Paper w/ @KonitzerTobias @_JenAllen @stephnie Howland @markusmobius @duncanjwatts "Comparing Estimates of News Consumption from Survey and Passively Collected Behavioral Data" academic.oup.com/poq/advance-ar… confirms & extends earlier results on over-reporting news consumption in surveys
Way more people report watching FNC & MSNBC in past week than ever watch in a month, roughly as many report FNC most used station as ever watch in a month, but MSNBC more realistic reporting of most used 2/9 
Plausible that social desirability bias of over-reporting news consumption much more important for Republicans reporting FNC, than other non-politically aligned news consumption patterns 3/9
Massive differences exist between "FNC Viewers" & "non-FNC Viewers" in sentiment & knowledge tricky: half of self-reported "FNC Viewers" don't actually watch Fox News, so either real difference is much bigger or actual treatment comes from larger Republican ecosystem 4/9
Results extend to online & social media news consumption, where survey self-reports far outstrip reasonable estimates from passive data. Note flexibility of passive data that can easily answer various types of outcomes a researcher may reasonability wish to explore 5/9
Passive data also captures trends missed by survey data 6/9
Sometimes survey data may be technically correct but misinterpreted by researchers: e.g. Pew defines news consumption as “information about events & issues that involve more than just your friends or family” which technically includes lots most of us would not consider "news" 7/9
Further, in eagerness to promote impact of new news sources, reporting on survey data about news consumption on social media tends to cherry-pick & exaggerate survey results, in way that survey did not technically report: e.g. of NYT misreporting 8/9
Paper goes on at length to discuss how future needs to be mixture of passive data for behavior and survey data for sentiment: together they hold the key to understanding the impact of news on the general population academic.oup.com/poq/advance-ar… 9/9
• • •
Missing some Tweet in this thread? You can try to
force a refresh
