News networks are better at crushing (or lifting) outsider campaigns than they are at helping major candidates

By Kent R. Kroeger (August 16, 2019)

July 1, 2019 Democratic Debate ( CNN/YouTube Screen capture)

The Systematic Bias of the National News Media is Undeniable

Recently, Michael Tauberg, an engineer by day and data journalist at night, published data on the tone of online news coverage for each of the 2020 Democratic presidential candidates from January to April 2019.

Source: Michael Tauberg
Source: Michael Tauberg

Changes in cable news coverage can change candidate support — but is it predictable?

Every political campaign I’ve worked on had a candidate and staff that complained incessantly about their campaign’s news coverage. One nasty news story could get a reporter barred from future interviews. But what candidates and campaign managers feared most — more than being on the receiving end of a negative story — was getting ignored by the media. Nothing kills a campaign faster than not being covered. Advertising and door-knocking can help build name recognition and promote a candidate’s core messages, but the credibility and visibility conveyed to a campaign through a national news media filter is irreplaceable — even in the age of social media.

The Data

The following is a very preliminary look at this question regarding news media influence on political candidate support. Using APIs to query a GDELT Project database on daily cable TV news coverage from January 4 to August 6, 2019, and downloading Democratic nomination polling data from RealClearPolitics’ data repository, I conducted a time-series analysis (vector autoregression) to determine whether or not changes in the volume of a candidate’s cable news coverage causes changes in a candidate’s popular support.

Data Source: RealClearPolitics; Polling data displayed and analyzed as 7-day moving averages
Data Source: The GDELT Project; Polling data displayed and analyzed as 7-day moving averages
Data Sources: RealClearPolitics and the GDELT Project; data smoothed using a 7-day moving average

The Statistical Method

Noble Prize winner in economics, Clive Granger, defines a relationship between two variables as causal (X1 Granger-causes X2) if prior changes in X1 predict future changes in X2, independent of past values in X2 and while controlling for other potential causal factors.

Model Specification

The VAR models estimated here for popularity (Y) and volume of news coverage (X) specify a p-order = 7, which means the ten models (two models for each candidate) looks back 7 days to assess the relationship between Y and X. All variables were measured at the daily level and smoothed using a 7-day moving average. The variables were also differenced in order to meet VAR’s stationarity requirement. This means we are, in fact, testing whether changes in X cause changes in Y and vice versa.

The Results

The bottom line up front: For three of the five candidates (see Figure 6), increases in their volume of cable news coverage caused small but significant increases in candidate popularity. Also, for three of the five candidates, increases in candidate popularity caused small but significant increases in the volume of their cable news coverage.

Data Sources: RealClearPolitics and the GDELT Project

Final Thoughts

Perhaps the most interesting candidates in this analysis are Bernie Sanders and Kamala Harris. There was little evidence of any causal relationship between their volume of cable news coverage and their popularity. It doesn’t surprise me that Sanders doesn’t even get a meager lift anymore from positive news coverage. It does surprise me that Kamala doesn’t. Just a visual inspection of the relationship between her popularity and her cable news mentions reveals what appears to be a strong (positive) relationship between the two (Figure 9).

Data Sources: RealClearPolitics and the GDELT Project; data smoothed using a 7-day moving average

APPENDIX: Additional Graphs

Figure A.1: Candidate Support and Cable TV Coverage (Biden)

Data Sources: RealClearPolitics and the GDELT Project; data smoothed using a 7-day moving average
Data Sources: RealClearPolitics and the GDELT Project; data smoothed using a 7-day moving average
Data Sources: RealClearPolitics and the GDELT Project; data smoothed using a 7-day moving average
Data Sources: RealClearPolitics and the GDELT Project; data smoothed using a 7-day moving average
Data Sources: RealClearPolitics and the GDELT Project; data smoothed using a 7-day moving average
Data Sources: RealClearPolitics and the GDELT Project
Data Sources: RealClearPolitics and the GDELT Project
Data Sources: RealClearPolitics and the GDELT Project
Data Sources: RealClearPolitics and the GDELT Project
Data Sources: RealClearPolitics and the GDELT Project

I am a survey and statistical consultant with over 30 -years experience measuring and analyzing public opinion (You can contact me at: kroeger98@yahoo.com)

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store