republican one, or vice versa, from election to election. However, with the changing composition in the bases of support for the 2016 election candidates–from how minority groups are allocating their vote to the educational divide that has defined much of the support behind Hillary Clinton and Donald Trump–it’s worth looking what might be developing in terms of each state’s voting pattern. The below map gives a current sense of these changes relative to the 2012 election.

Data from U.S. Election Atlas and HuffPost Pollster.

In this map, each state’s hue is based on the difference between Clinton’s current polling margin and Barack Obama’s margin in 2012. Both margins come from Democratic Party vote share minus Republican Party vote share. The colors range from blue, which represents where Clinton most overperforms Obama’s 2012 result, to red, where she most underperforms his margin.

Utah is excluded (colored in grey) because of how the candidacy of Evan McMullin has complicated things. Subtracting Trump’s vote share in polls–currently at 36 percent–from Clinton’s vote share–26 percent–would lead to the misleading conclusion that Clinton has drastically improved on Obama’s performance in the state of Utah (and thus the most blue color on the map). In reality, she has currently culled just one more percentage point than Obama in 2012, who gained 25 percent of the vote in Utah. Margins work as a good indicator of strength when there isn’t a formidable third party presence, such as with McMullin, who’s currently at 22 percent in Utah, and when there isn’t a comparison between years where only one sees this type of third party success.

Clinton is currently beating Obama’s state margins most in Idaho (+12), Oklahoma (+11), Kansas (+10), and Texas (+10). That’s a bit misleading, however, since it has more to do with Trump underperforming Mitt Romney’s 2012 support in those states. But there are some noteworthy results here in that her strength relative to 2012 in Texas and Arizona comports with the notion of possibly rising Hispanic shares of state electorates and support for Democratic candidates. She also seems to be consolidating support well in certain parts of the South, such as in Alabama, Georgia, and South Carolina.

On the other end, Clinton is currently most underperforming Obama’s 2012 margin in Hawaii (-18), Rhode Island (-16), Iowa (-9), and Vermont (-8). For a lot of these states, Clinton is failing to reach Obama’s level of support (rather than Trump gathering more support than Romney in 2012). Interestingly, among the top four here, three are states she lost in the Democratic Primary. Her four best states listed above–and thus Trump’s four worst states relative to Romney’s 2012 margin–also all represent states lost by Trump in the Republican Party. Perhaps there are lingering effects in certain states of a hotly contested primary season for both parties. Notably, the sets of ID/OK/KS/TX and HI/RI/IA/VT represent strongholds for the Republican and Democratic Parties, respectively, in the last few elections. That these now mark where the greatest change is occurring suggests the lack of the usual strong acceptance of party nominees across all states (and partisans).

These potential changes of course become a lot clearer and more concrete in a matter of days. There remains a sizable non-major party vote in most if not all states, and that could be contributing to some of the more surprising processes at work. Democrats and Republicans could come home to their parties in larger numbers in the days to come, as we should expect given the high levels of partisanship in the U.S. Nevertheless, the map here should provide some sense of where any electoral college changes could materialize.

Changes in the Electoral College Map 5 Days Before Election Day