,
8 tweets,
3 min read
Read on Twitter
Today @VinArceneaux & I posted our new pre-analysis plan @OSFramework that introduces our take on how & when physiological threat sensitivity is associated with social conservatism. Here some thoughts on what we do & what we learned from writing the pre-analysis plan (1/8)
In our recent working paper (N=635 in US and NL) we failed to find evidence for the positive association between threat sensitivity (physiological measures) and social conservatism documented in Science science.sciencemag.org/content/321/58… and provided in path b below 
We expect path B might only happen among people who are phsyiologically aroused & self-report this (quadrant #2). People in #4 are not responding and might be more liberal. Expectations for off-diagnal tricky. Those in #1 might have down-regulated response. #3 might be signalling
We test path b: We measure social conservatism & show people images (snakes, spiders, attacks) while measuring skin conductance response during exposure as well as their self-reported arousal & valence after each image. This allows us to create index of concordance per image
To get at path D we also conduct an Implicit Association Test tapping into ideology. Physiological threat sensitivity might correlate stronger with other implicit attitudes. To get at path A we measure Openness and Conscientiousness and expect to find well-known associations.
The full pre-analysis plan can be downloaded here: tinyurl.com/y26ouptt. If you have any suggestions, we can incorporate them now.
Recently @claire_gothreau ran 15 respondents through study. Using this dataset we created code to test hypotheses. We learned that analysis strategy was not always correct test of our hypotheses. We adjusted models where necessary & became more precise formulating expectations
Our @OSFramework page contains pre-analysis plan, scripts to create qualtrics and e-prime experiments. Code intended to compile, create and analyze data. Whole project can be found here: tinyurl.com/y3aqhtry Are we missing anything?
