Every election year there are those who publish fake news or cast doubts about the importance of electoral research. And there are people who questionwhether the polls are relevant, since, not infrequently, the results of the polls may be different from what the surveys indicated. This, in fact, can happen. But it is important to remember: the data that appears in voting intention surveys are like a snapshot of the moment, they show who is ahead and point to downward or upward trends.
ADVERTISING
This video from Politize! explains the meaning of a pre-election survey:
Therefore, voting intention surveys do not have the power to predict the outcome of the election, they only show paths and trends. Everything can change in the face of new facts, in weeks, days and even hours. Who intended to vote for X decides, at the last minute, to vote for Y? And the polls themselves can influence this change in thinking, therefore, they are an important part of the electoral scenario.
What is it for?
Polls assess voting intentions for the next election and serve as a thermometer for the electorate, since, in some cases, voters change their vote according to the candidate who is ahead in the polls.
ADVERTISING
This phenomenon can be called “useful voting”: the voter chooses the candidate who can defeat the one he does not want to see in the chair of president (or governor, mayor, senator and even deputy).
Pre-election surveys also serve as a thermometer for political parties, which measure the popularity of their candidates.
The surveys also help guide media outlets when choosing candidates who will participate in debates. Those with less than 1% voting intentions are left out. It's a way of optimizing the debate time and focusing the speeches on those who are most likely to win. For this reason, debates (TV, radio and websites) usually only present the first placed in the polls.
ADVERTISING
Margin of error
The margin of error, also called confidence interval, is a mathematical calculation made based on the sample in the survey (how many people were interviewed), to indicate whether the results are truly a reflection of what the total population thinks.
As research is not carried out based on absolute values but rather on estimates (statistics), there will always be a margin of error. By defining the number of respondents, it is possible to identify the survey's margin of error, that is, how close or far that result is from reality.
Standard indexes in the market are 2 percentage points of error with a 95% confidence level. That is, if the survey is repeated 100 times, in 95 of them the results will be within a variation of 2 percentage points more or less.
ADVERTISING
The margin of error in polls is often the origin of memes on the internet
Who pays for a survey?
Generally, research is commissioned by large financial institutions and has a high cost, between R$100 thousand and R$200 thousand. But there are also press outlets that finance their own research, such as Datafolha.
After ordering the survey from a research institute and agreeing on the price, the organization begins to select people by class, age, gender. This selection is called a variable and the selection takes place randomly, according to data from the Brazilian Institute of Geography and Statistics (IBGE). This is a sample of the Brazilian population.
ADVERTISING
The group can vary between large, medium or small cities. There is no exact number of participants for this type of research, but it is usually between 1 and 4 people.
Interviewers approach citizens in their homes, on the streets or by telephone. Around 20% of respondents receive feedback from researchers to check whether the past data is correct.
Poll on social media
Polls carried out on social media do not serve as statistics. In 2019, 82,7% of the population had access to the internet, but not everyone has access to social networks, according to IBGE data. In polls carried out on the networks, the voter searches for the research and the algorithm directs them to the “polls” with which users most identify.
In other words, a Lula voter, for example, is more likely to see more polls from left-wing pages than a Bolsonaro voter. On social networks, users can create and vote for more than one account and, for this reason, polls can be manipulated. Keep an eye!
Follow the research:
- The newspaper O Globo created the blog “Pulso”, which analyzes electoral polls based on the behavior of Brazilians. (The globe*)
- Power 360º has a mechanism that allows the assembly of graphs of electoral polls since 2000, whether for municipal elections, Senate, state governments and the Presidency. (Can)
- Estadão has a “calculator” which shows in graph the most likely scenario of the electoral dispute, based on research released in the last six months. (Estadão*)
Curto Curatorship:
- How is an electoral survey carried out? (Super interesting*)
- How do surveys work? (Politicize!)
- Consult registered searches (TSE)
- Type of electoral polls (To the Facts)
- How do Twitter bots work? (Public Agency)
Featured photo: Antônio Augusto/Ascom/TSE
(): may require registration and/or signature
Curto Explain: everything you need to know and are embarrassed to ask!😉
Click to see more explanatory content ⤴️
