How the Levada Center conducts its Surveys

(for surveys of which results are represented on the site)

The Levada Center conducts its political studies as Omnibus surveys of two types based on representative samples of the urban and rural population. Respondents aged 18 years and over are selected according to region, type of settlement, sex, age and educational level. To obtain full conformity of the sample structure with the official statistics, the Levada Center weights the data by sex, age, education, region and type of settlement.

Fieldwork is conducted by 26 regional offices of the Levada Center under supervision of regional and Moscow supervisors. The final data undergoes several levels of control (visual control of completed questionnaires, control of coding of open-ended questions, logical data control and sample control of route lists).

Sampling

Russian representative sampling is based on a multi-stage stratification scheme.

• First step - All urban settlements and rural areas are used as selection units to create a matrix stratification. Ten large economic regions are defined (North and North-West, Central Non-Earth-Black Zone, Central Earth-Black Zone, North Caucasus, Volgo-Viatka, Volga, Urals, Western Siberia, Eastern Siberia, and Far East). Inside each region several strata of settlements are defined, taking into consideration: administrative status, population size, and relation to autonomous republics inside Russia. In each stratum, Primary Sampling Units (PSU -- urban settlements and rural areas) are randomly selected out of the list of all urban settlements and rural areas of this stratum, with a probability which is proportional to the size of the adult population.
• Second step - Secondary Sampling Units (SSU) selection. In cities, electoral districts are randomly selected out of the list of electoral districts in the locality. The rural settlements are randomly selected out of the full list of settlements of the rural area.
• Third step - A household selection is made within the selected local points. The survey is carried out by random route sample.
• Fourth step - Within a household, the interviewer selects a respondent using sex-by-age and sex-by-education quotas.

Standard surveys

A monthly Omnibus -- "Express" -- takes place on third weekend of each month, results are available on 23-25 of each month. Sample size of Express is 1600 respondents. Express survey is conducted by the method of face-to-face interviews at home of respondent with a questionnaire 30-40 min. length.

83 PSUs are included in the sample (25 regional centres, 31 other urban settlements and 25 rural districts in 31 administrative regions of Russia). In total, about 120 SSUs are included into the sample.

A bi-monthly Omnibus -- "Monitoring" -- takes place at the second-third week of each January, March, May, July, September, and November. Results are available about 10th of the month following the survey month.

103 PSUs are included in the sample (35 regional centres, 38 other urban settlements and 30 rural districts in 38 administrative regions of Russia). In total, about 200 SSUs are included into the sample.

Monitoring surveys are conducted by the method of self-administrated questionnaires at home of respondent and, together with this, some sections of the questionnaire are filled in by face-to-face interviews with a questionnaire of about 40-45 min. length.

Sampling error and Elections

The sampling error is only a few percentage points on both types of surveys. Given the importance of the 5% threshold for winning proportional representation seats in the Duma, readers should regard parties endorsed by three to seven percent of respondents as 'having a chance' of Duma representation. The logic of random sampling means that in the event some will clear the barrier and others will not. Moreover, aggregate preferences will fluctuate between the date of interviewing and election day. In the presidential race, sampling error is specially important when candidates are within a few percent of each other in competition for second place, since the candidate actually finishing third will be eliminated while the second place candidate will have a chance to move on up to first in the subsequent ballot with only two candidates.

List of surveys used in this site