David Broockman
@dbroockman.bsky.social
Day job = Associate Prof. of Political Science at UC Berkeley. Tweets = personal views.
That's what we mean by "brand"! :)
May 28, 2025 at 4:36 AM
That's what we mean by "brand"! :)
Will add this to our list to look at! We do have candidate fixed effects so our main specification shouldn't contain any bias from that.
May 27, 2025 at 10:32 PM
Will add this to our list to look at! We do have candidate fixed effects so our main specification shouldn't contain any bias from that.
Yes, you can see some of the partisan asymmetry in Fig 1 -- R candidates are much more right than D candidates are left. Ds are to the left of voters, still, but voters see the Ds as even more left than they are -- sign of how the national party brand needs to improve.
May 27, 2025 at 9:12 PM
Yes, you can see some of the partisan asymmetry in Fig 1 -- R candidates are much more right than D candidates are left. Ds are to the left of voters, still, but voters see the Ds as even more left than they are -- sign of how the national party brand needs to improve.
Thanks for your questions! Fig 2 has endorsement knowledge -- in primaries it's much higher than policy position knowledge. Endorsements often go to different cands; we should quantify this. And I'll see if we have data on timing!
May 27, 2025 at 9:11 PM
Thanks for your questions! Fig 2 has endorsement knowledge -- in primaries it's much higher than policy position knowledge. Endorsements often go to different cands; we should quantify this. And I'll see if we have data on timing!
I'd put my money down for a revised version of #1: moderation on issues where your party is out of step
May 27, 2025 at 3:40 PM
I'd put my money down for a revised version of #1: moderation on issues where your party is out of step
The broader point: Super-informed/extreme primary voters or uninformed general election voters are unlikely polarization’s main drivers. Groups & other factors studied in the literature are more likely.
Feedback welcome!
Full paper available here: osf.io/7xbza
Feedback welcome!
Full paper available here: osf.io/7xbza
OSF
osf.io
May 27, 2025 at 2:07 PM
The broader point: Super-informed/extreme primary voters or uninformed general election voters are unlikely polarization’s main drivers. Groups & other factors studied in the literature are more likely.
Feedback welcome!
Full paper available here: osf.io/7xbza
Feedback welcome!
Full paper available here: osf.io/7xbza
Some caveats: This is observational data from 27 districts in 2024. Voters might care about other things like compromise or ideology that we didn't study. For issue voting, projection is a threat to causal inference–but we discuss why that’s unlikely to explain our findings.
May 27, 2025 at 2:07 PM
Some caveats: This is observational data from 27 districts in 2024. Voters might care about other things like compromise or ideology that we didn't study. For issue voting, projection is a threat to causal inference–but we discuss why that’s unlikely to explain our findings.
Parties face a dilemma. Nominating moderates in close districts helps win—but most voters infer that candidates hold the party’s typical positions. To win some seats, a party needs a *nationwide* moderate reputation. But groups & others might not want to build it (e.g., by moderates in safe seats).
May 27, 2025 at 2:07 PM
Parties face a dilemma. Nominating moderates in close districts helps win—but most voters infer that candidates hold the party’s typical positions. To win some seats, a party needs a *nationwide* moderate reputation. But groups & others might not want to build it (e.g., by moderates in safe seats).
Summarizing: primary voters actually struggle to identify which candidates match their views. So they rely on other cues—especially from interest groups who might help drive polarization. But general election voters then *do* punish extremism.
May 27, 2025 at 2:07 PM
Summarizing: primary voters actually struggle to identify which candidates match their views. So they rely on other cues—especially from interest groups who might help drive polarization. But general election voters then *do* punish extremism.
FINDING #3: Evidence that groups contribute to polarization. Group endorsements have major influence. Voters who learn about them are ~15 pp more likely to vote for endorsed candidates. This effect is driven by endorsements from liked groups—negative cues barely register.
May 27, 2025 at 2:07 PM
FINDING #3: Evidence that groups contribute to polarization. Group endorsements have major influence. Voters who learn about them are ~15 pp more likely to vote for endorsed candidates. This effect is driven by endorsements from liked groups—negative cues barely register.
This creates a problem for primary voters: party cues are useless in primaries where everyone's the same party. So primary voters stay confused about who's closest to them, even as they learn what candidates of their party stand for.
May 27, 2025 at 2:07 PM
This creates a problem for primary voters: party cues are useless in primaries where everyone's the same party. So primary voters stay confused about who's closest to them, even as they learn what candidates of their party stand for.
We found evidence that voters learn about candidates partly by learning national party reputations: a) voters somehow learn just as much about no-name candidates, & b) voters are 3x more likely to learn "stereotypical" positions (like Dems supporting healthcare) than unusual ones
May 27, 2025 at 2:07 PM
We found evidence that voters learn about candidates partly by learning national party reputations: a) voters somehow learn just as much about no-name candidates, & b) voters are 3x more likely to learn "stereotypical" positions (like Dems supporting healthcare) than unusual ones
Why do general election voters know more? Party cues.
