Implicit Association Test

Implicit Association Test: An Overview : 

Implicit Association Test: An Overview Doug Ewing February 23, 2010

Overview : 

Overview Implicit Attitudes and Measures Running an IAT Considerations Administration Prepping for Analysis IAT Issues & Criticisms

Implicit Attitudes and Measures : 

Implicit Attitudes and Measures

Implicit Attitudes : 

Implicit Attitudes Potentially different from stated (i.e. explicit) attitudes. Generally considered to be less accessible than explicit. Underlying +/- associations. “Real” attitudes. Slower to form and change. Less vivid but persistent. Guide behavior under cognitive constraint . A source of cultural influence on behavior Moors and DeHouwer (2006); Rydell and McConnell (2006); Buttle et al. (2005); Gibson (2008); Rudman (2004); Olson and Fazio (2009).

Implicit Attitudes : 

5 Implicit Attitudes Associative-Propositional Evaluation (APE) Gawronski and Bodenhausen (2006) Implicit and explicit attitude change: Explicit att. via “syllogistic reasoning”. Implicit att. via change in association and/or activation. May converge/diverge and/or mediate one another. Relevant to priming, conditioning, persuasion, etc.

APE Model : 

6 APE Model Source: Gawronski and Bodenhausen 2006, p.697

Implicit Methods : 

Implicit Methods Implicit Association Test (IAT) Greenwald et al. (1998); Greenwald et al. (2003) Brief and Single Category IATs Sriram and Greenwald (2009); Karpinski and Steinman (2006) Go/No-go Association Task Nosek and Banaji (2001) Affective Priming Task Fazio et al. (1995) Extrinsic Affective Simon Task DeHouwer and Eelen (1998)

IAT : 

IAT 8 Coke or Good Pepsi or Bad Pepsi or Good disgusting E I E I Key: Coke or Bad

IAT : 

Source: Greenwald, Nosek, and Banaji (2003) IAT

SC-IAT : 

SC-IAT 10 Gallo or Good Bad Good Gallo or Bad disgusting E I E I Key:

IAT/SC-IAT : 

IAT/SC-IAT What is being measured? Essentially, strength of associations in memory. Shorter response times indicate stronger associations. Implicit attitude given by relative strength of +/- associations. May or may not be “implicit” (cf. DeHouwer 2009).

Running an IAT : 

Running an IAT

Running an IAT : 

Running an IAT Preliminary Considerations Software DirectRT, Inquisit. Pencil-and-paper IAT is a possible option. Ss Familiarity w/ IAT Stable lab environment.

Running an IAT : 

Running an IAT Preliminary Considerations Matched pairs Comparable attitude objects (e.g. brands, products, people, etc.) IAT vs. SCIAT Stimuli easy to process? Pictures vs. text. Comparability of pictures.

Running an IAT : 

Running an IAT Confronting DirectRT Something of a “blackbox” program. Runs in a Direct-X environment. Editing and output mostly separate from program. Relatively simple and straightforward to implement. Excel rather than MediaLab interface hassles. You can see and edit exactly what DirectRt will present to Ss. Dealing with output can be intimidating. Need to go from raw response times to an IAT score. Occasionally some detective work required.

Slide 16: 

Input .csv File Output .csv Files

Running an IAT : 

Running an IAT Most editing of an IAT procedure uses Excel. A “.csv” file provides input to DirectRT. DirectRT accesses a “stim” directory for all on-screen pictures, words, etc. Style (e.g. color, font) edited in Direct RT. MediaLab is not required for an IAT but integrates well if needed.

Slide 20: 

Input .csv File Output .csv Files

Stim Directory : 

Stim Directory Instructions Simple .txt files Introductions, transition instructions, reminders Lists Simple .txt files Word lists, image lists Random sampling (w/ or w/o replacement), draw in order, etc.

Stim Directory : 

Stim Directory

Stim Directory : 

Stim Directory

Stim Directory : 

Stim Directory “Good” Words ~beautiful ~celebrating ~cheerful ~excellent ~excitement ~fabulous ~friendly ~glad ~glee ~happy ~laughing ~likable ~loving ~marvelous ~pleasure ~smiling ~splendid ~superb ~paradise ~triumph ~wonderful “Bad” Words ~angry ~brutal ~destroy ~dirty ~disaster ~disgusting ~dislike ~evil ~gross ~horrible ~humiliate ~nasty ~noxious ~painful ~revolting ~sickening ~terrible ~tragic ~ugly ~unpleasant ~yucky

Slide 26: 

Input .csv File

Slide 28: 

Output .csv File

Prepping for Analysis : 

Prepping for Analysis Merge individual Ss data files DirectRT provides a utility for this purpose. Many files in  one file out. Choose Scoring Method(s) Greenwald et al. (2003) provides a variety of scoring methods with recommendations. Karpinski and Steinman (2006) if SC-IAT.

Prepping for Analysis : 

Prepping for Analysis Check to make sure DirectRT ID and Medialab ID match in all cases. Compute IAT Scores for all Ss Prepare to get intimate with the SPSS COMPUTE and AGGREGATE commands. If possible, consider using Greenwald’s syntax available on his website.

Slide 32: 

Source: Greenwald, Nosek, and Banaji (2003)

Slide 33: 

Input .csv File Output .csv Files

IAT Issues & Criticisms : 

IAT Issues & Criticisms

IAT Issues & Criticisms : 

IAT Issues & Criticisms Not amenable to “real world” use or field studies. Time and facility intensive ($$$). Needs a relatively specific environment. Possible to “fake” IAT responses. Ss often are aware of what an IAT is doing. With some effort, evidence suggests IAT responses can be influenced (DeHouwer et al. 2007)

IAT Issues & Criticisms : 

IAT Issues & Criticisms IAT and implicit phenomena. IAT does not necessarily imply measurement of implicit phenomena. Other ways to assess effects of implicit phenomena. The implicit/explicit distinction. Always worthwhile? Remember, IAT has many uses but it is only a method.