The  Congress 2018 column I published in early April looked at how the Democrats and Republicans are doing in the race for Congress this fall through a question entitled the National Generic Congressional Ballot: Are you planning to vote Democrat or Republican for the House of Representatives this fall? The column compared the results of four late March 2018 polls to the Results of Exit Polls for the last non-Presidential election year race for Congress, 2014. The 2014 Election Exit Poll reported that Americans who voted in House elections that year voted 53% Republican to 47% Democratic in a straight-match-up between the two parties.

The four late March 2018 polls, from Quinnipiac, The Economist, Marist, and Fox News, showed 5-6 point leads for the Democrats, with about 15% of the electorate still undecided or claiming they would vote for another political party. If the undecideds broke in the same proportion as those voters making a choice at that time, the Democrats would be close to a seven-point lead.

Nate Silver, the national polling analyst who founded, an online publication focused on politics and polling, has said the Democrats need to win the National Congressional vote by seven full percentage points against the Republicans (in the results, not just the polls!) to have a break even chance of taking the majority in the House of Representatives, given the distribution of the electorate across the country among the individual 435 districts. That’s what I call the ” 7% ” solution.

The Democrats still hold a lead in early May 2018 polls on the National Generic Congressional ballot question, but it’s smaller. (The National Generic Congressional ballot question, by the way, and most other polls are updated on a daily basis on the fivethirtyeight website. )

As we all know, polls are uncertain animals and a look at these early May polls exposes some of the shortcomings of polling, but still demonstrates the potential for a Democratic victory, much of which will ultimately depend on the intensity of the campaigns in the individual districts. The starting point for analysis remains a comparison with what happened in 2014, and then moving on to the new polls.The 2014 Exit Polls showed the following results:


The four early May Congressional ballot polls show the Democrats with two five-point leads one three point lead, and one with six points*. This graph compares these results with the 2014 Exit Polls.

CONGRESS 2018 May Update Overall-page-001

CNN claimed that its poll, with the Democrats holding a small three point lead, showed the Republicans were closing the gap from 2017, when many polls were showing the Democrats with leads of ten points or more. Two other polls, Morning Consult and The Economist/You Gov, show five-point leads for the Democrats. The Rasmussen six-point  Democratic lead is interesting because Rasmussen polls a different segment of the electorate, likely voters, compared to the others, which poll all registered voters. Likely voters are defined as a group with a regular voting history in non-Presidential year elections. Rasmussen also uses this group for its Daily Presidential approval rating, which has shown Trump sometimes at 50% approval, higher than virtually all other polls. One explanation for this apparent contradiction is that voters may be frequently responding with approval for a specific Trump action, which may get reflected in the bumps up and down of the daily ratings, but still seek a check on him with a Democratic Congress.

This next graph compares a breakdown of Exit Polls for white voters with the new May 2018 polls for white voters:Congress 2018 May Update Whites-page-001

The Democrats lost the white vote 60-38 in 2014, according to the Exit Polls. The May 2018 Polls remain similar to the March 2018 polls- the four May polls show the Democrats holding an average of 39%, similar to how they did in 2014, while Republicans in the four polls average 46% among whites. The remainder are undecided or claim they would vote for another party. If the undecideds broke in the same proportion as the samples making choices now, the Democrats could reach 44% of the white vote, substantially better than they did in 2014.

The next graph compares Exit 2014 Polls for black voters with the results of three of the four May polls for black voters. CNN consolidated polling results for non-white voters and is shown in the graph.Congress 2018 May Update Blacks-page-001

The May 2018 polls for black voters show similar huge margins supporting the Democrats as 2014, although 2 out of 3 of the polls show slight improvements for the Republicans. Interpreting the results is difficult when polls reach only a few members of a group that is smaller than the sample as a whole. A poll of 1000 voters might have 120 black voters – similar to their share of the electorate- but a small sample of 120 could have a margin of error of +- 10, meaning the result of the poll compared to the real population of that particular group could be ten points higher or lower than the result of the poll. The prior results from 3 March polls for black voters were 83-7, 73-15, and 86-6, with the lopsided margins for the Democrats. Volatility in poll results for small samples demonstrates some of the limitations of polling.

