Nicolai Berk
@nicolaiberk.bsky.social
Postdoc at ETH Public Policy Group & Immigration Policy Lab. Political Communication, Public Opinion, NLP.
Very excited to announce the next term of the TADA speaker series!
This term revolves around the validity of LLM generated data in the Social Sciences. We have a set of four great speakers to enlighten us on different aspects of this issue.
If you'd like to join, sign up here: TADA.cool
This term revolves around the validity of LLM generated data in the Social Sciences. We have a set of four great speakers to enlighten us on different aspects of this issue.
If you'd like to join, sign up here: TADA.cool
October 1, 2025 at 12:58 PM
Very excited to announce the next term of the TADA speaker series!
This term revolves around the validity of LLM generated data in the Social Sciences. We have a set of four great speakers to enlighten us on different aspects of this issue.
If you'd like to join, sign up here: TADA.cool
This term revolves around the validity of LLM generated data in the Social Sciences. We have a set of four great speakers to enlighten us on different aspects of this issue.
If you'd like to join, sign up here: TADA.cool
July 30, 2025 at 7:44 PM
July 30, 2025 at 7:41 PM
However, voters move more towards their own party's position when learning ANY party's position.
April 11, 2025 at 12:55 PM
However, voters move more towards their own party's position when learning ANY party's position.
The results of my experiment prove me wrong:
while voters change their minds about the issue in response to the news frames, these changes do not translate into party preferences.
This finding holds irrespective of a respondent's attachment to a given party.
while voters change their minds about the issue in response to the news frames, these changes do not translate into party preferences.
This finding holds irrespective of a respondent's attachment to a given party.
April 11, 2025 at 12:55 PM
The results of my experiment prove me wrong:
while voters change their minds about the issue in response to the news frames, these changes do not translate into party preferences.
This finding holds irrespective of a respondent's attachment to a given party.
while voters change their minds about the issue in response to the news frames, these changes do not translate into party preferences.
This finding holds irrespective of a respondent's attachment to a given party.
Existing theories of political behavior give contradicting answers to this question.
Rational choice theories expect that voters become more inclined to vote for a given party as their attitudes move closer to that party's issue position.
Rational choice theories expect that voters become more inclined to vote for a given party as their attitudes move closer to that party's issue position.
April 11, 2025 at 12:55 PM
Existing theories of political behavior give contradicting answers to this question.
Rational choice theories expect that voters become more inclined to vote for a given party as their attitudes move closer to that party's issue position.
Rational choice theories expect that voters become more inclined to vote for a given party as their attitudes move closer to that party's issue position.
I rule out alternative explanations, such as readers abandoning the tabloid after the editorial shift. Instead, I find that immigration attitudes became much less volatile following the 2015 refugee crisis. I argue that these 'crystallized' attitudes are less susceptible to shifts in media framing.
February 9, 2025 at 3:19 PM
I rule out alternative explanations, such as readers abandoning the tabloid after the editorial shift. Instead, I find that immigration attitudes became much less volatile following the 2015 refugee crisis. I argue that these 'crystallized' attitudes are less susceptible to shifts in media framing.
Leveraging large-scale panel data from ~15,000 respondents in the excellent German Longitudinal Election Study Panel, I examine whether this stark shift affected readers’ immigration attitudes.
The surprising result: attitudes toward immigration and related variables remained unchanged.
The surprising result: attitudes toward immigration and related variables remained unchanged.
February 9, 2025 at 3:19 PM
Leveraging large-scale panel data from ~15,000 respondents in the excellent German Longitudinal Election Study Panel, I examine whether this stark shift affected readers’ immigration attitudes.
The surprising result: attitudes toward immigration and related variables remained unchanged.
The surprising result: attitudes toward immigration and related variables remained unchanged.
Analyzing ~2.5 million news articles, I show that this editorial change heavily affected the tabloid’s immigration coverage: Bild increased their attention to crime frames in their migration coverage by 42%, compared to other daily newspapers.
February 9, 2025 at 3:19 PM
Analyzing ~2.5 million news articles, I show that this editorial change heavily affected the tabloid’s immigration coverage: Bild increased their attention to crime frames in their migration coverage by 42%, compared to other daily newspapers.
