|
| Implicit Testing in Behavior Analysis: Where’s The Science? |
| Monday, May 25, 2009 |
| 3:00 PM–4:20 PM |
| North 131 A |
| Area: CSE; Domain: Experimental Analysis |
| Chair: Amanda Gavin (University of Tesside) |
| Abstract: The current series of papers together provide an outline of a behavior-analytic research program into means of “implicit” testing without the necessity for mentalistic concepts or the use of poorly understood stimulus presentation and response derivation methodologies. The first paper outlines a series of behavior-analytic concerns over the widely employed Implicit Association Test and provides several suggestions for researchers interested in understanding and developing implicit tests that are acceptable to behavior analysts. The second paper reports on an experiment that illustrates how implicit test effects are both easily malleable and have a questionable ability to make reliable behavioral predictions. The third paper provides a theoretical model, along with supporting empirical evidence, that a behavior-analytically understood test can effectively predict and explain subjects’ performances during a computer game designed to test racial prejudice. Finally, the fourth paper, describes an experiment that aimed to assess the effectiveness of a novel behavioral test at identifying differences in verbal histories of a sample of male and female subjects. This test represents a functionally-understood counterpart to several popular implicit tests. |
| |
| “Implicit” Testing: No Science, No Process, No Function. |
| BRYAN T. ROCHE (National University of Ireland, Maynooth), Maria R. Ruiz (Rollins College), Amanda Gavin (University of Tesside) |
| Abstract: The current paper outlines the Implicit Association Test and reviews its meteoric rise in popularity in the absence of a process-based account. Most users of the test understand little about it except how to administer the test and generate scores. From a behavioral perspective, the test appears to serve as little more than a test for the relative rates of response function acquisition by pairs of verbally related and unrelated stimuli. These differences point to variances in the histories of specific verbal stimulus classes, such as differences in class strength. They do not reasonably point to such constructs as “unconscious bias” as users of the test widely claim. Moreover, the test is imbued with procedural artifacts and statistical scoring techniques that blur the reported reaction time and accuracy measures leading to contrived behavioral indices that can not be accurately described as response times, response accuracy, response rate, or even fluency, even though they may be presented a such. Now that implicit testing has begun to be employed within the field of behavior analysis, researchers need to be keenly aware of the limitations of these tests as a measure of anything at all of interest to psychologists, let alone behavior analysts. |
| |
| IAT Effects and Behavioral Probability: How Related Are They? |
| MARIA R. RUIZ (Rollins College), Bryan T. Roche (National University of Ireland, Maynooth), Sara Jacobsen (Rollins College), Melissa Bernardo (Rollins College), Amanda Gavin (University of Tesside) |
| Abstract: Subjects were first exposed to a word-picture association training phase using a respondent conditioning preparation. Specifically, two nonsense syllables were paired with images of plants and animals, respectively, as well as with the colors red and blue, respectively. Subjects were also exposed to an equivalence training procedure which led to the formation of two three-member equivalence relations, each containing one of the two nonsense syllables. They were then exposed to a derived transfer of functions test that probed for the derived transfer of stimulus functions (i.e., colors and images) to other equivalence class members. Subjects were then exposed to an IAT-type test consisting of images and stimuli from the established stimulus equivalence relations. The laboratory controlled history led to the successful creation of a non-socially-established IAT effect. This effect was then undermined across repeated exposures to the IAT (i.e., via practice effects), after which the test for derived transfer of functions was re-administered. The results show that the power of the IAT to predict responses during the derived transfer of functions test is easily compromised through simple laboratory interventions. |
| |
| Can the IAT Predict the Decision to Shoot in the Shooter’s Game?: A Behavioral Model |
| Maria R. Ruiz (Rollins College), Bryan T. Roche (National University of Ireland, Maynooth), MELISSA BERNARDO (Rollins College), Sara Jacobsen (Rollins College), Amanda Gavin (University of Tesside) |
| Abstract: In the Shooter’s Game (Correll, Park, Judd & Wittenbrink, 2007), subjects are instructed to shoot images of Caucasian or African American males holding a gun. Subjects are required to make rapid decisions whether or not to shoot across trials in which images are presented of Caucasian and African American males holding a wallet, a beer can or a gun. More erroneous shooting responses are made when subjects are presented with the African American compared to the Caucasian. The current paper presents a behavioral model of the Shooter Game effect in which it is conceptualized in terms of contextual control over derived classes of discriminative stimuli for avoidance or aggression. The model also allows behavioral predictions to be made regarding responses during game play on the basis of IAT-style test results. Preliminary data supporting the model will be presented.
References
Correll, J., Park, B., Judd,C. M., Wittenbrink, B. (2007). Across the thin blue line: Police officers and racial bias in the decision to shoot. Journal of Personality and Social Psychology, 92, 1006-1023. |
| |
| A Transparent and Coherent Implicit Test for Verbal History: Just Functional Behavior Analysis, No Sleight of Hand. |
| AMANDA GAVIN (University of Tesside), Bryan T. Roche (National University of Ireland, Maynooth), Maria R. Ruiz (Rollins College) |
| Abstract: The current paper reports on an experiment that demonstrates the utility of a functionally-understood test for verbal history. The test is procedurally transparent, relies on traditional behavioral principles and is free from theoretical constructs. Specifically, the test assesses the relative rates of acquisition of common response functions to word pairs considered compatible for a normal population compared to words considered incompatible. Subjects (N=20) were required to respond to exemplars of child, adult, sexual and nonsexual stimulus categories in one of two specified ways. For exemplars of two of the four categories presented on a computer screen, subjects were instructed to respond with a red key press, while for exemplars of the remaining two categories subjects were instructed to respond with a blue key press. In another block of testing the requirements were juxtaposed so that the combination of stimuli requiring a common key response was altered. There was more rapid acquisition of common response functions across suspected compatible exemplars (child and nonsexual) than incompatible (child and sexual) exemplar pairs. This test provides a behavior-analytic counterpart to the Implicit Association Test and other similar tests. |
|
| |