Pandemics and Political Behavior in Authoritarian Regimes: Russia and COVID-19

The first RCT involves approximately 80% of the sample of 1600 respondents (so approximately N=1280), randomly dividing this 80% into 4 groups of about 320 each. The first group serves as a control and is not asked the first question in the trial at all, while the other three groups each receive a question that asks people to reflect on a different consideration widely linked to the COVID-19 pandemic: intrinsic personal threats linked to the virus’s spread, economic threats associated with the virus crisis, and leaders’ performance in combatting the virus. These three treatment groups are then asked to say how closely they are following news related to the specific threat they are assigned to think about.

The second RCT comes after the first and involves all 1600 respondents in the survey, dividing them randomly into one control group and one treatment group. The treatment group is primed to think about what they believe to be Putin’s position on the coronavirus crisis without being prompted to specify what it is. The control group receives no prime.

The third RCT, which comes after the second RCT in the interview, involves all 1600 respondents and randomly bifurcates the two groups from the second RCT, yielding four groups of about 400 respondents each. Among respondents who were in the second RCT’s control group, one subgroup receives no prime (being a control for the third RCT) and the other subgroup is read a statement from a doctor recommending certain health-related behavior that could curb the spread of the virus. Among respondents who were in the second RCT’s treatment group (those who were asked to reflect on Putin’s position on the virus), one subgroup is read a statement by Vladimir Putin recommending the same kind of health-related behavior that the quoted doctor was recommending that could curb the spread of the virus, while the other subgroup is read both Putin’s and the doctor’s statements.

2020-04-24

2020-04-30

The primary outcomes of interest in the first RCT are four-point agree-disagree scales on three statements of support for Russia’s political leadership. The primary outcomes of interest in the second and third RCTs are 11-point scales on the likelihood the respondent will engage in certain health-related behaviors widely believed capable of curbing the spread of the virus (social isolation).

The larger outcome of interest in the first RCT is the difference in patterns of responses between the control group, on the one hand, and the three treatment groups, on the other, and secondarily the difference between the three treatment groups. For the second RCT, the larger outcome of interest is the difference in patterns of answers between the treatment and control groups. For the third RCT, the larger outcome of interest is the difference in patterns of answers between the group receiving both the Putin statement and the doctor’s statement, on one hand, and the group receiving only the doctor’s statement, with the control group providing a measure of baseline opinion without any endorsement. Also of interest will be whether any “rally effect” induced by the virus crisis goes on to shape how people respond to a health-behavior endorsement from the leader potentially benefiting from the rally. The larger outcome of interest here will be the degree of effect heterogeneity in the second and third RCTs observed across the groups from the first RCT.

The experiment (full text attached to this registry) is embedded in the regular “omnibus” survey of Russia’s highly reputable Levada-Center agency. The sample is selected through a probability sampling method with random digit dialing (RDD) to generate mobile phone numbers within the fixed range of prefixes and numbers functional in Russia proportionally to each federal district. The total coverage of the target population is 97 percent by the mobile frame sample. No socio-demographic quotas are used. It will include about 1600 residents of Russia age 18 or older. Interviews will be conducted by phone using the CATI method.

Randomization in the assignment of respondents to control and treatment groups is achieved through the randomization function in Levada’s survey software.

Individual

No

0

1600 adult residents of Russia

For the first RCT, there will be approximately 320 in the control group and 320 in each of the three treatment groups. For the second RCT, there will be approximately 800 in the control group and 800 in the treatment group. For the third RCT, there will be approximately 400 in the control group and 400 in each of the three treatment groups.

Analysis Plan