I made the case on Monday that while Sen. Elizabeth Warren (D-Mass.) made extraordinary improvement between the first of the year and the end of September, her numbers have been flat since October, and in some cases, the margin between her and other candidates has narrowed.

More data points released late Monday add some weight to this analysis. In the national Morning Consult Political Intelligence poll, both former vice president Joe Biden and Sen. Bernie Sanders (I-Vt.) picked up four points from the previous week in early state polling. Warren’s number did not move. That puts her comfortably in third place. (Biden has 33 percent, Sanders 22 and Warren 17.)

If you turn to state polls, a new poll from the Nevada Independent shows: “Biden was backed by 29 percent of likely caucusgoers in the poll, while Warren and Sanders were each favored by 19 percent of respondents.” Nevada voters are much more diverse than New Hampshire and Iowa, and that helps Biden quite a bit:

Biden tended to have higher favorability ratings among those who identified themselves as somewhat liberal, moderate, or conservative, women and those above the age of 40, with relatively even perceptions between non-college and college educated individuals. He also was viewed slightly more positively among those who identified as white or non-white than Hispanic and those who have caucused in the past.
Warren was viewed most favorably by liberals, women, those 40 and up, college educated individuals and those who identified themselves as white.

Even Sanders’s base is more diverse. (“Sanders had the highest favorability ratings among self-described liberals and with people between the ages of 18 and 39. He also was viewed more positively by non-college educated individuals and voters who identify themselves as Hispanic or non-white.”)

Nevada voters at this stage sound more pragmatic. “The top issue for respondents was electing a candidate who can beat Trump — which was roughly twice to three times as popular as the second top choice issue, supporting someone who can work with both parties.” Additionally, “An overwhelming plurality of respondents, between 31 and 33 percent, said that the most important issue for them when choosing a Democratic presidential nominee is who has the best chance to beat President Donald Trump.” A word of caution however: As in other polls, a high percentage (55 percent) say they could change their mind.

This tells us a few things about the race. First, South Bend, Ind., Mayor Pete Buttigieg, at 7 percent, takes in some of the white, college-educated voters Warren needs. (“The South Bend mayor had higher favorability ratings with college educated individuals and white voters.”) She will need to capture them and/or appeal to nonwhites and non-college-educated voters if she is going to compete, especially as the race moves from homogeneous states such as New Hampshire and Iowa to states that look more like the Democratic Party as a whole (e.g., Nevada, South Carolina, southern Super Tuesday states). Second, “momentum” and “surges” can subside and shift as voters look more carefully at candidates and as they reveal personal and policy strengths and weaknesses. What was true in September may look very different in November.

It is a truism that the presidential primary is a marathon, not a sprint. The field is too big for a candidate to rely purely on one type of voter (male/female, white/nonwhite, college/non-college-educated). Unlike the GOP, which is extraordinarily monochromatic and dominated by evangelical Christians, the Democratic Party requires coalition-building — an exciting progressive who nevertheless does not scare moderates, a candidate who draws on older, reliable voters but also finds some support with younger voters, etc.

Until such a coalition-builder pops up, the race will be unsettled and segmented by these subgroups of voters. The one who can string a bunch of them together likely will win the nomination.

Read more: