IISEPS NATIONWIDE PUBLIC OPINION POLLS

More than 20 years IISEPS has been conducting opinion polls the results of which obtained wide recognition in Belarus and abroad. During these years over hundred national opinion polls on the basis of multistage random sampling have been conducted representing the population of the country from the age of 18 and older. The standard volume of selection amounts approximately to 1500 respondents. This means that Institute had surveyed a total more than 150,000 Belarusians.

IISEPS interviewers’ network
The interviewers’ network of IISEPS for carrying out national opinion polls includes over 100 skilled interviewers and regional supervisors, who have been specially trained, have long experience of doing different types of research and are given instructions before each poll. Over 80% of interviewers took part in conducting of more than eight national opinion polls. 90% of interviewers have higher, mostly humanitarian, education; the rest of them have secondary and vocational education in different professions. Almost a half of interviewers are younger than 30, other age groups are represented proportionally. Interviewers’ main places of work are also different: they work as sociologists, psychologists, lawyers, accountants, insurance and commercial agents. Almost one third of them are students of institutes of higher education, about 15% work at different educational institutions and libraries. There are as many housewives and state employees, and about 10% of the interviewers are pensioners.
Under the guidance of the network manager supervisors, who work in Minsk and all regional centers, organize and control the work of regional groups of interviewers up to 15 people in number. They systematically give instructions; work on selecting and training of new (or reserved) interviewers. In general, these are professional sociologists or specialists who have been doing this work for no less than five years; those who have sufficient expertise and theoretical experience. Interviewers and supervisors are not members of any political parties, and in this sense they are not exposed to political influence and do not exert such influence upon answers of respondents.

Sampling
The number of places where a poll is conducted and the quantity of respondents in each type of settlement and in each region is determined in accordance with population and volume of a sampling. The next pattern is considered basic:
– Minsk (the capital);
– all the regional centers;
In each region they choose:
– 1 to 2 cities (with population over 50 thousand people);
– 2 to 3 towns and small towns (20-50 and 5-20 thousand of people);
– 3 to 5 villages.
Thus, on the whole a sampling includes on average about 50 settlements.
In order to choose respondents’ address the random route method is used (with the probability proportional to the size of a settlement). At the selected addresses respondents are chosen with the “nearest birthday” method under the control of a unified precept of quotas on gender, age, level of education and field of work to ensure representativeness of each stratum of the society. Later the facts are also considered according to gender, age, level of education, region of residence and the type of settlement. Statistical error in the given polls does not exceed 3%.

Procedure of polls
A poll is conducted domiciliary with the face-to-face method of interview: an interviewer reads out questions and variants of answers and marks the variants chosen by a respondent or (in case of an open question, i.e. a question to which respondents give their own answers) writes down the answers. If a certain question has great volume (many variants of answer, difficult for aural perception), an interviewer gives a respondent a corresponding card. Interviewers are trained for overcoming “barriers of primary distrust”, for maintaining a polite and friendly atmosphere of the conversation and for holding respondents’ attention. Though they often have to face negative reaction or fear, their expertise as a rule helps them to solve these problems. On average the number of respondents refusing to answer the questions does not exceed 20%. In such cases, according to the instruction, an interviewer chooses other respondents. A third of them refuses to answer because of the fear of possible repressions for disagreeing with the official point of view, another third, on the contrary, because of the reluctance to communicate with any non-state structures and actions (thus, these two groups of politically motivated denial in a way “neutralize” each other), about the same number of refuses are motivated by domestic factors (fatigue, family conflicts, work around the house, etc.). In these cases, according to the instructions, the interviewer selects other respondents. The timing of the interview is on average from 40 minutes to 1 hour.

Control of the data quality
Control of the data quality is carried out in each research in the following way:
– by control of the interview blank fill quality, execution of instructions, control of route sheets and closing statements by regional supervisors and the network manager;
– by spot check (up to 10%) of the conducted interviews over the phone or by supervisors’ (or by a third party) second visits;
– by logical control (or by control with the help of special programs) of the aggregate results;
– data re-weighing.
Compliance with these standards as a result allows to obtain reliable, objective information. As an example, we can show the characteristics of representativeness of the IISEPS survey conducted after the presidential elections in December 2010, the results of which caused a wave of emotion in the country and abroad – according to the poll voted for Lukashenko 58% of the number of voting, that was much more than the opposition announced (31%), and much less than the announced government (almost 80%):

Table 1. The results of re-weighing data of nationwide public opinion poll conducted in December 2010

As can be seen, only one parameter – primary education – the difference between data before and after data re-weighing moved beyond sampling error (because of the known difficulties to poll of this social group – usually rural pensioners). Specifically for neutralize such deviations is carried out the procedure data re-weighing.