In general elections, party labels give voters huge clues about candidate positions. In primaries, all candidates have the same party label—so these cues are useless for choosing between them.
In general elections, party labels give voters huge clues about candidate positions. In primaries, all candidates have the same party label—so these cues are useless for choosing between them.
May 27, 2025 at 2:07 PM
Why do general election voters know more? Party cues.
In general elections, party labels give voters huge clues about candidate positions. In primaries, all candidates have the same party label—so these cues are useless for choosing between them.
In general elections, party labels give voters huge clues about candidate positions. In primaries, all candidates have the same party label—so these cues are useless for choosing between them.
BUT: b/c they know more, general voters much more successfully identify the closest candidate. They vote for the closest candidate on an issue 78% of the time; primary voters only do 18% of the time
The opposite of conventional wisdom about incompetent general voters & aware primary voters
The opposite of conventional wisdom about incompetent general voters & aware primary voters
May 27, 2025 at 2:07 PM
BUT: b/c they know more, general voters much more successfully identify the closest candidate. They vote for the closest candidate on an issue 78% of the time; primary voters only do 18% of the time
The opposite of conventional wisdom about incompetent general voters & aware primary voters
The opposite of conventional wisdom about incompetent general voters & aware primary voters
When primary AND general voters learn a candidate agrees with them on an issue, they're ~14 percentage points more likely to vote for that candidate.
In generals, this includes when the closest candidate is an outpartisan–party loyalty isn’t everything.
In generals, this includes when the closest candidate is an outpartisan–party loyalty isn’t everything.
May 27, 2025 at 2:07 PM
When primary AND general voters learn a candidate agrees with them on an issue, they're ~14 percentage points more likely to vote for that candidate.
In generals, this includes when the closest candidate is an outpartisan–party loyalty isn’t everything.
In generals, this includes when the closest candidate is an outpartisan–party loyalty isn’t everything.
FINDING #2: Using Lenz’s identification strategy and exploiting our panel data, we also estimate issue voting.
Both primary AND general election voters change their votes when they learn candidate positions—by a very similar amount…
Both primary AND general election voters change their votes when they learn candidate positions—by a very similar amount…
May 27, 2025 at 2:07 PM
FINDING #2: Using Lenz’s identification strategy and exploiting our panel data, we also estimate issue voting.
Both primary AND general election voters change their votes when they learn candidate positions—by a very similar amount…
Both primary AND general election voters change their votes when they learn candidate positions—by a very similar amount…
FINDING #1: General election voters know MORE about candidate positions than primary voters.
By election day, general election voters correctly identify 40% of candidate positions vs just 22% for primary voters.
By election day, general election voters correctly identify 40% of candidate positions vs just 22% for primary voters.
May 27, 2025 at 2:07 PM
FINDING #1: General election voters know MORE about candidate positions than primary voters.
By election day, general election voters correctly identify 40% of candidate positions vs just 22% for primary voters.
By election day, general election voters correctly identify 40% of candidate positions vs just 22% for primary voters.
This gives us a ton of unique data.
We measure knowledge & learning of 122 candidate issue positions in the 2024 Congressional primaries and 269 candidate issue positions in the 2024 Congressional generals.
We measure knowledge & learning of 122 candidate issue positions in the 2024 Congressional primaries and 269 candidate issue positions in the 2024 Congressional generals.
May 27, 2025 at 2:07 PM
This gives us a ton of unique data.
We measure knowledge & learning of 122 candidate issue positions in the 2024 Congressional primaries and 269 candidate issue positions in the 2024 Congressional generals.
We measure knowledge & learning of 122 candidate issue positions in the 2024 Congressional primaries and 269 candidate issue positions in the 2024 Congressional generals.
We monitored debates, websites, & news in real time during the 2024 primary and general elections to identify issues & endorsements where the candidates *in each district* differed. We then conducted pre-election surveys 2-3 mo before each election & re-surveyed again on e-day.
May 27, 2025 at 2:07 PM
We monitored debates, websites, & news in real time during the 2024 primary and general elections to identify issues & endorsements where the candidates *in each district* differed. We then conducted pre-election surveys 2-3 mo before each election & re-surveyed again on e-day.
Conventional wisdom blames:
• Primary voters who closely follow politics & prefer extremists
• General election voters who are too ignorant of candidate positions—or too “intoxicated” by party loyalty—to vote for moderates over extremists
But our data tells a different story…
• Primary voters who closely follow politics & prefer extremists
• General election voters who are too ignorant of candidate positions—or too “intoxicated” by party loyalty—to vote for moderates over extremists
But our data tells a different story…
May 27, 2025 at 2:07 PM
Conventional wisdom blames:
• Primary voters who closely follow politics & prefer extremists
• General election voters who are too ignorant of candidate positions—or too “intoxicated” by party loyalty—to vote for moderates over extremists
But our data tells a different story…
• Primary voters who closely follow politics & prefer extremists
• General election voters who are too ignorant of candidate positions—or too “intoxicated” by party loyalty—to vote for moderates over extremists
But our data tells a different story…