NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
Anti-Asian Racism during COVID-19 (Pilot)
Initial registration date
June 24, 2020
June 25, 2020 11:44 AM EDT
This section is unavailable to the public. Use the button below
to request access to this information.
Johns Hopkins University
Other Primary Investigator(s)
Johns Hopkins University
Additional Trial Information
This is a registry for our pilot survey. We will study anti-Asian racism in the US in the context of COVID-19. Specifically, we will study how COVID-19 and its media coverage affect anti-Asian racism. We will implement several information RCTs to study the role of belief on the severity of COVID-19 and biased media report. To understand the role of publicity on racially discriminatory behaviors, we will also randomize different messages about anonymity of their survey responses before starting a survey module on discriminatory behaviors. We will measure truthful attitude toward racism using a donation experiment, in which participants will receive extra monetary reward if they authorize making a donation to a randomly chosen politician with different racial attitudes.
In pilot survey, we will randomize different survey formats to understand how they will affect the estimation of continuous latent variables, such as attitude and perception. Half of survey sample will respond to a questionnaire using Likert scale and another half will respond to a questionnaire using Slider. Using survey responses from two different formats, we will estimate the nonparametric densities of latent variables and we will discuss which survey method will give responses which are more aligned with standard identification assumptions in measurement error literature.
We will have four RCTs in one survey. The first RCT is an information RCT giving precise information about local COVID-19 statistics for the treatment group. Control group receives no information. The second RCT is another information RCT asking respondents to read racially biased / neutral / no news articles about COVID-19. The third RCT is another information RCT giving different messages about anonymity of their survey responses. The treatment group receives a message that they may be followed up by our research team on their later responses. The control group receives a message that their survey responses are completely anonymous. The fourth RCT is a donation experiment, assigning different politicians with different racial attitudes. We will ask respondents whether to authorize a donation to a randomly chosen politician for them and if they authorize the donation, they can receive small extra monetary reward. In pilot, we will vary the monetary reward amount to design the main survey. We aim to find the amount which will be enough to obtain truthful attitude, but is not too expensive.
In addition to the four RCTs, we will randomize the survey formats to understand their impact on the estimation. Half of survey sample will respond to a questionnaire using Likert scale and another half will respond to a questionnaire using Slider. Using pilot survey responses from two different formats, we will estimate the nonparametric densities of latent variables and we will discuss which survey method is better to apply the standard econometric techniques to study measurement errors.
Intervention Start Date
Intervention End Date
Primary Outcomes (end points)
We have three primary sets of outcome variables.
The first set of outcome variables is about anti-Asian sentiments. The second set of outcome variables is about perception about racism in the US. The third set of outcome variables is about hypothetical choices about discriminatory behaviors, and an outcome of a donation experiment.
Primary Outcomes (explanation)
We collect multiple measurements for each set of outcome variables, not to cherry-pick the most responsive survey items later, but to apply nonparametric estimation techniques to estimate measurement errors. Having multiple measurements is one of the core identifying assumptions.
The first set of outcome variables about anti-Asian sentiment is the shorter version of realistic threat measure, intergroup anxiety and prejudice measures used in Stephan et al. (1999). The second set of outcome variables about perception about racism in the US is measured by asking how much respondents agree to five statements. The statements are (1) whether typical American citizens might feel annoyed at racial stereotype jokes, (2) whether typical American citizens will follow a manager with racial bias as long as the person delivers results, (3) whether a person who publicly claims to be a racist can maintain good social relationships with most people, (4) whether typical American citizens will tolerate racial bias to some degree, as long as it does not violate the law evidently, and (5) if a person refers to the novel corona virus as “China virus”, the person will face severe criticism. The third set of outcome variables is about discriminatory actions. We will ask whether respondents have racial preference in hiring decisions, doctor choices, school choices, and neighborhood choices even when all objective conditions are the same but race / racial composition. To have one incentivized response about discriminatory behavior, we will do a donation experiment, asking respondents whether to authorize a donation to a randomly chosen politician. We chose Tom Cotton and Dan Whitfield as two politicians with different racial attitudes.
Reference Stephan, Walter G., Oscar Ybarra, and Guy Bachman. "Prejudice toward immigrants 1." Journal of Applied Social Psychology 29.11 (1999): 2221-2237.
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
To remove the order effect, we randomize the order of information RCT about local COVID-19 statistics and media article RCT. Half of sample will do local COVID-19 statistics RCT first and news media article next and another half of sample will do them in reverse order. We measure anti-Asian sentiment and perception about racism after implementing COVID-19 statistics RCT and news article RCT. We randomize the format of anti-Asian sentiment and perception questions in either Likert scale or slider. To not contaminate other parts of the survey with the effect of anonymity message RCT, we implement this RCT in the last part of our survey, just before asking about hypothetical choices about discriminatory actions and a donation experiment. Every RCT in our survey will have half of sample in treatment group and half of sample in control group but a donation experiment. To increase power, we will assign 90% of sample to a politician with racial bias and 10% of sample to a politician without racial bias. Among 90% of sample, we vary the donation and extra monetary reward between $0.25, $0.5, $1.
Experimental Design Details
Randomization is done using the randomizer in Qualtrics survey program.
Unit of randomization is an individual.
Was the treatment clustered?
Sample size: planned number of clusters
Sample size: planned number of observations
Sample size (or number of clusters) by treatment arms
The sample size for treatment arms differ by RCTs. We implement 50:50 randomization for every RCT but donation experiment. To increase power in donation experiment, we assign 90% of sample for treatment arm and 10% of sample for control arm.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
This is a pilot study. Power calculation for main survey will be done using pilot survey data.
INSTITUTIONAL REVIEW BOARDS (IRBs)
The Homewood Institutional Review Board (Johns Hopkins University)
IRB Approval Date
IRB Approval Number