Thread by @balexanderstats

My next session at #AAPOR is “Things that Divide Us: Ideology, Identification, and Information”. The first talk is “Fake News Interventions: Effective for Both Strong and Weakly Identified Partisans?” presented by Joseph Sandor.

Joseph first explains that while people have a motivation to share accurate information on social media sometimes people share things that are inaccurate. Joseph is presenting the results of an experiment where users rated a headline’s accuracy before sharing. #AAPOR

Next Joseph is walking through the literature on why people share fake news and how sometimes people are motivated to share fake news for political reasons. #AAPOR

Now Joseph shows the experiment design to see the effect of accuracy nudge and identity manipulation. The experiment was conducted on Mechanical Turk with an even number of Republican and Democrats. #AAPOR

There was not support for the hypothesis that accuracy nudges work or that partisan identity threat mattered. #AAPOR

Next we have “Measuring Susceptibility to News Misinformation By Political Party and Media Consumption” by Linley Sanders of @YouGovAmerica

Linley starts by explaining that the purpose of this study to see if American can identify real articles and fake articles by headlines. The fake articles were AI generated. The study was conducted on about 1500 US adults on YouGov panel. The headlines were randomized. #AAPOR

About 70% of headlines were correctly identified. There are differences in accuracy of detecting of age, political party, news attention, and news outlet they consume. They who consume news on social media did perform worse than those who use more traditional sources. #AAPOR

Next we have “Are Centrists Even Real? Combining Survey Self-Reports and Web Tracking Data to Improve Our Understanding of Left-Right Ideology” by Oriol Bosch. #AAPOR

Oriol first explains how web tracking data is and how it can be used to measure news consumption and it’s affect on political opinion. Oriol mentions that tracking data can help supplement self reports. #AAPOR

Oriol says his presentation focuses on Spain because the media outlets are polarized across the political spectrum. The tracking panelists are placed an a social economic and political spectrum. #AAPOR

Next Oriol compares the results from the web tracking and surveys and the results were similar but survey is a better predictor of political ideology. Next Oriol discusses a hidden markov model to better classify the panelists. #AAPOR

The survey data does seem to do a better job at identifying political ideology over the web tracking data. However Oriol mentioned that the tracking data is an alternative to surveys but surveys are better. This his twitter for reference. @orioljbosch #AAPOR

Next we have “Mass Media and Spatial Polarization Examining the Relationship between Media Consumption, Identities Connected to Place and Sentiments Towards Rural and Urban Americans.” By David Coppini.

First David provides an overview about the rural urban divide and social identity framework. David states his presentation focus in the rural urban divide. #AAPOR

The study was conducted with @ipsosus KnowledgePanel in 2019. The survey measured media consumption, rural/urban identities and animosity towards people in rural/urban areas. #AAPOR

The studies did find a urban/rural divide. #AAPOR

The next presentation is “Mass Media and Spatial Polarization Examining the Relationship between Media Consumption, Identities Connected to Place and Sentiments Towards Rural and Urban Americans” presented by David Rothschild. #AAPOR

This presentation is about getting a comprehensive view of political identity then a more basic political identification. Questions include what demographic features matter the most to you? The data comes from Prodege with 2000 respondents. #AAPOR

The models include multi-class logistic regression with cross validation to predict political preferences with different demographics. #AAPOR

David says he will focus on predicting if people are Republicans and Democrats. Adding expanded demographics and identity questions improves the accuracy of predicting party affiliation. #AAPOR

#AAPOR tagging @DavMicRot
David now is summarizing his results. There was significant variation in prediction across across different groups.