Democratic and Republican Vote Shares: 2016- 2020
| 2016 | 2018 | 2020 | |
| Protestant | D: 39 R: 58 Margin: -19 | D: 36 R: 62 Margin: -26 | D: 38 R: 61 Margin: -23 |
| Catholic | D: 45 R: 52 Margin: -7 | D: 49 R: 49 Margin: 0 | D: 49 R: 50 Margin: -1 |
| Mormon | D: 25 R: 61 Margin: -36 | D: 29 R: 67 Margin: -38 | D: 24 R: 71 Margin: -47 |
| Jewish | D: 71 R: 24 Margin: 47 | D: 71 R: 28 Margin: 43 | D: 68 R: 30 Margin: 38 |
| Muslim | D: 82 R: 15 Margin: 67 | D: 80 R: 17 Margin: 63 | D: 64 R: 35 Margin: 29 |
| Something Else | D: 62 R: 29 Margin: 33 | D: 65 R: 31 Margin: 34 | D: 61 R: 36 Margin: 25 |
| Non-Religious | D: 60 R: 26 Margin: 34 | D: 70 R: 27 Margin: 43 | D: 72 R: 26 Margin: 46 |
Data Sources
Most 2016 estimates provided by Edison Research Exit Polling, with the exception of Mormons and Muslims, for whom data was unavailable through Edison.
Estimates of Muslim 2016 vote preference derived from the Cooperative Congressional Election Study (CCES). Estimates of 2016 Mormon vote preference provided by Pew Research.
For latter years, estimates of vote preference for all faiths provided by AP VoteCast (2018, 2020), a nationally-representative survey of more than 110 thousand Americans, launched in 2018, which includes a more detailed breakdown of religious voters — including, for instance, Mormons and Muslims (whose voting behavior is not typically described in Edison Research exit polling).
Margins
The margins between Democrats and Republicans were derived by subtracting the Republican vote share from the Democratic vote share for a particular constituency in a given year.
Positive margins therefore favor Democrats on balance. Negative margins therefore favor Republicans. Zero represents parity between the two parties. Longitudinally speaking, positive margins which trend towards zero over time indicate a constituency shifting towards Republicans. Negative numbers shifting towards zero indicate a constituency moving towards Democrats.
Margins are a useful way of visualizing the trends over time, because they simultaneously account for changes in both the Republican and Democratic vote shares. Often these movements are not symmetrical: Republicans can lose more voters than Democrats gain (because people vote for third parties instead, for instance); or both parties can simultaneously gain voters (as third-party voters shift to the Democrats and Republicans, albeit often at different rates); or both parties can simultaneously lose voters (for instance, by suffering attrition to third parties, again, often at different rates). The margin incorporates any or all of these trends into a singular number.
That said, it is often illuminating to graph or analyze trends in vote share among Democrats and Republicans separately too (or instead), as I’ve done previously. But here, for simplicity’s sake, I focused on the margins.
Midterm v. Presidential Election Years
Typically, the electorates are very different for midterm years as compared to cycles in which a presidential election is being held. Fewer people tend to vote overall — and these reductions tend to be especially pronounced among many traditionally Democratic-leaning groups: young people, minorities, lower-income people, etc.
For this reason, it is sometimes difficult to directly compare trends across midterms and presidential years, and it can make sense to analyze or visualize them separately — as I have done previously — because it makes trends in the data easier to discern.
In 2018, however, turnout during the midterms reached a 50-year high — rendering the midterm electorate more comparable to that of presidential years than it would typically be.