ANES 2012 and Media Variables

The American National Election Studies have historically been weak on media questions.  For the 2012 survey, released Monday, this is not the case. From a media perspective, here are a few new (and old) twists in the preliminary release of the ANES 2012 Time Series Study:

• Exposure and Attention.  There's a difference between exposure to a medium and attention to that medium.  At times the ANES has included both measures.  They're back to some degree.

• Source Specificity.  Questions get at specific programs and news sources (i.e., NYTimes, Good Morning America).  This is extensive and useful.  Keep in mind, however, many respondents are coded as "inapplicable" due to random assignment to questions.

• Separation.  Like some previous surveys, the 2012 version separates "print" from "online" news sources.  Important.  This is becoming the default for many surveys today.

• Candidate websites.  This version asks a number of questions about respondent visits to campaign websites.  Useful.

• Social Media.  Yes, Twitter and Facebook show up in the survey in a couple of different ways.  Again, useful.

I've only downloaded the codebook and data today, so I can't write extensively yet about results. This is a huge survey, the 29th in a series, with the largest ever sample size (5,916 respondents) with random samples in a fact-to-face and Internet modes.  The "Big Five" personality traits are included, as well as some other neat tricks.

For those budding mass comm scholars out there, a warning.  ANES (and to a lesser extent, Pew) data can be daunting when you first sit down with a codebook and your favorite statistical package.  So many questions.  So many confusing question experiments.  This is particularly true of the 2012 version.  Read the explanations carefully, weight the data appropriately,  and beware the various splits and splices in the ANES questions.  It's easy to screw up hours of analysis.  Yeah, I've done it myself.

The strength of the ANES stuff has always been the large collection of variables for analysis.  The weakness, from a polcomm perspective, has been its corresponding weak collection of media variables.  This has changed in the last few releases.  Perhaps it's time I offered a graduate class just in the careful analysis of secondary data.  Our students seem oblivious to the potential here.  I can feast for a year or more on the 2012 data alone.