Russian Presidential Voting Behavior 2018

All respondents in a sample of adults in the Russian Federation designed to be nationally representative are given a set of four non-sensitive items with which they might agree. A randomly selected third of the sample is given a fifth (potentially sensitive) item with which they might agree, a statement that they had voted for Vladimir Putin in the March 18, 2018, presidential election in Russia. Another randomly selected third of the sample is given a fifth item that is non-sensitive and that previous research has established that almost everyone will agree with, a statement that they have watched at least one television question during the previous week. All respondents are asked to tell the enumerator a single number that represents the total number of items with which they agree and explicitly instructed not to tell the enumerator the individual items making up this count.

2018-06-21

2018-06-26

The outcome variable is a count of the number of items with which each respondent agrees, without knowledge of the specific items that make up this count for any given individual.

The outcome of primary interest is the difference in the counts produced by the control subsample (the one with only 4 non-sensitive items) and by the “treatment” subsample (the one with the added item on voting for Putin). This difference will enable the researcher to estimate the share of the population that voted for Vladimir Putin in the March 2018 Russian presidential election. Additionally, the difference in counts between the control subsample and the subsample with the fifth non-sensitive item (on watching TV) will provide leverage in assessing the extent to which (or whether) the outcome of primary interest (the vote estimate) is driven by the addition of the item of interest (the voting item) as opposed to other features of the experimental design.

The experiment is embedded in the monthly omnibus survey of the highly reputable independent Russian firm Levada Market Research (LMR), using their regular multi-stage area probability sample* designed to be representative of the adult Russian Federation population. Interviews are conducted face-to-face. For the experiment in question, all respondents are handed a card with a list of four non-sensitive items: 1. I usually read no fewer than one newspaper or journal in a week; 2. I want to see Russia be a country with a high standard of living; 3. I can name the first name of the chair of the Constitutional Court of the Russian Federation; 4. I am satisfied with my income level. A randomly selected third of the sample receives on their card a fifth item, positioned between items 3 and 4 in the list above: "In the last presidential elections of March 18 I voted for Vladimir Putin." Another randomly selected third of the sample receives on their card the following fifth item, also appearing between items 3 and 4 in the list above: "Last week I watched at least one television program." All respondents are then asked: "Think, please, about which of the statements on this card you are prepared to agree with. Do not tell me with which of these you agree and disagree, only tell me the number of statements on the card that you are prepared to agree with. Give your answer in a single number, from 0 to [4/5]." The enumerators are instructed to record only a single number (or codes for refusal to answer or hard to say).

*Based on data from the Russian state statistical agency in 2015, LMR’s sample is distributed among the country’s eight federal districts and the capital city Moscow, with each district divided into five strata proportionally to adult population size. All cities with a population of over one million are included as self-representative units; in the remaining (non self-representing) strata, probability proportional to size (PPS) is used to elect 1-10 urban settlements (or rural districts in rural areas). The number of interviews in a given stratum is divided equally among selected settlements. In total, 137 primary sampling units (PSUs) are drawn, including 99 urban settlements and 38 rural districts in 48 subjects of the federation (official federation-forming regions). In each selected settlement, two electoral districts (or two villages in a rural district, 18 districts in Moscow, and 8 districts in St. Petersburg) are selected at random from a complete list of electoral districts (or villages where appropriate), resulting in about 280 secondary sampling units (SSUs). Selection of households is accomplished by the random route method using route lists, and one eligible adult per household is selected according to gender, age, and education level. People institutionalized in prisons or hospitals, people conscripted into the military, the homeless, and people living in very remote, difficult-to-access, or extremely small settlements are excluded from the sample.

The researcher uses the randomization feature of the software Microsoft Excel to randomly assign one of the three versions of the list question described above to the number of each questionnaire to be administered as part of LMR’s omnibus survey. The researcher then creates single-page “inserts,” with each insert containing only the correct version of the question to be administered for its assigned questionnaire and the number of that assigned questionnaire. The inserts are sent to the survey firm (LMR), which inserts each insert into the questionnaire assigned to that insert. The questionnaire itself (independently of the insert) contains not the actual question, but instructions for the enumerator to use the insert and instructions for recording the respondent’s response (count).

Individual

No

0

1,602 adult residents of the Russian Federation

534 control (with only four items), 534 with fifth item on Putin vote, 534 with fifth item on watching television