Buttigieg's rise has cost Biden and Warren the most early-state support
Buttigieg has appealed to voters nervous about Warren's liberalism and Biden's age
The takeaway: As measured by 2020 primary polling, the rise of Pete Buttigieg has cost Joe Biden and Elizabeth Warren the most support. Buttigieg has ticked upward while fortunes for the likes of Bernie Sanders and minor candidates have also improved, indicating that he might not be the only reason for their decline, however.
(PS: Anyone else spending their Saturday coding and playing the new Pokemon game? Bueller? Bueller? PPS, here’s the code for this analysis...)
Editor’s note:
Thanks for reading my thoughts on this subject. And thanks for subscribing! Your membership adds up and makes all this newslettering possible (reminder: I do all this work independently). Please consider sharing online or with a friend; the more readers, the merrier!
As always, send me your tips about what you’d like to read about next. Or what you don’t want to read. Or your feedback otherwise. I’m elliott@thecrosstab.com, or @gelliottmorris on Twitter.
Thanks all,
—Elliott
Pundits (at least the ones I talk to) have been captivated by Pete Buttigieg’s improving primary prospects over the past few weeks. By their reckoning, he is emerging as an alternative to Joe Biden, the former Vice President and current 2020 Democratic front-runner.
This is mostly believable, I think. Buttigieg and Biden are both moderates, at least when compared to the crop of other Democratic candidates. The media have characterized both as candidates who could be successors to Obama’s presidency.
If Buttigieg can prove himself a suitable, “electable” alternative to Biden in Iowa, he will surely rise, some analysts say. Of course, down the line, he will have to prove his ability to compete in more diverse states—something he has failed to do so far. For now, let’s set aside his weakness with non-whites and focus on Iowa.
The main question I’m interested in is whether Buttigieg’s rise has actually come at a cost to Biden. If so, we might have some empirical evidence that he is emerging as a moderate alternative to the former vice president. I will look at national polls as well as those in Iowa and New Hampshire for evidence.
To begin answering this question, I calculated averages of each candidate’s polls in Iowa, New Hampshire and nationally. On any given day, their average depends on an average of the previous polling in the state, weighted for both sample size and recency (I punish old polls using an exponentially decaying weight). Those averages look like this:

We can derive some insights by exploring patterns in the data visually. At first glance, Buttigieg’s rise in Iowa has come while Biden has fallen there. The same thing has happened in New Hampshire, but for Warren as well as Biden. Indeed, All of the leading candidates besides Buttigieg have fallen in Iowa over the last month.
This would normally be enough evidence for, say, a tweet, or a paragraph in a story, but I really want some hard statistical data this time. We can use something called a variance-covariance matrix to tell us whether Buttigieg’s poll numbers are positively or negatively correlated with other candidate’s. But before running that analysis, let’s plot those relationships on a graph:

Again, it looks like Buttigieg’s rise has cost Warren in New Hampshire and Biden in Iowa. But it doesn’t look like averages for Sanders, Harris or other candidates have also fallen. Indeed, the numbers for other candidates have seemingly improved alongside Buttigieg’s.
Okay, but what about this fancy variance-covariance stuff? Does it confirm or deny the visual analysis? Below, I’ve put the relationship between Buttigieg’s averages and each other candidate’s averages in every state into a table. Positive covariances mean that an increase in Buttigieg’s numbers has paralleled increases in the other candidate’s numbers. Negative numbers mean the opposite—that the other candidate has suffered when Buttigieg’s numbers have increased. This table contains covariances only for the past month of polling.

Notable in the table is that Biden and Warren have fared especially poorly while Buttigieg has risen.
However, these data themselves don’t mean that Buttigieg has explicitly cost Warren and Biden support. That’s a causal claim, but right now, we only have correlational data.
But importantly, we see that Buttigieg hasn't gained support from supports of candidates outside the top five. That’s represented by his positive and null covariance with the averages for Democrats who support other (or no) other candidates.
In other words, if Buttigieg hasn’t taken supporters from contenders like Andrew Yang, Beto O’Rourke, etc, then he must have taken support away from someone in the top 5. And based on these data, we think Biden and Warren are the most likely suppliers of these new Buttigieg voters.
I will save some speculation about why Buttigieg has cost Warren and Biden support for another time. This quick post has already gone on too long.
To really be sure about who has suffered at the hands of Mayor Pete, we’d need data on preferences of individual voters. Lucky for you, dear reader, CBS News—which has been tracking the same voters over the course of the primary—will be releasing new estimates tomorrow morning. I expect to have some coverage of these data, either here or on Twitter.
Happy Saturday.
—E