'Indirect Measures of Social identities: pre-registered experiment' (AsPredicted #141,514)
Author(s) This pre-registration is currently anonymous to enable blind peer-review. It has 4 authors.
Pre-registered on 2023/08/22 00:27 (PT)
1) Have any data been collected for this study already? No, no data have been collected for this study yet.
2) What's the main question being asked or hypothesis being tested in this study? The present research tests whether two established indirect measures of social identity differentiate between social identification and disidentification. The study manipulates the meaning of the group and test its effects on an identity IAT and the MMP. We add a non-identification condition for explorative purposes because non-identification theoretically is the zero point of social identification and disidentification. Moreover, we add a non-categorization condition where people are not member of any of the social groups. This results in four levels of the meaning of the group (disidentification vs. non-identification vs. social identification vs. non-categorization). We predict that both the MMP and the identity IAT differentiate between the ingroup conditions and the non-categorization condition, while only the MMP will differentiate between disidentification and identification with the ingroup.
3) Describe the key dependent variable(s) specifying how they will be measured. 1) The identity IAT-D-effect: The identity IAT follows the procedure by Greenwald et al. (2002). We compute the IAT-D-effect following the improved scoring algorithm. More positive IAT-D-effects indicate stronger associations of self and ingroup (and other and Japanese) than self and Japanese (and other and ingroup).
2) The ingroup and outgroup match-mismatch effects in the Match-mismatch paradigm: The MMP will be conducted similar to the paradigm by Smith and Henry (1996). We compute the ingroup and outgroup match-mismatch index from the response latencies (see Coats et al., 2000; Sassenberg & Matschke, 2010). To do so, ratings for the groups and the self are dichotomized (1-3 = no, 5-7 = yes). The reaction-time based group match-mismatch effect is computed by subtracting the mean response times to matches from the mean response times to mismatches. Thus, for every participant, the content of the matches and mismatches is an individual set of traits. Higher values of the match-mismatch effect are commonly interpreted as stronger inclusion of the group into the self-concept.
In addition, we assess social identification with 15 items (Leach et al., 2008; Roth & Mazziotta, 2015) and measure disidentification with the 11-item scale of Becker and Tausch (2014).
4) How many and which conditions will participants be assigned to? There will be four levels of the meaning of the group: disidentification (DIDC) vs. non-identification (NIDC) vs. social identification (SIDC) vs. non-categorization (NCATC).
5) Specify exactly which analyses you will conduct to examine the main question/hypothesis. 1) A MANOVA with Meaning as independent variable and self-reported social identification and disidentification as dependent variables will test whether the conditions differ significantly in social identification and disidentification.
2) An ANOVA with the IAT effect as dependent variable and Meaning as independent variable will be conducted in order to test differences between the conditions.
3) To test whether the match-mismatch ingroup (but not outgroup) effect differs between the conditions, we conduct a mixed model ANOVA with the between-group factor Meaning of the Group and the repeated measure factor Group (i.e., ingroup vs. outgroup).
6) Describe exactly how outliers will be defined and handled, and your precise rule(s) for excluding observations. We apply the following exclusion criteria: participants (1) who do not follow instructions (i.e., who indicate more than one social group and those who are not a member of the indicated group), (2) who show less than 10 ratings above or below 4 (i.e., indifference) in the trait assessment of ingroup, outgroup, or self, as part of the MMP (3) who have missing values for either matches or mismatches for the ingroup or outgroup (i.e., match-mismatch indices cannot be calculated), (4) who are outlier in the MMP index (> 3 SD, Sassenberg & Matschke, 2010), and (5) who have more than 10% reaction times below 300 milliseconds in the IAT (Greenwald et al., 2003). In addition, for the calculation of the individual match-mismatch index, reaction times below 300 and above 5000 milliseconds are excluded from the calculation (Smith & Henry, 1996). Finally, for the unlikely circumstance that anyone of the participants is a member of the specified outgroups, we add the exclusion criteria of not being Vietnamese or Japanese along with the added exclusion criteria of being a member of the ingroup in all ingroup conditions in this experiment.
7) How many observations will be collected or what will determine sample size? No need to justify decision, but be precise about exactly how the number will be determined. We conducted separate a priori power analyses (Faul et al., 2007) for the predicted effect in the MMP and in the identity IAT. A priori power analyses for between-within interaction require the estimation of additional parameters, which comes with added uncertainty and may result in too optimistic sample-size estimates. Following recommendations of Döring and Bortz (2016; p. 848), we chose a conservative way to analyze power by conducting a between subjects ANOVA for the MMP. This way, the planned power will be achieved irrespective of the factual correlation between repeated measurements. This ANOVA (fixed effects, special, main effects and interactions) with α = .05, Power = .90, and an estimated effect size of ηp2 = .09 (based on the smallest effect size in the pilot data), numerator of the interaction df = 3, number of groups = 8, for the interaction of Meaning (DIDC, NIDC, SIDC, NCATC) × Group (ingroup vs. outgroup), indicated a necessary sample size of N = 148. The power analyses for the follow up pairwise comparisons with a t-test for independent means (effect size d = 3.64, based on the smallest effect size in the pilot data, α = .05, Power = .90, allocation ratio = 1), and the priori power analyses for the overall effect in the ANOVA to analyze the IAT effect (with α = .05, Power = .90, and an effect size of ηp2 = .32) indicated smaller sample sizes. To allow for about 20% of exclusions, and to power the analyses adequately for the analyses with the weakest effect, we aim to assess a sample size of N = 178.
8) Anything else you would like to pre-register? (e.g., secondary analyses, variables collected for exploratory purposes, unusual analyses planned?) Nothing else to pre-register.