I focus on a unique case involving the largest German tabloid, Bild. In 2017 the chief editor - who had hosted a refugee family - resigned over allegations of sexual harassment. His successor would severely change the outlet's migration coverage.
February 9, 2025 at 3:19 PM
I focus on a unique case involving the largest German tabloid, Bild. In 2017 the chief editor - who had hosted a refugee family - resigned over allegations of sexual harassment. His successor would severely change the outlet's migration coverage.
Or to put it more bluntly: unless we want tech billionaires to control what is considered true and what not, we need to define regulations for the production of LLMs, similar to medical products. A certification process is in order @ec.europa.eu !
January 28, 2025 at 8:35 AM
Or to put it more bluntly: unless we want tech billionaires to control what is considered true and what not, we need to define regulations for the production of LLMs, similar to medical products. A certification process is in order @ec.europa.eu !
📚💡How can we study innovation and specialization using 19th-century library data and NLP methods? Join us next Wednesday when Caterina Chiopris presents "The Diffusion of Ideas" at TaDa RG!
📅 Next Wednesday, 27.11.24, 5pm Berlin time
For more info subscribe to our newsletter: tada.cool
📅 Next Wednesday, 27.11.24, 5pm Berlin time
For more info subscribe to our newsletter: tada.cool
November 24, 2024 at 7:47 AM
📚💡How can we study innovation and specialization using 19th-century library data and NLP methods? Join us next Wednesday when Caterina Chiopris presents "The Diffusion of Ideas" at TaDa RG!
📅 Next Wednesday, 27.11.24, 5pm Berlin time
For more info subscribe to our newsletter: tada.cool
📅 Next Wednesday, 27.11.24, 5pm Berlin time
For more info subscribe to our newsletter: tada.cool
We will host four more amazing speakers covering a variety of NLP problems and applications over the spring:
April 5, 2024 at 12:34 PM
We will host four more amazing speakers covering a variety of NLP problems and applications over the spring:
Dear TaDa-Nerds, we are back !
Sarah Masud will make the kickoff next Wednesday 5pm (Berlin), joining from New Delhi to explain the relevance of pre-training variables in the performance of LMs in hatespeech detection!
More info on Slack: join.slack.com/t/tadapolisc...
Sarah Masud will make the kickoff next Wednesday 5pm (Berlin), joining from New Delhi to explain the relevance of pre-training variables in the performance of LMs in hatespeech detection!
More info on Slack: join.slack.com/t/tadapolisc...
April 5, 2024 at 12:34 PM
Dear TaDa-Nerds, we are back !
Sarah Masud will make the kickoff next Wednesday 5pm (Berlin), joining from New Delhi to explain the relevance of pre-training variables in the performance of LMs in hatespeech detection!
More info on Slack: join.slack.com/t/tadapolisc...
Sarah Masud will make the kickoff next Wednesday 5pm (Berlin), joining from New Delhi to explain the relevance of pre-training variables in the performance of LMs in hatespeech detection!
More info on Slack: join.slack.com/t/tadapolisc...
The variable name to start your weekend.
February 23, 2024 at 2:02 PM
The variable name to start your weekend.
🚨 *Last session* of our TaDa term: next week we are joined by @dirkwulff.bsky.social (Uni Basel/MPI Berlin), who will share a fresh perspective on how to study cognitive semantic representations 🧠 with embedding models!
📅 Wed. 21st, 17:00 CET
Join us via Slack: join.slack.com/t/tadapolisci/
📅 Wed. 21st, 17:00 CET
Join us via Slack: join.slack.com/t/tadapolisci/
February 16, 2024 at 10:45 AM
🚨 *Last session* of our TaDa term: next week we are joined by @dirkwulff.bsky.social (Uni Basel/MPI Berlin), who will share a fresh perspective on how to study cognitive semantic representations 🧠 with embedding models!
📅 Wed. 21st, 17:00 CET
Join us via Slack: join.slack.com/t/tadapolisci/
📅 Wed. 21st, 17:00 CET
Join us via Slack: join.slack.com/t/tadapolisci/