Data and scripts for 'A Critical Evaluation of Tracking Surveys with Social Media: A Case Study in Presidential Approval' by Ferg, Conrad, Gagnon-Bartsch

Data sets:

• approval_topline2019-08-30.csv: Daily survey data as downloaded from https://projects.fivethirtyeight.com/trump-approval-ratings/?ex_cid=rrpromo on August 30, 2019.
• placebo_sentiments2019-08-25.csv: Daily mean sentiment of tweets containing each of the placebo words from January 16, 2017 through August 25, 2019.
• trumpSent2019-08-25.csv: Daily mean sentiment of the 1000 randomly sampled tweets per day used in the tracking analysis from January 17, 2017 through August 23, 2019.
• Tweets from our set of users in the longitudinal study are under 'Releases'. To protect the privacy of these users, we do not include any identifying information about the users or their tweets.

Scripts:

• Cross-Sectional Analysis.R: R script for cross-sectional analysis. Using 'approval_topline2019-08-30.csv', 'placebo_sentiments2019-08-25.csv', and 'trumpSent2019-08-25.csv' data files, calculates optimal correlation between tweets and survey data, finds reference distribution, and how these change over time as more data is gathered.
• Longitudinal Analysis.R: R script for longitudinal analysis. Creates plots for frequency and sentiment of tweets of users in the longitudinal study from 2016 through mid-2017.

We also provide code for how we compute the sentiment scores from csv files of raw tweets in the 'Creating_data_files' folder. To protect the privacy of users, we do not provide the tweets themselves.

A docker image of the above analyses can be found at: https://hub.docker.com/r/johanngb/twitter-presidential