The second wave of the CBS/YouGov Delegate Model was released this morning on Face the Nation. We are still six months away from any delegates actually being chosen, but the data do suggest how the race is shaping up.
We have a "top tier" of candidates polling in double digits: Joe Biden, Kamala Harris, Bernie Sanders, and Elizabeth Warren, with Pete Buttigieg just below that level. The remaining twenty or so candidates don't seem to be getting any traction with Democratic primary voters.
Between the first wave, conducted May 31-June 12, and the second, conducted July 9-18, there were a pair of debates on MSNBC featuring 20 of the two dozen or so major candidates. In the press following the debates, the performances of Harris and Warren were praised, while those of Biden and Sanders were panned. In our polls, the two women have gained at the expense of the two men, so, for now at least, it looks like a fairly tight four-way race.
Of course, a month ago it seemed like Joe Biden held a commanding lead and the race would soon be over. Voters change their minds. Which candidate is leading the horserace is less interesting than how voters are switching their preferences overs the candidates. Because our delegate tracking contains large numbers of reinterviews of the same voters, we are able to identify which types of voters actually switched.
A note of caution: this analysis is based only upon the 4,718 democratic and democratic-leaning voters who were interviewed in both the June and July waves. We reweighted the sample to match demographic and candidate preferences in the baseline wave. This eliminates the effects of differential recontact rates, but the numbers are slightly different from the full-sample results in the Delegate Model. Also, the sample differs from other polls because it is limited to the 18 early primary and caucus states, not Democrats nationally.
The biggest loser between the June and July surveys was Joe Biden, who dropped almost five points (from 30.0% to 25.5%). Bernie Sanders fell a bit less than two points (from 16.0% to 14.5%) as did Pete Buttigieg (from 7.9% to 6.2%). The biggest gainer was Kamala Harris, who picked up almost six points (from 10.7% to 16.5%). Elizabeth Warren picked up almost four points (from 17.2% to 20.9%). There was some winnowing of the remaining twenty or so remaining candidates, who dropped collectively by two points (from 18.3% to 16.3%).
Together, Harris and Warren gained ten points. Where did this gain come from? For Harris, the pickup came almost entirely from voters who previously preferred Biden or one of the second tier candidates (anyone other than the top five). Warren's smaller gain was distributed more evenly. She picked up about a point each from prior supporters of Biden, Buttigieg, and Sanders. She lost about half a point to Harris, but this was offset by a gain of the same size from second tier candidates.
|Preference||Preference in July|
Who were the switchers? We divided Democrats into three groups based upon their self-described ideology: very liberal, somewhat liberal, and moderate or conservative, which will be referred to as left, liberal, and moderate. Liberals are the largest group among Democratic primary voters and moderates the smallest.
First, Biden is losing support from the left. Only 60% of June Biden supporters in the left category still supported him in July. He retained over 70% of the support of liberals and moderates.
Sanders, on the other hand, lost more support from liberals and moderates than the left. 75% of June Sanders supporters from the left still supported him in July, compared to only 60% of liberals.
Finally, the support that Harris picked up from Warren was from liberals, not the left. 14% of Warren's liberal supporters switched to Warren versus 10% of Harris's liberal supporters going in the other direction (for a net loss of 4%). Among left supporters of Warren, only 8% switched to Harris, while 10% of Harris's supporters switched to Warren (for a net gain of 2%).
One metaphor for primary races is candidate "lanes," where, for example, left-wing candidates (like Sanders and Warren) compete for left-wing voters, while more moderate candidates (like Biden and Harris or Buttigieg) compete for moder moderate voters. As the data above show, some lane switching is indeed occurring. But one should not exaggerate. Most preference switching does not follow this pattern. Net switching is fairly small and not well predicted by ideology.
You can see details on the polling and methodology here: CBS/YouGov Delegate Model Results and Methodology - July 21, 2019