Tuesday, October 18, 2005

Reliability woes

I'm in the middle of working on a content analysis of award-winning radio ads. For those of you who don't know, a content analysis is where you systematically count the presence of certain types of content in media messages. The content can be mainifest--in other words, so obviously present that there is little question. Or it can be latent--meaning it is more open to interpretation.

When doing a content anlysis you develop a code book which describes the different categories of content you are looking at. You then train the coders what to do with the content. Then, you basically test the understandability of the codebook by having all your coders analyze the same messages and test how similar their answers are.

Well, that's right where we are now. We are doing pretty good on recognizing the manifest content. You know--we're pretty good at recognizing when ads have multiple speakers and background music. However--we were also pretty good at some things that I thought we might have trouble with. Like whether it was an ad for a product or service . Or the type of product category it falls in.

But, we have trouble with some things that require much more interpretation. Tonight I have to work on how we define sex appeal and humor.

---But--for right now, I'm watching a TIVO'd version of Everybody Hates Chris. If you haven't seen it yet, I suggest you check it out.


IUAngelini said...

Believe me, reliability of questions where a personal interpretation is necessary is so difficult. We're still struggling with it!

-Content Analysis King

Anonymous said...

Isn't that "Content analysis queen?"