Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/2006 WORKSHOP Implicit Measures Copy of slides available at : Jan De Houwer Ghent University, Belgium Amsterdam ICPS - 13 March 2015
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/2006 Overview 1.Examples of different implicit measures 2.Pros and cons of different implicit measures 3.Issues to consider when constructing your own implicit measure - What do you want to measure? - Select stimuli - Program the task - Collect the data - Analyze the data - Interpret the data Amsterdam ICPS – 13 March 2015
Implicit measures are measurement outcomes that reflect the to-be- measured construct in an automatic manner (i.e., by virtue of processes that are uncontrolled, unintentional, goal-independent, purely-stimulus- driven, autonomous, unconscious, efficient, or fast). (DH & Moors, 2007) Procedure – Person – Outcome Automatic (in certain sense)
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/2006 airplane bad Conscious Goal: go to ICPS Controlled processing Automatic Processing Behavior Amsterdam ICPS - 13 March 2015
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/ Examples of implicit measures 1.1. Implicit Association Test (IAT; Greenwald et al., 1998): => 2909 citations - Procedure: Four categories assigned to two responses - Effect: Faster if associated categories are assigned to same response => see DEMO Task 1: POSITIVE + SMOKING - NEGATIVE + NOT-SMOKING Task 2: POSITIVE + NOT-SMOKING - NEGATIVE + SMOKING Amsterdam ICPS – 13 March 2015
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/ Evaluative priming (Fazio et al., 1986) - Procedure: Respond to target preceded by prime - Effect: Faster if prime and target share same valence Congruent:CANCER-DIRTY Incongruent:CANCER-HAPPY 1.3. AMP (Payne et al., 2005) - Procedure: Evaluate Chinese characters preceded by prime - Effect: valence of prime influence evaluation of character => see DEMO Amsterdam ICPS – 13 March 2015
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/ R-SRC task (Mogg et al., 2003): See Demo - Procedure: Move manikin toward or away from stimuli - Effect: Faster to move to positive and away from negative 1.5. IRAP (Barnes-Holmes et al., 2010) - Procedure: Respond “as if” you believe something - Effect: Faster to respond “as if” if you actually believe it 1.6. RRT (De Houwer et al., in press) - Same idea as IRAP but more user friendly => See talk on Saturday, 14:00, Veilingzaal Amsterdam ICPS – 13 March 2015
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/2006
Many other measures Name letter effect (e.g., Koole, Dijksterhuis, & van Knippenberg, 2001; Nuttin, 1985) Semantic priming (Wittenbrink, Judd, & Park, 1997) Affective Simon effect (De Houwer & Eelen, 1998) Go-NoGo Association Test (GNAT; Nosek & Banaji, 2001) Stereotypic explanatory bias (Sekaquaptewa, Espinoza, Thompson, Vargas, & von Hippel, 2003) Extrinsic Affective Simon Task (De Houwer, 2003) Single-target IAT (Wigboldus, 2001; Karpinski & Steinman, 2006) Implicit association procedure (Schnabel, Banse, & Asendorpf, 2006) Single association test (Blanton, Jaccard, Gonzales, & Christie, 2006) Approach-avoid task (Rinck & Becker, 2007) Sorting paired features task (Bar-Anan, Nosek, & Vianello, 2008) Brief IAT (Sriram, & Greenwald, 2008) Recoding-free IAT (Rothermund et al., 2009)... Amsterdam ICPS – 13 March 2015
2. Pros and cons of different implicit measures 2.1. IAT is still the standard a) Pro: reliable, user friendly, evidence for added value b) Contra: - (sometimes) questions about validity: What is it measuring? (e.g., associations? attitudes? salience? similarity?) - relative (2 target concepts: e.g., alcohol vs soda) => ST-IAT, “not-alcohol” 2.2. Evaluative priming a) Pro: more “personal” and implicit evaluative response? b) Contra: Low reliability Amsterdam ICPS - 13 March 2015 Amsterdam ICPS – 13 March 2015
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/ AMP a) Pro: reliable, user friendly, structurally very different from IAT b) Contra: Automatic? See debate between Bar-Anan vs. Payne. => But still a good additional measure 2.4. R-SRC a) Pro: reliable b) Contra: - not many studies (mostly addiction) - relative (2 target concepts: e.g., alcohol vs soda) => “not-alcohol” Amsterdam ICPS - 13 March 2015 Amsterdam ICPS – 13 March 2015
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/ IRAP a) Pro: can capture beliefs (e.g., I AM good vs I WANT TO BE good) b) Contra: user unfriendly (e.g., 20% dropout in students) 2.6. RRT a) More user friendly way to capture beliefs b) Contra: - not many studies - how implicit is it + what is added value compared to explicit? Use measures that can capture the to-be-measured representations Use several measures!
