Implicit Association Test (Race)

Introduction

Social psychologists have long been interested in the measurement of human attitudes. Attitudes reflect a general evaluation of a class of individuals or objects (Smith & Fabringer, 2000). Despite the inherent drawbacks, psychologists have traditionally used self-report measures to measure attitudes.

Recently, researchers have begun to use less direct measures of attitudes of the cognitive processes underlying judgment. Greenwald, McGhee and Schwartz (1998) pioneered the use of implicit association tests. Implicit association tests (IATs) measure the relative ease with which people are able to make associations between certain groups of people (e.g., older adults) and the concepts of "good" and "bad." Ease of association, measured by judgment speed, is taken as evidence for an implicitly-held attitude toward that social group. For example, researchers interpret the finding that people are quicker to associate the term "good", with the term "young", rather than with the term "old" as evidence of a generally-held bias in favor of youth. The IAT method is useful for measuring a variety of attitudes including gender, race, and political constructs (Nosek, Greenwald, & Banaji, 2005). The version of the IAT in this experiment measures implicit associations of race, which was one of the first topics investigated by the developers of the IAT--Anthony Greenwald and Mahzarin Banaji (1995).

Design

Participants complete two types of tasks. Each task requires participants to categorize words or pictures. A single keystroke (either 1 or 2) is used to make face categorizations and word categorizations. The first type of task requires correct racial classification of a person in a photograph, as shown in Figure 1. Similarly, participants classify words on the basis of 'good' or 'bad' as shown in Figure 2.


Sample image from the Implicit Associations (Race) experiment
Figure 1

Sample image from the Implicit Associations (Race) experiment
Figure 2

The second task requires each participant to classify a photo or stimulus using a combined set of terms (e.g., is the person in the photograph African American or is the connotation of the adjective Good; is the person in the photograph European American or is the connotation of the adjective Bad). For example, if a photo of an African American is presented as a stimulus, the participant must decide whether to indicate this photo as African American/Good or European American/Bad. An example of this task is presented Figure 3.


Sample image from the Implicit Associations (Race) experiment
Figure 3

In the first set of test trials, European-American is paired with Good and African-American with Bad. This type of test trial is called a "stereotype congruent trial" because the racial stereotypes that have been held historically by the majority culture in the USA have assigned negative attributes to Blacks and positive attributes to Whites. The other type of test trial, illustrated in Figure 3, is a "stereotype non-congruent trial", in which European-American is paired with Bad and African-American with Good. Thus during the course of the experiment, the 1-key might be used for European-American faces and Good words, African-American faces and Good words, or any of the other combinations of Face and Word categories. Based on the data from these two tests, one can determine how the category combinations affect speed of performance.

Example Data Analyses

The data collected from the IAT study is shown in Figure 4.

Sample image from the Implicit Associations (Race) experiment
Figure 4

The first column provides the participant ID number, or the number that each person receives upon completion of the experiment. Data in columns 2-5 are relatively self-explanatory: participant's class affiliation (linked by an Class ID), gender, age, and date of participation.

The column labeled preference is an indication of how the person reported their feelings about individuals on the basis of race. If the participant declined to indicate his or her warmth toward a particular group, you will see the word "Decline" in the column next to the participant's information. The possible values represent the following conditions:

	1 - I strongly prefer Whites to Blacks
	2 - I moderately prefer Whites to Blacks
	3 - I like Whites and Blacks equally
	4 - I moderately prefer Blacks to Whites
	5 - I strongly prefer Blacks to Whites

In the column labeled warmth, participants reported how warmly they feel toward each group. The ratings range from 0 (coldest) to 10 (warmest). The Warmth number is the difference between the two ratings made on the Black and White scales. If the difference score is positive it means the Warmth score was greater for Whites than Blacks. Negative scores imply more warmth toward Blacks and scores of 0 indicate equal warmth toward both races. Finally, data are in columns labeled congruent and incongruent. Data in these columns reflect the average time (averaged across trials) it takes to complete the task. Faster times reflect implicitly held attitudes because the response times are believed to reflect strength of association in semantic memory.

This experiment measures implicit attitudes about race. An investigator can use a repeated measures t-test to compare the effects of the stereotype congruent and the stereotype non-congruent stimuli.

Applications/Extensions

Many people are surprised by the results of the IAT. Someone may not hold explicit biases based on race, yet still pair 'good' terms more quickly with faces representing one race than those representing the other. Most often, particularly when the participant is White, the bias is in the direction of the stereotype-congruent pairing (when white faces are paired with "good" words and black faces with "bad"). This type of activity illustrates the pervasiveness and potential subtlety of cultural stereotypes in our society, and provides opportunity for discussion of factors that may underlie performance on the IAT as well as the nature of the constructs the test purports to measure. More information about the validity of and questions associated with the IAT appears on A. Greenwald's homepage.

References

Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: 
	Attitudes, self-esteem, and stereotypes. Psychological Review, 102, 4-27.
	
Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998). 
	Measuring individual differences in implicit cognition: The implicit
	association test. Journal of Personality and Social Psychology, 74, 1464-1480.

Nosek, B. A., Greenwald, A. G., & Mahzarin, B. R. (2005).
	Understanding and using the implicit association test: II. Method variables and
	construct validity. Personality and Social Psychology Bulletin, 31, 166-180.

Smith, S. M., & Fabringer, L. R. (2000). Attitudes: An overview. In A. E. Kazdin (Ed.), 
	Encyclopedia of Psychology (pp. 303-305). Washington, DC: APA.