'ORE-2 -- Stereotypes about required effort -- March 2023' (AsPredicted #124627)
Author(s) Bethany Lassetter (New York University) - bethany.lassetter@nyu.edu Natalie Hutchins (New York University) - natalie.hutchins@nyu.edu Vivian Liu (New York University) - vl845@nyu.edu Natalie Toomajian (New York University) - nt2190@nyu.edu Andrei Cimpian (New York University) - andrei.cimpian@nyu.edu
Pre-registered on 2023/03/09 - 06:53 PM (PT)
1) Have any data been collected for this study already? It's complicated. We have already collected some data but explain in Question 8 why readers may consider this a valid pre-registration nevertheless.
2) What's the main question being asked or hypothesis being tested in this study? STEREOTYPES: We will assess how hard children believe boys and girls have to work (i.e., how much effort they have to put in) to do well in math and reading/English (hereafter referred to as "reading"). We predict that children will report that girls have to work harder than boys in math, and potentially in reading as well. We will explore whether these stereotypes are stronger for older or younger children, and whether they hold more strongly by girls or boys.
MOTIVATION LINKS: We will assess children's self-efficacy (e.g., how easy is math/reading for you?), interest (e.g., how much do you like math/reading?), and anxiety (e.g., how nervous would you feel if you had to take a big test in your math/reading class?) in math and reading. We predict that greater endorsement of the stereotype that one's gender has to work harder to do well in a domain (i.e., math and reading) will be associated with lower self-efficacy and interest, and greater anxiety among members of that gender. These predictions may be more likely to be confirmed among older children than among younger children. We will also explore whether they are confirmed more strongly in one domain vs. the other and in one gender vs. the other.
3) Describe the key dependent variable(s) specifying how they will be measured. Stereotype DVs will be children's average coded responses across a 4-item scale (five response options; min=0, max=4) used to assess effort stereotypes in math and reading, for boys and girls (four total composites). Primary motivation DVs will be children's average coded responses across the 2-item scales (five response options; min=0, max=4) for self-efficacy, interest, and anxiety in math and reading (six total composites).
4) How many and which conditions will participants be assigned to? 2 gender conditions (within-person): Effort beliefs about (1) girls and (2) boys
2 domain conditions (within-person): Questions about (1) math & (2) reading
5) Specify exactly which analyses you will conduct to examine the main question/hypothesis. STEREOTYPES:
effort_stereotype ~ gender_condition(girls vs. boys)*domain(math vs. reading)*child_gender(girl vs. boy)*age + (1 | child)
Our primary prediction will be informed by the gender_condition main effect and the gender_condition*domain interaction. The presence of a relative difference in stereotypes about girls vs. boys will be inferred by comparing the marginal means from the model for gender_condition overall, as well as within a particular domain. We will calculate and compare the within-domain marginal means for girl and boy stereotype targets regardless of whether the gender_condition*domain interaction term is significant. We may also calculate the gender_condition*domain marginal means within each child gender separately and at particular ages. Age-related changes will be inferred from the coefficient on age and its interactions with gender_condition and/or domain. We will interpret all other significant effects in the model (e.g., main effects of domain).
MOTIVATION LINKS:
self-efficacy ~ effort_stereotype_recoded*domain(math vs. reading)*child_gender(girl vs. boy)*age + (1 | child)
interest ~ effort_stereotype_recoded*domain(math vs. reading)*child_gender(girl vs. boy)*age + (1 | child)
anxiety ~ effort_stereotype_recoded*domain(math vs. reading)*child_gender(girl vs. boy)*age + (1 | child)
The relation between stereotypes and motivation is indexed by the "effort_stereotype_recoded" term in the model, coded to represent stereotypes about the child's own gender. We expect that the relation between stereotypes about effort (i.e., child's own gender has to work harder to do well) and motivation (self-efficacy, interest, anxiety) will be negative. It is possible that these relations become more differentiated with age (i.e., a stereotype*age interaction). We will explore whether these relations are more differentiated for one gender than for the other (i.e., a stereotype*gender interaction) or in one domain than in the other (i.e., a stereotype*domain interaction).
6) Describe exactly how outliers will be defined and handled, and your precise rule(s) for excluding observations. We will exclude data from children whose parents interfered with testing or who encountered technical issues during testing (e.g., a poor internet connection that prevented the child from hearing the experimenter's questions). We will also exclude data from children whose tester made mistakes when they administered the study. Although we won't prevent nonbinary children from participating, we will not include their data in primary analyses because we anticipate we will not have enough children with this gender identity to analyze their data as a separate group. The planned sample size (see below) pertains exclusively to children who identify as girls or boys. We will exclude data from any children who are not in school yet or who are homeschooled, for whom some of the questions will not be applicable.
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. 246 children between the ages of 6 and 12 (approximately half girls and half boys) will be recruited to participate.
8) Anything else you would like to pre-register? (e.g., secondary analyses, variables collected for exploratory purposes, unusual analyses planned?) We will conduct reliability analyses for each measure (e.g., Cronbach's alpha, factor analysis). If reliability is low, we may analyze items separately and/or drop items that do not cohere with the rest. For example, if the self-efficacy, interest, and anxiety items have a high alpha overall (>.70), we may create overall composites (i.e., combining the three groups of items) for each domain.
We may explore whether children's race/ethnicity, SES, and environment (as assessed via variables linked to zip code) moderate the predicted effects.
For analyses detailed in (5) above, we may explore patterns separately for children who report having reading vs. English class in school. (It is possible that not enough children will report having English class in school, since our sample is young. If so, we will just focus on those children who report having reading class in this supplemental analysis.)
We may include random slopes for domain (math vs. reading) and gender_condition (beliefs about boys vs. girls) in the mixed-effects models. We may also simplify the mixed-effects models if any of the fixed-effects terms do not show significant main or interactive effects. If any of the mixed-effects models do not converge, we will follow the recommended steps for resolving convergence issues (https://rpubs.com/bbolker/lme4trouble1); if these fail, we will gradually simplify the models until they converge.
We have tested 30 children (roughly 12% of the planned sample) to ensure that there are no major issues with the procedure. We are now confident that the procedure runs smoothly and no questions are causing confusion, so we are preregistering the study and proceeding to test the full sample.
