#100068 | AsPredicted

'Survey Clickbait: June 2022 study'
(AsPredicted #100068)


Created:       06/15/2022 05:12 AM (PT)

This is an anonymized version of the pre-registration.  It was created by the author(s) to use during peer-review.
A non-anonymized version (containing author names) should be made available by the authors when the work it supports is made public.

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?
This study examines the effects of "clickbait polling." Exposure to headlines about this type of polling…
H1. … cause respondents to rate Americans as less intelligent.
H2. … cause respondents to view Americans as less qualified to vote.
H3. … reduce confidence in the U.S. system of democracy.
H4. … increase support for restricting uninformed peoples' ability to vote.
H5. … reduce trust in the public opinion polls.
H6. … reduce trust in the news media.

3) Describe the key dependent variable(s) specifying how they will be measured.
• H1 will be measured using a battery of trait ratings: informed, patriotic, selfish, tolerant. Each is measured on a 4-point scale. We only expect effects on intelligence.
• H2 will be measured by two items: confidence in Americans to cast informed votes (4-point scale) and agreement with the statement "most Americans are well-qualified to vote" (5-point scale).
• H3 will be measured using one item on confidence in the U.S. system of democracy (4-point scale).
• H4 will be measured using (a) a three-item battery on support for restrictions on uninformed voters and (b) an item on whether uninformed voters or people who do not vote are a greater threat to American democracy. We will interpret the latter as tapping a tradeoffs people plausibly weigh in their mind when they decide whether to support the former.
• H5 and H6 will be measured using three-item batteries regarding the accuracy, trustworthiness, and informativeness of public opinion polls and the news media.
• For H2, H4, H5, and H6, we will report effects on the separate measures, as well as on indices that combine the specified items. For H4 the item designed to measure what is the greatest threat (denoted (b) just above) will not be included in the indices.

4) How many and which conditions will participants be assigned to?
Using simple random assignment, we will allocate respondents to one of three conditions: a treatment group, a placebo control group, or a pure control group that sees no stimulus at all.

5) Specify exactly which analyses you will conduct to examine the main question/hypothesis.
We will estimate treatment effects using OLS with robust standard errors using lm_robust() in R. Our main model will be:
Yi = a + b*treatment_i + b*placebo_i + Σ b*x_i + epsilon_i
where treatment_i is an indicator for treatment status (0 = control, 1 = treatment) and x_i is a pretreatment covariate. We will use LASSO to select pretreatment covariates that predict the outcome variable in the control group, with the tuning parameter (i.e., lambda or the penalty) selected by cross-validation. We will also report unadjusted estimates (i.e., excluding the b*x_i terms).

As all hypotheses specify directional expectations, we will conduct one-sided tests. We will use a significance threshold of p < 0.05 as a reference point, not a sharp threshold.

6) Describe exactly how outliers will be defined and handled, and your precise rule(s) for excluding observations.
So that all treatment effect estimates represent the same group of respondents, we will exclude all respondents who do not reach the final page of outcome questions.

We will drop all respondents who complete the questions that are common to all respondents in 1/3 or less of the median time.

We will also drop respondents who fail at least two of the following quality checks:
- Indicates that they use Doromojo, a fictitious social media platform.
- Has a mismatch of 5 years or more between their age and year of birth.
- Has a mismatch between their state of residence and their ZIP code.
- On the open-ended most important problem question, provides a non-response or non-sequitur response.

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.

2400

8) Anything else you would like to pre-register?
(e.g., secondary analyses, variables collected for exploratory purposes, unusual analyses planned?)

No