Wednesday, May 22, 2013

By National TV News, I Mean ...

When we ask survey respondents how much they watch national television news, what goes through their mind? Do they calculate some generic TV watching total? Do they weight their answer depending on how much they watch, say, Fox News versus CNN versus other networks?

I was curious, so of course I turned to the data. 

This is part of a bigger paper I'm writing, but I thought I'd share a few results. The idea is simple -- in a survey we may ask the generic question (national television news) and we may or may not get into specifics (ABC, MSNBC, etc.).  Which is better?  Is there any real difference?  This matters both methodologically (we can save money with a single question) and theoretically (generic TV viewing may predict different outcomes than network-specific TV viewing).  Watching Fox News, we can all agree, is somewhat different than watching MSNBC.

First, lemme get at the overlaps.  A national survey asked respondents how whether they watched the three broadcast networks, the three cable networks, and also included a generic "national television news" question.  As you'd expect, answering "yes" to one meant you were likely to answer "yes" to the others.  The lowest overlap?  No surprise -- Fox News and MSNBC (5.4 percent said "yes" to both).  The greatest overlap? NBC and CBS (19.5 percent "yes" to both).  And how do these overlap with the generic TV question?  Pretty well, most around the 20 percent overlap mark except MSNBC (9.9 percent), perhaps because it's not available on as many cable systems.

The real question here, from a PhDweeb standpoint, is whether the generic question is as good a predictor of various outcomes as the specific network measures.  I'm in the middle of this, but I can tell you a few small differences do emerge, but not as many as you might expect.  Few are surprising. Below, I sketch out some regression model results that control, statistically, for a bunch of other factors (education, political interest, etc.) to predict stuff like political knowledge.  See below.
  • Political Knowledge -- Three are negative (national TV news and watching ABC and CBS).  The others, no relationship.  In other words, the generic question predicts less knowledge (only barely, by the way), as do two of the three broadcast networks.  MSNBC comes close to significance and is positive.  That's interesting.
  • Rate Obama -- No surprises here.  Generic national TV viewing and all of the cable networks, save one, predict positive feelings toward President Obama.  The exception, of course, is Fox News, which is negative.  Duh. If you ever wanted quantitative evidence of Fox News exceptionalism, here it is.
  • Talk to Friends -- This is a fun one.  Does watching the news predict your likelihood to talk politics with friends?  The generic TV news viewing measure is a strong predictor even after statistical controls (beta = .14, p<.001 for the nerds among you), but not a single specific network is related to talking except Fox News, and it's positive too.  That's fascinating.  I'm guessing people watching Fox gives those viewers something to talk about, no doubt about Obama and not in a good way.
I'm even digging down to the program level (The O'Reilly Factor, etc.) to see how much those differ from their generic networks.  In other words, is there something special about watching Sean Hannity as compared to generic Fox News?  Maybe. Maybe not.  From an academic standpoint, your dependent variables may drive the use of specific networks versus a generic TV news exposure measure.  On some, it'll matter.  On others, it may not.  I have a lot more data to crunch and writing to do before this will go out to an academic journal, but it has some strong possibilities.

No comments: