I´m currently working on an IAT experimental design (using OpenSesame). One of the central points of my study is to test how implicit and explicit measures correlates. In short, I´m using an IAT to test implicit associations between voyeurism and social networks. My idea is to add (at the end of the IAT Open Sesame task) one "sketchpad" with the following question: “In what degree do you associate social networks with voyeurism?” This question would be accompanied by a 5-point Likert scale ranging from 1: “no association at all” to 5: “strong association”. However, at this point, I have some doubts on how is the best way to deal with explicit measures in such experimental design. In particular, do you think it is better to ask for explicit measures at the end or at the beginning of the implicit task? Thanks!
Well, you're correct that self-report is the usual "explicit" measurement technique. But if you want to have a decent explicit measure, you probably ought to use more than a single item. Instead, you should carefully define what you're trying to measure (something you may already have done), write a number of items that tap into said construct, and run a pilot study or so using just the self-report. The results may then be analyzed on the item level (using classical test theory or item response theory), the "bad" items discarded, and the "good" ones retained. ("Good" here means mainly that they correlate with one another to form a coherent scale - for example, in terms of an internal consistency statistic like Cronbach's alpha.)
As for the before vs. after question, perhaps the best advice would be to counterbalance order of presentation so that half the sample gets it one way, the other in reverse.
Hi Stephen, thanks for your reply. I will definetely consider your suggestions.
Best regards,
Antonio
Hi Antonio,
I totally agree with Stephen. You will need a better explicit measure, and then counterbalance the order of presentation.
Best,
Ruben.
Assuming that you're most comfortable in the area of cognitive experimental research, this would be a good opportunity to collaborate with a departmental colleague who has expertise in scale development.
Hi everyone,
I am facing a similar problem. However, in my design I took in account Stephen's and Rubén's suggestions using a multi-item 7-point Likert explicit measure counterbalancing the implicit-explicit order. Nevertheless, I still have doubts regarding the best way to analyze/model the relationship between the implicit measure DV (RT, a continous variable) with the explicit measure DV (7-point Likert means). Any suggestion will be more than welcome. Thanks in advance!