The Project on Law and Mind Science recently launched a policy-oriented implicit association test (IAT).
The IAT is an experimental method designed to measure associative information that people are either unwilling or unable to report. The test was first published by Greenwald and colleagues in 1998. The IAT builds on the implicit-explicit distinction in memory. It reflects the observation that because much social cognition occur in an implicit mode, measuring unconscious cognition likely provides the “missing ingredient” necessary to support efficient testing and development of psychoanalytic, behaviorist, and cognitive theories.
Development & Status of the Implicit Association Test
PRE-IMPLICIT ASSOCIATION TEST RESEARCH: 1990 – 1997
In the early 1990s, Greenwald and Situationist contributor Mahzarin Banaji, in collaboration with others, co-authored a number of papers that explored the unconscious operation of stereotyped beliefs, prejudicial attitudes and discriminatory behavior in the context of social groups, personal judgment, and gender. The authors also investigated automatic stereotyping and implicit social cognition generally. Greenwald proposed in a 1990 article that robust attitude effects can be easily discovered when attitude is involved only indirectly. In 1993, Banaji and others refined Greenwald’s 1990 proposal, arguing that stereotype information is influential only when the social category of the target makes the information relevant to judgment. For example, whereas priming individuals with stereotypes of aggression influences judgments about men, it has little influence on judgment about women. Alternatively, judgments about women are influenced when individuals are primed with stereotypes for dependence. In that 1993 article, Banaji and her co-authors also argued that an individual’s social category influences use of primed stereotype information.
BIRTH OF THE IMPLICIT ASSOCIATION TEST: 1998
Greenwald, McGhee, and Schwarz first published the IAT in 1988. Since then, “the IAT has been used repeatedly to measure implicit attitudes and other automatic associations.” IAT subjects sort stimuli representing four categories using two responses: usually keys on a computer keyboard. Each response is assigned to two of the four categories. The test rests on the assumption that participants will sort stimuli faster where they experience paired categories as more closely associated.
In their 1998 paper, Greenwald and colleagues presented three experiments that showed that when highly associative categories (e.g. flower + pleasant) share a response key, performance on the test is faster than when less highly associated categories share the same key. In the first experiment, subjects were asked to sort stimuli for flowers, musical instruments, insects, and weapons. This experiment tested the hypothesis that associations can be revealed by mapping two discrimination tasks alternately onto a single pair of responses. The results confirmed this assumption– superior performance was observed when associatively compatible categories were mapped onto the same response. The second experiment extended the test to a domain more typically attitudinal than the first experiment: using the test to discriminate differences between Japanese Americans and Korean Americans in their evaluative associations toward Japanese and Korean ethnic groups. The researchers hypothesized that ethnically Korean subjects would find it more difficult to perform the Japanese + pleasant than the Korean + pleasant combination and vice versa. This hypothesis was based on the history of Japanese-Korean antagonism. The results of the IAT confirmed the expected pattern. The third experiment combined the tasks of classifying Black and White names and discriminating pleasant versus unpleasant word. Twenty-six White American students participated in this test. At the end of the test, each subject responded to questionnaire measures for race-related attitudes and beliefs. The results of the third experiment “clearly revealed patterns consistent with the expectation that White subjects would display an implicit attitude difference between the Black and White racial categories. More specifically, the data indicated an implicit attitudinal preference for White over Black, manifest as faster responses for the White + pleasant combination than for the Black + pleasant combination.”
Since its initial publication in 1988, the IAT has been applied in a diverse array of disciplines including social and cognitive psychology, clinical psychology, developmental psychology, neuroscience, market research, health policy, business and consumer research, and law.
Policy Application for the Implicit Association Test
Although since its initial publication in 1988, the IAT has been applied in a diverse array of disciplines, application to the legal and policy arenas has been minimal. This is the case even though the dominant schemas that shape law and policy are not unlike attitudes, stereotypes and other forms of implicit cognition that IAT is so often harnessed to measure. In a series of articles, Situationist contributor Jon Hanson and his collaborators have endeavored to identify the dominant knowledge structures, schemas, and categories that shape law and policy. Based on their research, the most influential policy scripts boil down to a very simple two-part proposition: “markets are good, regulation is bad.” It seems likely that the IAT can shed light on this dominant policy script.
Hanson and Situationist fellow Mark Yeboah have developed an IAT-based study to investigate the the presence and strength of policy scripts across the ideological spectrum. The experiment hopes to shed light on a number of questions, including the extent to which implicit associations correspond with explicit attitudes about markets and regulation and how those associations might vary across various ideological and political dimensions.
* * *
To participate in the Policy IAT, click here.
To read a fascinating interview of professors Banaji and Greenwald about the history and significance of the IAT, go to the five-part series: Part I is here; Part II is here; Part III is here; Part IV is here; and Part V is here.
To review all of the previous Situationist posts discussing implicit associations click on the “Implicit Associations” category in the right margin, or, for a list of such posts, click here.