The PRL Animosity Survey is an ambitious and expensive undertaking. That’s why we enthusiastically provide all of our collected survey data to the public.
If you use our data in your work, we humbly ask that you cite the project:
How do you calculate “polarization”?
Affective polarization is the difference between {inparty} and {outparty} ratings. We calculate polarization exactly as follows:
- Filter for Republicans and Democrats (using
pid7)- capturing Strong, Leaning, and Not very strong partisans
- For Democrats, calculate:
democrat_therm_1-republican_therm_1 - For Republicans, calculate:
republican_therm_1-democrat_therm_1 - Remove anyone with a negative score
What are the date/time variables?
The starttime and endtime columns mark when a participant starts and finishes our survey. We often find that participants start our survey to claim a spot in the YouGov panel, but then opt to finish the bulk of the survey later. For this reason, we often use end times for any time-series analyses.
How do you check for participants who aren’t paying attention?
before June 2023
In our original engagement check, we show participants a passage about a wildlife funding program for a particular state. The passage is then removed, and participants are asked what state (of 7 possible options) the passage referred to. We’ve recoded this data to label participants as “engaged” or “not engaged” based on whether they guessed the correct state.
after June 2023
After June 2023, we changed our engagement check as follows. The survey begins with a made-up news story about a particular person from a particular location experiencing a particular event. The four key features of the story (name, occupation, location, and event) are randomized. The story is removed, and the respondent must correctly identify one of the four features.
If a participant fails the engagement check, the survey stops. Consequently, data after June 2023 only contains respondents who passed the engagement check.