Here are the results of the four May polls for Hispanics and/or nonwhites.Other-page-001

Only two of the four May 2018 polls separate Hispanics in the electorate; the other two group either group all nonwhites together or group Hispanics with other nonwhites except blacks. All four show the Democrats holding substantial leads among Hispanics and among nonwhite voters in general, within range of the 2014 results, but once again these results come from smaller samples. Three March polls among Hispanics had results of 64-30, 46-23, and 43-31, somewhat similar to the May polls. Regardless of these poll results, Hispanics were only a 7% share of the electorate in 2014- Democrats must boost the number of Hispanic voters substantially in 2018 to improve their chances of winning a majority, as well as win a share of their vote similar to 2014. Hispanics were 11% of the electorate in the Presidential election, compared to 7% in 2014.

*- Three of the polls reported updates on May 23rd. The Economist reported a Democratic margin of 5 again, 43-38. Morning Consult reported a 6 point lead for the Democrats, 42-36, while the lead for Democrats in Rasmussen closed to one point, 43-42.

The Democrats maintain a small lead in the National Generic Congressional Ballot, but a large share of the population remains unsure of who they will vote for. Right now it seems that if the undecideds broke in the same proportion as those who had made choices in these polls, the Democrats would not quite gain the seven point margin. But the volatility of the Black, Hispanic, and other minority group poll results create some uncertainty in the direction of the polls. The Democrats are competing heavily in many more districts this year than they were in 2014, and still have an intensity that should increase their turnout.


The flow of polling across the country about politics includes a question regularly asked of the electorate called the Generic Congressional Ballot. It simply asks how do you plan to vote in the upcoming 2018 election for the House of Representatives­—Democrat or Republican? With seven months to go before the November election, this column looks at how the Democrats are doing on this national question by reviewing four recent polls and comparing them to what actually happened in the 2014 House elections. Looking at what really happened in the previous midterm election is a good reality test in thinking about what could happen in the next midterm. Generic Congressional ballot polls are challenging in their predictive value: after all, they are just snapshots in time with tiny samples of the huge American electorate. National congressional polls also do not provide data for the 435 individual Congressional districts as to how the electorate might vote in each of those districts.

Nate Silver, the statistician turned national polling expert with his own online analytical firm,, also provides a general rule of thumb for understanding the meaning of a generic Congressional ballot poll. In an article he wrote for 538 after the November 2017 election, where the Democrats had done well in Virginia and New Jersey, he said, “Democrats also face a big disadvantage in the way their voters are distributed across Congressional districts, as a result of both gerrymandering and geographic self-sorting. Although these calculations can vary based on incumbency advantage and other factors, my back-of-the-envelope math suggests that Democrats would only be about even money to claim the House even if they won the popular vote for the House by seven percentage points next year.“ This means that polls showing Democratic margins of seven points or better (as long as they are accurate when it comes to the actual result), suggest that Democrats have the potential to attain a Majority in the House.

Using the 2014 House elections as a starting point to compare current polls shows the challenge for the Democrats to win the Majority. Before that election, the Democrats held 201 seats and the Republicans held 234 as a result of the 2012 election, where the Democrats won eight seats in the year President Obama won re-election. 129 million people voted on the Presidential line that year, and 122 million votes were recorded as the sum of all the ballots in the House elections. But in 2014, the number of votes cast on the Congressional lines across the country dropped to 78 million, and the Democrats lost a net of 13 seats. The drop-off in the vote from the Presidential election to the off-year Congressional one is staggering.

