'Follow back and partisanship on twitter' (AsPredicted #40835)
Author(s) Mohsen Mosleh (Massachusetts Institute of Technology) - mohsen.mosleh@gmail.com David Rand (Massachusetts Institute of Technology) - rand.dave@gmail.com Cameron Martel (Massachusetts Institute of Technology) - cmartel@mit.edu
Pre-registered on 2020/05/11 - 11:56 AM (PT)
1) Have any data been collected for this study already? No, no data have been collected for this study yet.
2) What's the main question being asked or hypothesis being tested in this study? Here we test the probability of follow back of users on Twitter given their partisanship and extremity of partisan account followed them.
3) Describe the key dependent variable(s) specifying how they will be measured. Our main dependent variable is whether the user follows back the account that followed them. 1= follow-back, 0=no-follow-back
4) How many and which conditions will participants be assigned to? Subjects are randomly assigned to one of 4 conditions based on the bots’ partisanship (matched with user or mismatched with user) X extremity of partisanship (extreme, weak), in a 2x2 between-subject design.
We will create 8 bots, 2 bots per condition: We create 4 democrat bots and 4 republican bots where for each partisanship we vary the extremity of partisanship on 2 levels as follows:
Stronger - account name has Trump 2020 or Biden 2020, profile says they are dem/rep, and recent prior tweets contain left/right leaning content.
Weaker - account name does NOT mention politics, but profile says they are dem/rep, and recent prior tweets contain left/right leaning content
All bots follow a set of elite account given their political ideo and randomly retweet from them
In order to estimate the subject's political ideology we use Eady et al. 2019 based on the content they shared.
We assigned subjects using blocking (Higgins et a. 2016) based on the following covariates: 1) subject’s ideology, 2) log transform of number of followers, 3) number of days with at least one tweet in past 14 days (to measure recent activity on the platform), 3) reciprocity rate (#mutual friendship/#followers).
5) Specify exactly which analyses you will conduct to examine the main question/hypothesis. One linear regression model, where DV is whether the subject followed back the bot, and IVs are
Co-partisanship (concordant vs discordant, z-score)
User ideology (democrat vs republican, z-score)
Partisan extremity of bot (stronger vs weaker, z-score)
All interactions
We will also calculate p-values using Fisherian Randomization Inference.
6) Describe exactly how outliers will be defined and handled, and your precise rule(s) for excluding observations.
We excluded subjects with more than 15k followers.
7) How many observations will be collected or what will determine sample size? No need to justify decision, but be precise about exactly how the number will be determined. Our target sample will be 6000 after exclusion equally across conditions.
8) Anything else you would like to pre-register? (e.g., secondary analyses, variables collected for exploratory purposes, unusual analyses planned?) We will also record timestamps of follow back. I.e. when the subject was followed and when the user followed back our account.
