Recall vs. Recognition Knowledge

And so in a moment of PhDweebishness I turn to the a favorite question of mine -- is there a difference between recall and recognition when it comes to measuring political knowledge.

By recognition, I mean a multiple-choice test in which you're asked a question and given choices to pick from.  So if I asked you to name the Vice President of the U.S. you'd get four names, including Joe Biden.  By recall, I might ask you "Who is the Vice President of the U.S.?"  That second one is tougher.  You have to pull the answer out of your head, retrieve the information with no help, no cues.

Yes, I've written a lot on this topic (here, for example, and here). 

I'm polishing a study that looks (yet again) at this topic, basically the same as my first linked post in the sentence above.  Yes, TV news use predicts only recognition knowledge (beta = .27, p<.001), not recall knowledge (beta = .06, ns).  This controls for a bunch of other factors, the usual suspects of age and income and education and political interest and a bunch of others.  The point is simple, that watching TV news helps you to recognize through cued recall a correct answer, but it doesn't help if you have to retrieve the information without some cue.  So what does predict recall?  Among media, again controlling for all the various factors, only getting your news from Internet news sites (beta = .12, p<.01) but not reading paper newspapers (beta = -.07, ns).  Newspaper exposure has always been a modest predictor of political knowledge, but that seems to be changing.  The Internet is the new Newspaper, at least when it comes to predicting political knowledge.

And here's the interesting part -- TV really works best on recognition for less educated folks.  For those with greater education it also predicts recognition knowledge, but just not as strongly.  This fits nicely into the knowledge gap literature.

For those who care, a few other interesting findings from the multivariate models

• Age predicts recall, but not recognition knowledge.
• Income, education, and political interest predict both.
• Being registered to vote predicts recognition, not recall.

With luck, all of this and more will eventually come to an academic journal near you.

Recall and Recognition

I've blogged often (and published academic research) on the differences between the recall and the recognition of political information. These two forms of what people know differ not only methodologically but theoretically as well.  To put it simply -- and less PhDweebishly -- recall is one's ability to pull information out of memory with no help, a cognitively difficult task.  Recognition is like a multiple-choice test in which you're given a number of answers to choose from, a cognitively easier task.

Recall Example:  Who is Nancy Pelosi?
Recognition Example: Who is Nancy Pelosi?  Is she (a) Senate Majority Leader, (b) Speaker of the House, (c) a brand of underarm deoderant, (d) Secretary of State?

The first question, the recall example, is harder.  It would be even harder if you asked Who is the Speaker of the House? versus giving respondents a name to work with, but that's a different theoretical and methodological issue for another day (yeah, I did a paper on this, but it never got accepted ... damn those reviewers).

So why go on about recall vs. recognition?  There is a study in a 2009 issue of Political Behavior that examines knowledge about congressional candidates.  It's not actually a test of recall and recognition, but it includes both approaches and so I'm riffing off of it since it serves my purpose.  Full cite at bottom.

Here's a part I found kinda interesting.  The study includes a variable called media market concentration.  What's that?  When media markets straddle or bleed across congressional districts, you expect less knowledge about candidates since the coverage of any particular race is diluted.  When districts are better aligned with media markets, knowledge should improve (journalists are covering just one race).  That's the guts of this concept, though the method to create it is more complicated.  See the study for details.  

Their result?

When media market boundaries are well aligned with the boundaries of congressional districts, incumbents receive no particular benefit, but recognition of challengers is disadvantaged

Other than the awful, jargon-drunk use of disadvantaged here, people in districts where you'd expect higher knowledge because of alignment actually do a lousy job of recognizing challengers.  There's no relationship in recognition of incumbents (probably a ceiling effect).

In recall, they do not break it down by incumbents and challengers (not sure why), and in this case there is a negative relationship between accurate recall of all candidates and media concentration.  This is similar to above, where you have a non-significant relationship and a negative one, so it's no surprise -- I guess -- that an overall recall measure is negative.  This also points out the difficulty in improving recall versus recognition, especially in "media markets" defined by television, which is thought to improve recognition but not recall.

My take?  In reading the methods, the tables, and the study, at least when talking about media concentration, it seems to me the results support the idea that news media coverage helps a little on recognition but not at all with recall -- again supporting my discussion in the previous graph.

Full cite
Wolak, Jennifer (2009).  The consequences of concurrent campaigns for citizen knowledge of Congressional candidates.  Political Behavior, 31, 211-229.

Metamemory

Stick meta in front of a word and we move a step back. It becomes thinking about that word, whatever it is. Too PhDweebish for ya? Here's an easy example. Put meta in front of cognition and metacognition is thinking about thinking. Metamemory, according to a study I read this morning, is belief in one's own memory efficiency -- kinda thinking about your memory.

Yeah, I'm getting older, so I wonder how long the old brainpan is gonna work.

This also gets into feeling of knowing. We've all suffered this one. Some call it the tip-of-the-tongue phenomenon. Or maybe their different. Feeling of knowing is just what it sounds like: you know something, but you can't quite remember. It's like an annoying itch you can't quite reach. Tip-of-the-tongue is similar, a word or phrase or name that's right there, on the tip of your tongue, but you just can't quite grab it. There are probably conceptual and theoretical differences between the two, but it's too early, and I'm not caffeinated enough, to tease them out here.

Why am I going on about this? In part to demonstrate that there's a wide world of research out there on memory and the different ways it goes wrong, which can definitely play a role in how those of us who study political knowledge deal with survey respondents or experimental subjects. We know that TV news, for example, does a bit better with recognition of political information but not so well with recall of facts. There's a nagging feeling of knowing, a tip-of-the-tongue example for you right there. TV news alone probably leads to greater feelings that you know something, because of haphazard exposure to TV news while doing something else, thus leads to better recognition of information as opposed to easy recall. What might be interesting to study is any frustration that emerges from that nagging feeling that you know something but can't quite yank it out of trace memory because you watched it on the boob tube rather than read it online or on paper (thus, deeper processing).

Names

So I'm working on a piece of research with a simple underlying thesis -- names matter. This gets into how we structure knowledge, but the basic idea is that asking someone a political knowledge question that starts with the name (as in, who is Dick Cheney?) is easier to answer than one that starts with the office (who is the vice president?).

Simple, eh?

Peek under the hood, into the engine of the brain, and this has some interesting twists and turns and cables and hoses, but what I'm working on is a media connection, that people who rely on entertainment-based news or even TV news will be more successful at questions in which they are given the name versus questions in which they are prompted instead with the office. People who rely in print news, a more active process, should be better able to handle both kinds of questions. Again, this gets into how each form of news is remembered and how this in turn influences memory structure.

Okay, so what did I find? Some support, at least in my initial data analysis. I've got a lot of fine tuning to do, numbers to crunch, data to manipulate, and I've got to think through some complicated interaction effects. But I think I'm on to something.

There is little practical use for this beyond understanding how people structure political information in their heads, except that the kinds of knowledge questions you ask in a survey can strongly influence who is likely, or unlikely, to answer them. As people become more and more reliant on skimming the news, catching it on The Daily Show, or a few minutes of CNN they will become less able to answer the "hard" questions.

In some ways this resembles the results I found in my Journal of Broadcasting and Electronic Media piece a year or so ago, in which I found some support for the idea that watching comedy news programs was more associated with recognition of political information than with actual recall of political facts.