The national Congressional popular vote in 2014 dropped by 44 million, more than one-third, compared to the votes on the Congressional line in 2012, from 122 million votes in 2012 to 78 million in 2014. The 2014 vote was also 8 million votes below the previous off-year Congressional election, 2010, when 86.7 million Americans voted (and the Democrats lost 63 seats). The 2014 vote was even below the 81 million who voted in 2006, eight years earlier. The Republicans won the 2014 national Congressional popular vote by six points, 51.2 to 45.5, with about three points for other candidates. On a straight Democrat vs. Republican, excluding votes for candidates of other parties, the Republicans won 53% to 47%.

This Exit Poll graph for the outcome of the 2014 Congressional election breaks down the vote both in overall numbers, as well as by racial and ethnic group.


The Republicans won 60% of the white vote, to 38% for Democrats. Blacks voted 89-10 Democratic and Hispanics voted 62-36 Democratic. Asians and others voted 50-49 Republican. National Exit Polls also reported that 75% of the 2014 electorate was white, 12% was black, 7% was Hispanic, and 5% was Asian or other.

With Donald Trump in office more than a year, Democrats are hopeful that his unpopularity and the traditional poor performance of the party that holds the presidency in the off-year will help them regain the majority in the House and either win the Senate or hold the line there. The Democrats need 24 seats to win back the majority. Whether it is galvanizing younger and minority voters, moving affluent suburbanites appalled by Trump to vote Democratic, or persuading ancestral white working class voters to come back to the Democrats, the Democrats have many messaging challenges to craft for these diverse groups. Local Democrats running in special elections for vacant seats have been running ahead of Hilary Clinton’s presidential election performance in those districts, sometimes by very large margins, and there are many Republican retirements, giving Democrats even more room for optimism.

Four polling firms completed national generic Congressional ballot question polls around the same time period in late March 2018. The firms are Quinnipiac, Fox News, Marist, and The Economist/You Gov. Here are their overall results compared to the 2014 Exit Poll.

Graph 2.PNG

The four polls are showing either 5 or 6 point Democratic leads. The results don’t add to 100 full points because some voters are undecided, and small numbers say they won’t vote or will vote for another party. All the polls are of registered voters, rather than all American adults, and generally had margins of error of +- 3 points. If the undecided voters split their votes along the same proportions as the voters who made choices in these polls, it would bring the Democratic margins up to 6-7 points, meaning the Democrats are just barely reaching the Nate Silver 7 point threshold he said the Democrats would need to have a breakeven chance to win the Majority.

Here are the 4 polls broken down by race and ethnicity:

Graph Blacks.PNG

For Black voters, the four recent polls show similar margins to how Black Americans voted in 2014; overwhelmingly Democratic, and likely to vote in the same range, 90-10 Democratic, as they did that year (an updated The Economist Poll this week had the Black electorate at 72-6 Democratic).


The results for Hispanic voters from the recent polls once again are similar to 2014; Hispanic voters are voting Democratic nearly 2-1. Fox News combined all nonwhite voters into one group in its release of data, but 69-21 is overwhelming when Blacks, Hispanics, and Asians are combined and are actually several points higher for those groups combined than 2014. With Hispanics seven per cent of the turnout in 2014, a major challenge for the Democrats will be to increase Hispanic turnout, which was 11% of the vote in the 2016 Presidential election.


The results for white voters for the four recent polls compared to 2014 shows the Democrats getting a range of 38-41%, meaning they are holding their base of the white vote from 2014, 38%. The results for the Republicans show a range of 45-51%, below the 60% of whites who voted Republican in 2014. Undecided voters ranged from 9-17%. But if Democrats could win 40% of the undecided white voters, they would clearly do much better than 2014.

Many districts have not yet even nominated the person who will be the Democratic or Republican candidate. Many intangibles like candidate quality, local energy, sudden retirements of incumbents, fund-raising advantages, and the like can affect individual districts outcomes regardless of the overall national situation. But in the last week of March 2018 current polling shows the Democrats tantalizingly near where they need to be to win.