Author(s) Daniel Effron (London Business School) - deffron@london.edu
Pre-registered on 02/07/2017 10:44 AM (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? We have previous found that members of entitative groups, compared to members of non-entitative groups, receive greater license to express prejudice against outgroups. The purpose of this study is to (a) replicate this effect, and (b) examine the role of collective responsibility attributions in this process.
In overview, participants will be randomly assigned to one of six conditions in a 3x2 between-subjects design: (high collective responsibility vs. low collective responsibility vs. control) X (high entitativity vs. low entitativity). We will measure entitativity, collective responsibility, and license.
We have formulated the following hypotheses:
- H1: In the control condition, the entitativity manipulation will increase collective responsibility. This effect will be significantly smaller or absent in the high and low collective responsibility conditions combined.
- H2: In the control condition, the entitativity manipulation will increase license. This effect will be significantly smaller or absent in the high and low collective responsibility conditions combined.
- H3: The effect of entitativity on license in the control condition will be mediated by the collective responsibility measure. That is, in the control condition, the entitativity manipulation will increase collective responsibility, which will lead to greater individual license.
3) Describe the key dependent variable(s) specifying how they will be measured. We will include three main measures:
- Entitativity (6-item scale from Denson et al., 2006)
- Collective responsibility (7 items, i.e., how responsible the group should be held for each of 7 prejudiced behaviors)
- License (7-item scale from Effron & Knowles, 2015, i.e., how socially acceptable the average participant would think it is for a group member to commit each of 7 prejudiced behaviors)
4) How many and which conditions will participants be assigned to? There are six conditions total in this 3x2 between-subjects design: (high collective responsibility vs. low collective responsibility vs. control) X (high entitativity vs. low entitativity)
5) Specify exactly which analyses you will conduct to examine the main question/hypothesis. Before testing our hypotheses, we will conduct a manipulation check with a t-test. We expect entitativity ratings to be significantly higher in the high-entitativity condition than in the low-entitativity condition.
We will test the hypotheses with linear regression analysis including two planned, orthogonal contrasts for the collective responsibility conditions (contrast_1 compares the high- and the low-responsibility conditions, each coded +1, to the control condition, coded –2; contrast_2 compares the high- and low-responsibility conditions to each other, coded +1 and –1, respectively), an effect code for entitativity (high = +1, low = –1), and the two two-way interactions between entitativity and each of the contrasts for collective responsibility.
To test H1, we will submit the collective responsibility measure to this regression analysis. We expect the coefficient for the contrast_1 X entitativity interaction to be significant and negative, indicating that the effect of entitativity on collective responsibility is more positive in the control condition than in the other two conditions. We also expect that the simple slope of entitativity in the control condition will be significant and positive, indicating that control participants gave higher collective responsibility ratings in the high-entitativity condition than in the low-entitativity condition.
To test H2, we will submit the license measure to the same regression analysis. Once again, we expect the coefficient for the contrast_1 X entitativity interaction to be significant and negative, indicating that the effect of entitativity on license is more positive in the control condition than in the other two conditions. We also expect that the simple slope of entitativity in the control condition will be significant and positive, indicating that control participants gave higher license ratings in the high-entitativity condition than in the low-entitativity condition.
Because we are pre-registering these directional hypotheses, we will test significance with one-tailed tests.
Note that we have no hypotheses about whether the effect of entitativity on either responsibility or license will differ between the high- vs. the low-responsibility conditions. Thus, in both regression analyses described above, the test of the contrast_2 X entitativity interaction is exploratory.
To test H3, we will conduct a mediation analysis using the sgmediation function in Stata, limiting the data to participants in the control condition. The IV will be the entitativity condition (effect-coded as above), the mediator will the collective responsibility measure (mean-centered), and the DV will be the license measure. To test the significance of the indirect effect, we will compute the bias-corrected 95% CI using 5,000 bootstrap samples.
6) Any secondary analyses? We also expect the collective responsibility manipulation to have a linear main effect on the license measure, such that people grant the least amount of license in the low-responsibility condition and the most in the high-responsibility condition, with the control condition falling in between.
This prediction is contingent on the collective responsibility manipulation having a linear effect on the collective responsibility measure (manipulation check).
We will test these predictions with a linear regression model containing the following IVS: a linear contrast for the responsibility manipulation (low responsibility = –1, control = 0, high responsibility = +1), and an effect code for the entitativity manipulation as a covariate.
As a manipulation check, we will first run this model with the collective responsibility measure as the DV. We expect a significant positive linear trend of the responsibility manipulation on the responsibility measure.
We will next run the same model with license as the DV. Again, we expect a significant positive linear trend of the responsibility manipulation.
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 will post slots for 1,000 participants on MTurk, and we will run until all slots are filled.
8) Anything else you would like to pre-register? (e.g., data exclusions, variables collected for exploratory purposes, unusual analyses planned?) We will exclude data from people who:
- Fail an attention-check question that asks people to identify whether the “Ebbites” are a religious group, an ethnic group, a national group, or none of the above.
or
- Submit data from duplicate IP addresses or from a duplicate MTurk ID (in these cases, the first set of data submitted will be retained)
or
- Have non-US IP addresses
or
- Do not complete 25% or more of the scale items for any individual dependent measure
