Association Survey Design

CleffedUp

Hi all, I was wondering if you social science types have a particular canon or other good reference for sound survey design, i.e. minimizing confounding factors. Good sources on experimental design, in general, are also welcome.

I'd rather learn to fish than just have takeout, but the survey I have in mind is one of association. Let's say I want to study the words people associate with birds. I'm especially interested in how many people associate "yellow," "bill," "rubber," and "squeaky" with the word "duck" as well as "roast," "fried," and "cluck" with the word "chicken."

I might present a really long checklist with control words, I might present a long or short randomized subset of the checklist, or there might be a better approach entirely -- I don't know what I don't know.

Thanks for your time.

DanP

If you are interested in trait ratings, look into the experiments done by:

Asch (1946); Wishner (1960); Rosenberg & All (1968); Rosenberg & Sedlak (1972).

Those experiments have been done to get trait ratings for the description of other humans not animals, but probably they'll put you on the right track.

brainstorm

First, don't call people "types" because it sounds condescending.

Second, the thing that will help you do good research is to understand what you're modeling and why.

You have to know WHY you want to know how many people associate certain words with others. Once you know why you want to know, then you can come up with a model that addresses your real question.