This repository contains data, analytic code, and findings based on a collaboration between BuzzFeed News and Liliana Bounegru, an author of the forthcoming A Field Guide to Fake News.
The findings support portions of the BuzzFeed News article, "Fake News, Real Ads," published April 4, 2017. Please read that article, which contains important context and details, before proceeding.
This repository contains the following five CSV files:
-
data/observed-trackers.csv: The data BuzzFeed News received from the researchers behind A Field Guide to Fake News, slightly restructured. -
output/tracker-matrix.csv: For each website-and-tracker combination, whether the website had that tracker (a) before Nov. 2016, and/or (b) in March 2017. (Includes only "comparable" sites. See below for details.) -
output/tracker-counts.csv: Using the data inoutput/tracker-matrix.csv, counts the number of comparable sites containing each tracker during the two timeframes, and the net change. -
output/tracker-statuses.csv: Using the data inoutput/tracker-matrix.csv, classifies each website-and-tracker combination into one of four categories: kept, removed, added or never had the tracker. -
output/tracker-statuses-pivot.csv: Using the data inoutput/tracker-statuses.csv, counts the number of sites that kept, removed, added, or never had each tracker.
To make the two time frames — before Nov. 2016 and in March 2017 — comparable, we removed two types of sites:
-
Sites with no observed trackers in the "before" period
-
Sites that had disappeared by the "after" period
After doing so, we were left with 51 websites.
The Python code that processes the data can be found here.
Contact Jeremy Singer-Vine at jeremy.singer-vine@buzzfeed.com.
Looking for more from BuzzFeed News? Click here for a list of our open-sourced projects, data, and code.