Study Plan
Study Plan
Existing Data: Registration prior to accessing the data.
Explanation of existing data: The survey is being fielded after our submitting the pre-registration. No partial data have been accessed or analyzed, and the only awareness of the data we have is through the fielding pacing information provided to us through the survey provider.
Data collection procedures: Participants will be recruited using Forthright, who will be fielding and administering the study online as well as compensating participants. The sample is college students within the U.S. All respondents who are above the age of 18, currently attending a two or four year degree program and consent will be used for analyses. We have a comprehension check, and we will not drop those who fail it. Instead, we will use whether or not one passed the comprehension check as a covariate in our analyses.
Sample size
We are surveying 1,000 individuals.
Sample size rationale
The rationale is practical and logistical. The current sample size was determined based on what was feasible to support the research and test the hypotheses we propose. The results of this survey will be used to identify power analyses in future studies.
Stopping rule: Our survey provider will stop collecting data once we hit 1,000 completes.
We manipulate one variable: The disability about which the participants are responding (blindness, depression, or an unlabeled disability). Participants will indicate their agreement/disagreement with a series of questions about either blind students, students with depression, or students (unspecified/control).
Measured variables
Independent Variables: Disability Type. Physical Disability (e.g., blindness), Psychiatric Disability (e.g., depression), Unspecified Disability (control group).
Dependent Variables (DVs): Administrative Burden Index (see below). Accommodation Support Index.
Covariates (exploratory)
Gender. Hypothesized to affect support, with women expected to show greater support for accommodations and less tolerance for administrative burdens compared to men. Categories: Men, Women, Other/Prefer Not to Say.
Respondent’s Disability Status/Accommodation Status. Disabled students or those who presently have accommodations for their disability are hypothesized to be more supportive of accommodations and less tolerant of administrative burdens compared to non-disabled students. Categories: Disabled, Non-disabled.
Disability Status of Family Member. Students are hypothesized to be more supportive of accommodations and less tolerant of administrative burdens compared to non-disabled students. Categories: Disabled, Non-disabled.
Demographic Variables. Student Status (college student). Race/Ethnicities. Gender.
Administrative Burden Index: Baekgaard et al. validate a scale with four items (our items 1-4) as an internally consistent single-factor measure of administrative burden. We added three additional items that seemed potentially relevant to our specific policy application. We will create an index and explore the possibility that our added items differ by disability type and/or might be internally consistent with our specific policy application. The scale is a five-point Likert-type response ranging from "Strongly Support" to "Strongly Oppose."
Accommodation Support Index: Zhang et al. validate a scale called the “Provision of Accommodations” scale, which measures support for specific accommodations. The original scale was 13 items, though some were only relevant for specific disabilities. We pared that list down to eight items, changing them to make them more general and applicable to multiple situations. We will create an index of these eight items. The scale is a five-point Likert-type response ranging from "Strongly Support" to "Strongly Oppose."
Statistical models
Hypotheses 1 and 2 will be analyzed using ANOVA, where the conditions will be compared on the mean level of the indices, and ordinary least squares regression, whereby the accommodation and administrative burden indices are regressed on the condition variable. We will use the emmeans R package to test the marginal means of planned contrasts, using Tukey’s p-value adjustment. Note that the covariate for if one passed the comprehension check or not will be used throughout analyses. Gender and disability status will be added as covariates in some analyses to compare to a baseline model.
Transformations: No transformations will be done. The reference value for the manipulated factor will be the control.
Inference criteria: We will be using p-values and setting our alpha at .05. Two-tailed tests will be used. p-value adjustments for multiple testing will use Tukey’s HSD.
Data exclusion: Those who do not identify as over 18 and college students will be dropped from our survey respondents. Outliers will be included in analyses.