Go/No-Go Association task:

The GNAT (pronounced like the bug) is a flexible technique designed to measure implicit social cognition. Conceptually similar to other implicit measures like the Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, JPSP, 1998), the GNAT assesses automatic associations between concept (e.g., gender) and attribute (e.g., evaluation) categories. The GNAT has two features that distinguish it from other measures of implicit social cognition. First, the GNAT is designed to be use signal detection statistics in its calculation of automatic associations (d-prime), but can also be adapted to utilize response latency as its operational dependent variable. Second, the GNAT is flexible in the establishing of contextual characteristics for the evaluative situation. For example, the IAT requires an attitude toward one category (insects) be assessed relative to a second category (flowers). With the GNAT, experimenters can vary whether insects are evaluated in the context of a single category (flowers), a superordinate category (animals), a generic category (objects), or with no context at all.

The GNAT is often used with the intention of obtaining an implicit measure of self-esteem, sexism, racism, aggression, and so on. In some cases, test takers may be motivated to score in a particular way on the GNAT (e.g., not implicitly sexist or racist).


  • To what extent can the GNAT be faked?
  • If an individual does fake the GNAT, can the faking be detected?

1 Answer 1


I've never looked at the GNAT before but, even though it's just about as transparent as IAT, the critical thing making it difficult to fake bias is the deadlining. In order to fake one would would have to respond in no-go trials where one would not and that also misrepresented their association. That calculation would be difficult to do in the time window and possibly result in overall very high error rates. I'm not a fan that they just throw out RTs. Furthermore, calculating d' values does not by itself make a task better.

There's a general problem with many of these tasks in that the requested response is directly along the tested dimension. What's implicit about that? Asking you if bad goes with bug does not make for an implicit tests. However, with something like the ALPS (Lebrecht et al 2009) you select an orthogonal dimension for the response (lexical decision in that case). That seems much more like an implicit task and since people tend to be very response focused it makes it much less likely that subjects will pick up on how the test works and attempt to manipulate it. You could also deadline that to make it even less likely that it's cheated.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.