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/ Issues to consider when constructing your own implicit measure 3.1. What do I want to measure? a) Beliefs vs associations => Most measures capture only associations: I – GOOD => Some measures aspire to measure propositions: I AM GOOD - IRAP: - “propositionalized” IAT: e.g., I like, I want, … b) Relative vs “absolute” associations => IAT captures relative associations (e.g., ME, OTHERS, GOOD, BAD) => Other measures are less relative (e.g., ST-IAT, priming, EAST) but also less reliable => Is any association/belief “absolute”? d Amsterdam ICPS - 13 March 2015
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/2006 c) Personal vs societal views => Measure personal views: Personalized IAT (Olson & Fazio, 2004; - other labels: I Like / I Dislike vs Good / Bad - no error feedback (sometimes) => Measure societal view: Normative IAT (Yoshida et al., 2012) - other labels: Most people like / Most people dislike Amsterdam ICPS - 13 March 2015
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/ Select stimuli (for the IAT) a) Labels - are crucial: IAT measures associations between category labels, not between items in a category (e.g. De Houwer, 2001: British vs. Foreign) - you can use items and instructions to guide the way that participants interpret the labels b) Items - important because they determine interpretation of labels - better fewer good items (e.g., 1, 2, 3, 4) then many bad ones - avoid items that allow for an obvious recoding of labels / task (e.g., perceptual features, familiarity,...) Amsterdam ICPS - 13 March 2015
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/ Write IAT program Many programs available: Inquisit, E-prime, Direct-RT, Affect, … a) Inquisit is easy to use and cheap (but perhaps a bit less flexible) + has version for internet research day trial version - task library => screen shot b) FREE IAT: Meade, A. W. (2009). FreeIAT: An open-source program to administer the implicit association test. Applied Psychological Measurement, 33, 643. Amsterdam ICPS - 13 March 2015
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/2006 Amsterdam ICPS - 13 March 2015
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/2006 c) Affect 4.0: - Spruyt et al. (2010, Exp Psy) - Free - programs and tutorials available - SCR (manikin) & RRT: (downloads) Amsterdam ICPS - 13 March 2015
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/ Collect data a) With enough practice and clear instructions, almost everyone can do most implicit measures (e.g., depressed patients, 10 year olds) b) More practice if young, old, low education, … c) IAT typically has big effect size => no need for big samples (but not bad either) d) Data collection via internet is possible with most implicit measures, also via Apps. Amsterdam ICPS - 13 March 2015
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/ Analyze data a) Technically: - Inquisit output (.dat) file => ASCII - Most basic analysis: Calculate mean RT for each of the two test block - More sophisticated analysis: D600 (Greenwald et al., 2003), but not always the best results => conduct (and report) different analysis => R scripts available for calculating many scoring algorithms (soon on ) b) Item or category specific effects Technically possible but (with IAT) rarely done because effect for one category depends on other category (but Researcher Df) Amsterdam ICPS - 13 March 2015
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/ Interpret the data a) IAT = Relative measure: always different associations involved (e.g., more positive self- than other-esteem) b) Be careful with interpreting absolute direction of effect (zero is not necessarily neutral; Blanton & Jaccard, 2006) c) Don’t make claims about unconscious, “true” knowledge => point is “spontaneous thoughts / feelings” Amsterdam ICPS - 13 March 2015
Distinction Procedure, Effect, and Theory – Jan De Houwer - 09/06/2006 Good luck! Amsterdam ICPS - 13 March 2015