The main mistake of many novice analysts and political scientists is the direct interpretation of public tests. Many believe that a large difference in test figures over different periods indicates a fundamental change in public opinion. The recent studies by the Polling Observatory show that this is not at all the case.
Such correlations depend on the time of the survey and its audience. The Labor Party electorate tends to change its mind more frequently than the Conservatives, but the Conservatives usually represent a very narrow section of society about age, education, and gender.
Thus, changes in polling indicators do not indicate a change in public opinion, but that between the respondents, the ratio of different strata of society has changed.
Thus, turnout weighting is very important, which allows you to find out the percentage of representatives of different classes and estates between the respondents.
To normalize the received data, a filtering method is also used, which allows you to reduce the amount of “noise” and get more average data that helps you see a more realistic objective picture.
Also, analysts advise making allowances for major events that can instantly affect people’s opinions. Events like the adoption of a budget, the announcement of a coalition of parties, or the withdrawal from a treaty can affect the results of a survey.
In this regard, you need to consider when the survey was made and carefully analyze the data.