Mohamed A. Hussein
@mhusseinlab.bsky.social
Assistant Professor at Columbia. I study the psychology of persuasion, politics, and the intersection of the two. Ph.D. Stanford.
Will be sure to share the results as they become available!
September 15, 2025 at 7:16 PM
Will be sure to share the results as they become available!
The paper is now out, and you can read it here: authors.elsevier.com/a/1lhYz51f8w...
This is joint work with Zak Tormala and Christian Wheeler at Stanford.
This is joint work with Zak Tormala and Christian Wheeler at Stanford.
September 1, 2025 at 2:34 PM
The paper is now out, and you can read it here: authors.elsevier.com/a/1lhYz51f8w...
This is joint work with Zak Tormala and Christian Wheeler at Stanford.
This is joint work with Zak Tormala and Christian Wheeler at Stanford.
We see the DV of choice of extreme candidates as an understudied one in psychology. I hope we see more research on it.
We also think that studying how people assess whether a candidate is extreme or moderate would be an exciting future direction.
We also think that studying how people assess whether a candidate is extreme or moderate would be an exciting future direction.
September 1, 2025 at 2:34 PM
We see the DV of choice of extreme candidates as an understudied one in psychology. I hope we see more research on it.
We also think that studying how people assess whether a candidate is extreme or moderate would be an exciting future direction.
We also think that studying how people assess whether a candidate is extreme or moderate would be an exciting future direction.
This effect was robust to …
different descriptions of extreme candidates
👉different issues
👉controlling for other attitude dimensions (e.g., certainty, importance, moralization, knowledge).
👉Different methods (e.g., conjoint, vignettes, human-LLM interactions)
different descriptions of extreme candidates
👉different issues
👉controlling for other attitude dimensions (e.g., certainty, importance, moralization, knowledge).
👉Different methods (e.g., conjoint, vignettes, human-LLM interactions)
September 1, 2025 at 2:34 PM
This effect was robust to …
different descriptions of extreme candidates
👉different issues
👉controlling for other attitude dimensions (e.g., certainty, importance, moralization, knowledge).
👉Different methods (e.g., conjoint, vignettes, human-LLM interactions)
different descriptions of extreme candidates
👉different issues
👉controlling for other attitude dimensions (e.g., certainty, importance, moralization, knowledge).
👉Different methods (e.g., conjoint, vignettes, human-LLM interactions)
Is this just about group identity? Unlikely.
In another study, we used LLMs. They either prompted Ps to reflect on their views, or to connect those views to their identity.
When views were tied to identity, attitudes grew more extreme and so did support for extreme candidates.
In another study, we used LLMs. They either prompted Ps to reflect on their views, or to connect those views to their identity.
When views were tied to identity, attitudes grew more extreme and so did support for extreme candidates.
September 1, 2025 at 2:34 PM
Is this just about group identity? Unlikely.
In another study, we used LLMs. They either prompted Ps to reflect on their views, or to connect those views to their identity.
When views were tied to identity, attitudes grew more extreme and so did support for extreme candidates.
In another study, we used LLMs. They either prompted Ps to reflect on their views, or to connect those views to their identity.
When views were tied to identity, attitudes grew more extreme and so did support for extreme candidates.
The effect held even on issues people knew nothing about.
Saying John has a view on abortion doesn’t tell you if he’s pro-life or -choice.
So we made up an issue (“Prop DW”). Party had/no stance. That alone made it feel identity-relevant, pushing ppl to more extreme candidates.
Saying John has a view on abortion doesn’t tell you if he’s pro-life or -choice.
So we made up an issue (“Prop DW”). Party had/no stance. That alone made it feel identity-relevant, pushing ppl to more extreme candidates.
September 1, 2025 at 2:34 PM
The effect held even on issues people knew nothing about.
Saying John has a view on abortion doesn’t tell you if he’s pro-life or -choice.
So we made up an issue (“Prop DW”). Party had/no stance. That alone made it feel identity-relevant, pushing ppl to more extreme candidates.
Saying John has a view on abortion doesn’t tell you if he’s pro-life or -choice.
So we made up an issue (“Prop DW”). Party had/no stance. That alone made it feel identity-relevant, pushing ppl to more extreme candidates.
In a Conjoint study, we had people choose between different candidates (different ages, backgrounds, views on social issues).
We measured people’s identity relevance.
As identity relevance increased, people became more likely to choose the candidate who is extreme.
We measured people’s identity relevance.
As identity relevance increased, people became more likely to choose the candidate who is extreme.
September 1, 2025 at 2:34 PM
In a Conjoint study, we had people choose between different candidates (different ages, backgrounds, views on social issues).
We measured people’s identity relevance.
As identity relevance increased, people became more likely to choose the candidate who is extreme.
We measured people’s identity relevance.
As identity relevance increased, people became more likely to choose the candidate who is extreme.
Across six studies, we find that as people’s opinions on political issues become more part of their identity, they are drawn to extreme (vs. moderate) candidates.
September 1, 2025 at 2:34 PM
Across six studies, we find that as people’s opinions on political issues become more part of their identity, they are drawn to extreme (vs. moderate) candidates.
What is identity relevance?
It’s the degree to which your view on an issue feels like a reflection of who you are.
For some, views on climate change are core to identity.
For others, they may have strong views, but those views don’t define them.
It’s the degree to which your view on an issue feels like a reflection of who you are.
For some, views on climate change are core to identity.
For others, they may have strong views, but those views don’t define them.
September 1, 2025 at 2:34 PM
What is identity relevance?
It’s the degree to which your view on an issue feels like a reflection of who you are.
For some, views on climate change are core to identity.
For others, they may have strong views, but those views don’t define them.
It’s the degree to which your view on an issue feels like a reflection of who you are.
For some, views on climate change are core to identity.
For others, they may have strong views, but those views don’t define them.
Past work has focused on structural factors (e.g., primary elections, changes in supply of candidates).
In a new paper, we shift the conversation to *psychological* factors.
We test if the *identity relevance* of people’s attitudes cause them to choose extreme candidates.
In a new paper, we shift the conversation to *psychological* factors.
We test if the *identity relevance* of people’s attitudes cause them to choose extreme candidates.
September 1, 2025 at 2:34 PM
Past work has focused on structural factors (e.g., primary elections, changes in supply of candidates).
In a new paper, we shift the conversation to *psychological* factors.
We test if the *identity relevance* of people’s attitudes cause them to choose extreme candidates.
In a new paper, we shift the conversation to *psychological* factors.
We test if the *identity relevance* of people’s attitudes cause them to choose extreme candidates.
What a great read. Thanks for sharing!
June 24, 2025 at 2:24 PM
What a great read. Thanks for sharing!
Reposted by Mohamed A. Hussein
At the CredibilityLab (currently hosting Aspredicted and Researchbox) we have a new platform in the works, AsCollected, that will help with this. We welcome input from experienced parties.
Signup for alpha or beta testing or announcement of release at AsCollected.Org
Signup for alpha or beta testing or announcement of release at AsCollected.Org
AsCollected - Coming Soon
AsCollected.Org
May 20, 2025 at 1:12 AM
At the CredibilityLab (currently hosting Aspredicted and Researchbox) we have a new platform in the works, AsCollected, that will help with this. We welcome input from experienced parties.
Signup for alpha or beta testing or announcement of release at AsCollected.Org
Signup for alpha or beta testing or announcement of release at AsCollected.Org