Author(s) Marco Radojevic (University of Zurich) - radojevic@ipz.uzh.ch Alberto Lopez Ortega (University of Zurich) - a.lopez.ortega@vu.nl
Pre-registered on 2021/05/12 08:42 (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? Do voters discriminate against people of non-german background, transsexual politicians and gay politicians and do these forms of discrimination interact?
We expect the discriminatory preferences to be lower if explicit labels rather than implicit attributes are shown to respondents (social desirability): We expect voters to be more favourable to the aforementioned attributes (non german ethnicity etc.) in experimental group 1a and 1b than in group 2a and 2b.
We also want to explore the relationship of ideology and discrimination.
We have one main prediction:
Once party labels are introduced we expect left wing voters to prefer 'diverse' (non-german, LGBTQI, female) right wing persons and we expect right wing voters to prefer 'non diverse' (German, heterosexual, male) leftists. We expect diversity to be used as heuristic for left leaning positioning ceteris paribus.
We also want to add to intersectional theory of discrimination by testing a multiplicative vs. a additive view of intersecting discriminatory experiences.
We expect ideology to overcome discrimination: Once party labels are introduced (group 1b and 2b) the effect of the aforementioned attributes should be lower.
3) Describe the key dependent variable(s) specifying how they will be measured. We analyse our data in two stages. We have 4 experimental groups in which we all employ a conjoint analysis. Within each group respondents have to chose between two candidates in 5 rounds. They either have to chose 'Person A' or 'Person B'. The dependent variable is the mean likelihood of a candidate profile being selected given a certain set of profile attributes. (Marginal Means / Average Marginal Component Effect). In a second step we are also interested, in how the different experimental conditions influence the marginal means in the first stage of the experiment. We will explore whether the marginal means / average marginal component effect is significantly different depending on information shown (See Hypothesis and Conditions)
4) How many and which conditions will participants be assigned to? Participants will be assigned randomly to one of four groups. In the following, we explain what information will be displayed to the respondents in each group. The attribute levels (in brackets) of each conjoint attribute within each group will be randomly assigned and profiles will be generated out of this.
Group 1a 'Explicit labels': Name (Turkish Name vs German Name), Gender (Male, Female, Other), Sexual Orientation (Heterosexual, bisexual, homosexual), Political Experience (4 Years in a Local Council, 8 Years in a Local Council; 12 Years in a Local Council), Political Record (Kept 50 Percent of Political Promises; Kept 70 Percent of Political Promises, Kept 90 Percent of Political Promises). All attributes are displayed explicitly.
Group 1b 'Explicit labels + Party Labels': Name, Gender, Sexual Orientation, Political Experience, Political Record, Party (CDU / CSU. Wants the CDU / CDU to become politically more right wing, CDU / CSU. Wants the CDU / CDU to become politically more centrist, Greens. Wants the GREENS to become politically more centrist, Greens. Wants the GREENS to become politically more left wing. All attributes are displayed explicitly.
Group 2a 'Implicit labels': Picture of a person that contains information implicitly (Ethnicity, Gender, Sexual Orientation) + Political Experience, Political Record are displayed explicitly
Group 2b 'Implicit labels + Party Labels': Picture of a person that contains information implicitly (Ethnicity, Gender, Sexual Orientation) + Political Experience, Political Record + Party are displayed explicitly
5) Specify exactly which analyses you will conduct to examine the main question/hypothesis. We calculate how each profile attribute level affects the likelihood of a person profile being selected by the respondent (marginal means and average marginal component effects). We will than compare the marginal means and average marginal component effect between experimental groups and test whether a significant difference exists between them.
To test multiplicative vs. additive views on intersecting discrimination, we estimate interactions between different attributes which are associated with discrimination. If we find interactions, this supports a multiplicative view of intersecting discriminations.
6) Describe exactly how outliers will be defined and handled, and your precise rule(s) for excluding observations. We will include all possible observations in each analysis. We exclude people that failed an attention check (2+2) .
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 have a sample size of 2300 adult individuals from Germany. Depending on negotiations with the survey company sample might be 100-200 respondents smaller.
8) Anything else you would like to pre-register? (e.g., secondary analyses, variables collected for exploratory purposes, unusual analyses planned?) We include a battery on the party choice and ideology (0-10 Left Right Scale) of each respondent and we record demographic variables (Gender, Sexual Orientation, Migration background). Party Thermometer (How likely it is that you will vote for Party x)
