The left-liberal skew of Western media Emil O. W. Kirkegaard, Noah Carl, et al
Are the media left-wing biased? ● ●
Common claim by conservative pundits, but little published research (why is that?) Published research is almost entirely about US media ○
● ●
E.g. Groseclose & Milyo (2005)
But claims about media bias are common in many countries Hence, need to investigate across many countries
Method options ●
Funding analysis ○ ○
●
Content analysis ○ ○ ○
●
Curiously, used by both Marxists (bourgeoisie/capitalists) and Nazis (Jews) Funders usually not involved much in daily operations Manual is slow, biased? Machine learning works well, been done for US media Machine learning hard across languages
Media personnel analysis ○
Usually only data for journalists
What use is data about journalists? ● ● ●
Distortion theory/model (Groseclose 2011) Same reason to study academics political learnings -- beliefs affect behavior! For journalists, beliefs affect: ○ ○ ○
● ●
There can be no pretending that political views do not affect behavior Hence, conclusive evidence of biased leanings is conclusive evidence of political bias ○
●
Choice of stories -- often many different things one can write about Choice of sources -- sources have political leanings too, which to use? Choice of framing -- call Charles Murray “white nationalist”? “controversial”? “conservative”?
But hard to determine the extent/importance without content analysis
Definition: biased leanings = not representative of general population
Getting the voting data ●
Not much data in academic journal papers ○
●
But some can be found in more obscure sources: ○ ○ ○
● ● ●
Academics left-wing too and not interested in media bias because it benefits themselves Dissertations Books Local language sources such as newspaper articles and reports by media companies
But who can read 20 different languages? Many research assistants, international diverse team Find and contact local people with relevant knowledge
Getting the party data ●
Party information copied from English Wikipedia ○
● ● ●
Example: https://en.wikipedia.org/wiki/Sweden_Democrats
Semi-structured format, friendly to automatic analysis Political ideology tags: nationalism, socialism, feminism, anti-immigration etc. Political position: far left to far right, quantified as -3 to 3
The data, so far ●
17 countries covered: 16 Western + Russia ○ ○
● ● ●
Denmark, Sweden, Norway, Finland, Germany, Switzerland, Austria, Netherlands, Belgium, France, UK, Ireland, USA, Canada, New Zealand, Australia, Poland Missing large countries: Italy, Spain, Portugal, Greece, most east European countries
113 parties total Some missing party data, but about 100 parties with data All public and being updated ○
https://docs.google.com/spreadsheets/d/1X9g7Eh-EerrkjtabHazNu-a-niVC4t5dObok_aNUP0g/ edit?usp=drive_web
Quantifying party preferences ● ● ●
No standard method (as far as I know) Decided on pluralistic approach Delta percent point (d) ○
●
Relative ratio (RR) ○
●
Journalist% / general%
Odds ratio (OR) ○ ○ ○
●
journalist% - general%
Parties cannot get more than 100% of the vote So large parties cannot have high RR Odds ratio removes this issue, but harder intuitive understanding
Example: journalist% = 35, general% = 10 ○ ○ ○
d = 25 RR = 3.5 OR = 4.85
The left-wing skew
●
Missing data due to 1) lack of Wikipedia coverage, 2) no simple left-right position, 3) party support very small in general population (<2%)
Which ideologies do journalists like? Not much method variance means in cols 1-3 medians in cols 4-6
Multivariate analysis ● ● ● ●
Difficult because 20 predictors and only n = 100ish Work in progress Main predictors from tag analysis were consistent Left-wingness itself not predictive?
Conclusions ● ● ●
Large left-wing skew of journalist voting behavior, r’s about -.55 Tag analysis show obvious patterns one would expect Essentially, expectations were confirmed
References ● ●
Groseclose, T., & Milyo, J. (2005). A measure of media bias. The Quarterly Journal of Economics, 120(4), 1191-1237. Groseclose (2011). Left Turn: How Liberal Media Bias Distorts the American Mind.