Tuesday, July 30, 2013

Measuring Media Exposure

A standard measure in our field is news media exposure -- how much people watch TV news or read a paper, that sort of thing. It's been around since the 1950s, it remains in most Pew surveys, despite research suggesting respondents exaggerate their exposure and the question is full of measurement error.  One guy called news media exposure "one of the most notable embarrassments of modern social science."  Ouch.

And yet, here I am, studying news media exposure because, dammit, I'm a notable embarrassment.

I'm comparing specific sources of TV news (Fox, CNN, ABC, etc.) with the traditional 0-to-7 days a week of generic TV news watching.  Is it better to be specific?  Well, yeah .. except it costs you additional survey questions which anyone in the trade will tell you costs you additional money to pay for that time in the survey.

We obviously gain something by asking about specific news sources.  If nothing else, it's well established now in the literature that folks who watch Fox News come away with a very different version of the world than those who watch other news sources -- more so than those who watch MSNBC.  If you doubt this, see the latest Journalism and Mass Communication Quarterly article about "death panels" and sources of news.  Fox News matters.  I'll blog on this article more another day.

Anyway, back to my problem because it's all about me.  I'm comparing how exposure to generic television news is the same or different from exposure to the three broadcast and three cable news networks on five different criterion variables (political knowledge, likelihood to vote, etc.).  To put it in less PhDweeb terms, I want to know whether you get a better result from the generic single question than from the six specific network questions.  Oh, and even better, I've tossed in exposure to specific programs in the three cable news networks (The O'Reilly Factor, etc.).  Fun stuff.

But ... I'm also mulling over a different measure.  Take TV news exposure on a 0-to-7 days scale and combine it with a count of how many different TV networks you say you watch (yes-no question).  It's a weighted measure.  People who watch three networks but not very much (1 day a week) are different than those who watch three networks 7 days a week.  To reverse this, people who report watching TV news 7 days a week but from only one network should in some way be different than those who watch 7 days a week but three different networks.  The math here is tough, how to come up with a valid weighted measure.  Suggestions welcome.  Right now I'm working with a simple interactive term (# of sources x days generic exposure).  I haven't run the analyses yet to see how it fares against the more traditional generic measure of specific network measures, but I'm hoping to see something useful, something helpful for future scholars.

Or at least another pub.  Yes, this is what we academics do when not teaching and other stuff.

Again, methodological suggestions welcome.

No comments: