I wrote a few days ago about recall vs. recognition and the challenges of coding open-ended questions that attempt to measure a respondent's political knowledge. Without repeating at length that post, let me just say that measures of recognition are a lot like those multiple choice tests we remember so well from school, while recall are more open-ended kinds of questions.
I am in the middle of a study that looks at how people differ in their ability to answer either recognition (closed-ended) questions versus recall (open-ended) questions. As an example, I might ask you who is Speaker of the House and provide four possible answers, with the correct response (at the moment) of John Boehner included. You'd be likely to either know it or, perhaps, recognize the right person from the list. A recall version would simply ask what office is held by John Boehner? It's a cognitively more difficult task, that last one, because you have to pull it from memory without any hints or help.
Okay, fine. So how's the study looking? Read through this, because it gets to something I find kinda interesting.
Early in, as expected, I find people are much better at answering the recognition versus the recall questions.
In my John Boehner example above, for example, 70.4 percent correctly answered the recognition version, but only 49.1 percent correctly answered the recall version. The same is true for Joe Biden as VP (94.8 percent on recognition, 83.2 percent on recall), John Roberts as Chief Justice (69.9 percent on recognition, 33.1 percent on recall), and David Cameron as Prime Minister of the U.K. (46.5 percent on recognition, 23.1 percent on recall). These results are tentative.
So, why does this matter? In part it matters from a methodology point of view in how we attempt to gauge what people know. One method paints a dismal portrait of the public's knowledge, the other one that is less so dismal.
But for me, the interesting issue is in how use of various media may lead to greater success on one versus the other. For example, my underlying hypothesis is the ephemeral nature of television news (both in how news is presented and in how we tend to watch it) makes viewers more likely to do better on recognition questions as compared to recall questions.
So far, my analysis supports this idea, even after statistical controls for a host of other likely factors (education, political interest, etc.). I love it when data confirm a good theory. It doesn't happen often enough, at least for me.
Interestingly, use of Internet news sites seems to be replacing the reading of the print newspaper as the single best media predictor of knowledge. There's probably a study just in that topic.
More to the point, so far I'm finding that getting the news from Internet news sites is the only media exposure question that is associated with recall (open-ended) questions, while only television exposure is significantly associated with recognition (closed-ended) questions. That's a great finding if it holds up to more scrutiny. My models so far are fairly rigorous, but there's a lot of "under the hood" work to do before writing it all up and submitting the results to an academic publication. Still, I'm hopeful.
The news here? The medium in some ways remains the message. It's about depth of processing and how the news is presented, but most of all it's about how people understand the world. Relying on TV is great for recognizing someone, but not so great when it comes to pulling out a difficult piece of information from memory. Nothing beats reading, and apparently news from the Internet is supplanting news from paper newspapers in this regard.