UCL survey

[tristessa363]

If you took the Etter survey, they probably sent you a link to the UCL survey attitudestosmoking.com (ucjtcpm@ucl.ac.uk). This is a four-part survey is designed to interpret your responses to visual images of cigarettes/smoking. they change up the order but basically they show you the same 20 images...

some are of smoking/cigarettes
some have nothing to do with cigarettes

In the first quarter you are told to press a key designating the image of a cigarette as Like vs. Dislike.
In the second quarter they reverse this designation.
In the third quarter you are told to press a key designating the image of a cigarette as Good vs. Bad.
In the fourth quarter they reverse this designation.

I took it twice, so I know that they switch the order around.
At the end they state:
"If your response time for Like was faster than Dislike this suggests a positive implicit emotional attitude to smoking.
If your response time for Good was faster than Bad this suggests a positive moral attitude to smoking."

The first time I took it, the result was that I was morally against smoking (I am not - it was the first round of the test and I hadn't gotten the hang of it).

I took it again (having learned to look for the "correct" word vs. the location of the word/picture) and these were my results:

Average response time (in seconds):
Like for smoking: 0.452
Dislike for smoking: 0.498
This "suggests a positive implicit emotional attitude to smoking."

Good for smoking: 0.496 seconds
Bad for smoking: 0.915 seconds.
This "suggests a positive moral attitude to smoking."

Seriously? It's not about your attitude towards smoking, it's about reaction time and memory. I did what I was told; I did a second test and "improved." That has to say something about this survey's validity.

 

[Oliver]

You would always improve if you took pretty much any test you were trying to 'beat' again. That says nothing about the validity of the test.

I would say, however, that discussing the methodology may well produce confounded results if people go on to take the test having already read what's involved. Although these types of inplicit association test (IAT) tend to be quite robust.

There's also a whole series of other IATs you can take on Harvard's website: https://implicit.harvard.edu/implicit/demo/