How do we calculate the average ratings of Ukraine Elections?

Sociological companies poll the residents of Ukraine at various intervals in order to find out the level of support for the leading political parties and politicians of the country. The results of such surveys are used to assess public opinion. However, imperfect sampling and methodologies lead to noticeable inaccuracies in representativeness. In addition, sociological companies most often have politically interested customers, which additionally leads to distortion of results. It is very bad when so-called "fake polls" are published, which no one has ever conducted. You can be deeply misled if you rely on data from one particular case study in isolation.

Which polls are involved in calculating the average rating?

To minimize statistical bias and improve data reliability, we combine polls into a single average. Only those sociological services that disclose full information about the study are involved in the calculations: the method of the survey, the organizers and customers, the size and error of the sample, the period and place of the survey. Non-representative surveys or those with a small sample size are eliminated from the calculations. On the graph, the results of each survey involved are dispersed in the form of separate points - party ratings grouped by color

How is the average calculated?

We use an exponentially weighted moving average model to identify trend lines. Each next indicator is affected to a certain decreasing degree by each previous one. Survey data is counted along with sample size and margin of error - larger surveys with lower margin of error have a greater impact on the average. The corners of the graphs are smoothed using the cubic interpolation mode

What do we end up with?

The bias of each individual poll will be blurred by other polls, and if the results of some public opinion poll sharply deviate, then their impact on the average will be minimal. Thus, by combining more data, the average rating is much less susceptible to inaccuracies and better reflects the real picture of electoral sympathies at a certain point in time. On the chart, our trend indicator is shown as a continuous colored line