Brief overview of implicit measures

Is current?: 
Yes

Project Implicit offers a few different indirect attitude measures that you can use in your studies. Here is a very brief review about these measures. Later entries will describe how to add those to your studies. For more information, see this page.

 

The most known measure is the Implicit Association Test (IAT). The IAT compares between two pairs of associations. For instance, a Democrats-Republicans IAT compares the pair "Bad with Democrats, Good with Republicans" to the pair "Good with Democrats, Bad with Republicans". The only possible inference from the IAT is that one pair of associations is stronger than another. For instance, that the participant associates "Bad with Democrats, Good with Republicans" more than "Good with Democrats, Bad with Republicans". This is the same concluding that (A+B) > (C+D); There is still a lot that we do not know about A, B, C and D. For instance, it might be the case that a person who really likes Dems, but like Reps a little more will have the IAT score as someone who really hates Reps, but hate Dems a little more. 

 

The Single-Target IAT (ST-IAT, also known as the Single-Category IAT) was developed to measure attitudes toward a single category. It is exactly the same as the IAT, but one category is missing. Let's take for example, an ST-IAT that measures attitudes toward cats. In one block, people use the left key for negative stimuli (usually, unpleasant words), and the right key for positive stimuli or cat stimuli. In the other block, people use the left key for negative stimuli or cat stimuli, and the right key for positive stimuli. If they are faster in the latter than in the former, then it suggests that they dislike cats. 

 

The Go-No-Go Association Task (the GNAT) is very similar to the IAT. In each block, participants see two categories at the top of the screen, e.g., [Democrats or Good words]. Participants press space when they see a stimulus that belongs to one of these categories, and do not press anything when they see stimuli that do not belong to those categories. Discussion about the possible benefits of this task is beyond the scope of this entry.

 

The Brief Implicit Association Test (the BIAT) is a combination of the IAT and the GNAT. In each block, participants see two categories at the top of the screen, e.g., [Democrats or Good words]. Participants press the right key when they see a stimulus that belongs to one of these categories, and the left key when they see stimuli that do not belong to those categories. Discussion about the possible benefits of this task is beyond the scope of this entry.

 

The Evaluative Priming Task (also known as Affective Priming) is an indirect measure that is based on semantic priming procedures. In each trial, participants see two stimuli: a prime stimulus and then the target. Participants are typically instructed to ignore the prime and categorize the target as positive or negative. Often, the targets are words with clear pleasant or unpleasant meaning (e.g., horrible, hell, love, wonderful), and participants simpy categorize to the groups "Good words" or "Bad words" as fast as they can. It was found that it easier to categorize the targets if the prime stimulus matches the valence of the target. That is, people who like Democrats and Dislike Republicans categorize good words faster if the prime was a Democrat (e.g., Clinton) than if the prime was a Republican. That is, people who like Democrats and Dislike Republicans categorize bad words faster if the prime was a That is, people who like Democrats and Dislike Republicans categorize good word faster if the prime was a Democrat (e.g., Clinton) than if the prime was a Republican. (e.g., Bush) than if the prime was a Democrat. 

Evaluative priming often seems like the perfect measure: you can put as many different categories as you want as primes. For instance, you use stimuli of Chinese people, Black people, White people and Aliens and compare between them. Additionally, you don't have to mention the categories at all (simply put a picture of a white person to measure attitudes toward White people), so it is a very indirect measure. Unfortunately, the measure is the least reliable indirect measure that I know.

 

The Affective Misattribution Procedure (the AMP) is also based on priming. In each trial, participant are presented with the prime, a blank screen, a target and then a mask. The prime, blank screen and the target are presented for a very short duration (e.g., 100ms each). Then, the participants are instructed to rate the target as pleasant or unpleasant. The target is usually a Chinese pictograph. Participants are instructed to rate the pictograph as more or less pleasant than the average pictograph. It was found that participants tend to rate the target as pleasant when they like the prime stimulus, and to rate the target as unpleasant when they dislike the prime stimulus. This measure offers the same advantages mentioned before regarding Evaluative Priming, but it much more reliable than Evaluative Priming. More and more researchers are using the AMP as their preferred measure of implicit attitudes. 

The AMP has only one drawback. It seems that the good psychometric qualities of the AMP (reliability and validity) are based on a minority of the sample. That is, perhaps the AMP measures attitudes only for a small number of the participants (less than half). However, many editors have rejected the paper that tried to present these findings, so perhaps those findings are not as important as I think they are. 

 

In each trial of the Sorting Paired Features task (SPF), the participant sees a pair of stimulus and categorizes the stimuli to a pair of categories. For instance, in a Cats vs. Dogs SPF, one stimulus is always a pleasant or an unpleasant word, and the other stimulus is always an image of a cat or a dog. The participants categorize each pair to one of four pairs of categories: Good+Cats, Bad+Cats, Good+Dogs, and Bad+Dogs. There are four key responses, each represents one of the four category pairs. A faster response in a given trial suggests stronger association between the categories in that trial.

The most promising advantage of the SPF is there it measures all the associations in the same block. There is no risk of block-order effects (unlike the IAT, GNAT, BIAT and the ST-IAT). Also, it shows slightly better reliability and validity than Evaluative Priming, perhaps because participants must attend both stimuli. Unfortunately, the SPF is currently less reliable than most of the other measures. 

 

Other entries (will) explain how to add each of these measures to your studies in Project Implicit.