Did researchers make a mistake on the elections again?

For years and year, in different countries, various continents, I have been conducting researches and I have been showing them to the politicians. Most often, as a firs slide they ask for political party rating, and as a second they want politician ratings – and that is what mostly what their interest is. Honestly, some of them look in which way does the country go and eventually the problems. They go through all other matchings and follow-up slides with a phone in their hands or with the “hurry up – why are you bothering me with this… “ look in their eyes.

When we talk about rating of political parties there are only two reactions of the politicians: our rating is great and this research is not good at all, we have more for sure. In Serbia, BiH, East Africa (not in Bulgaria, which is interesting) bad ratings, provided by the researchers is usually followed up by the sentence: they have never guessed…

This is why I was interested in how  accurate were the researches for the 2016 elections in Serbia, especially because agencies were saying that there were 20% of those that would vote, but are not declaring for who would they vote. 

I took 5 researches by different agencies done from 20.04 to 23.04. Results are shown in the graph:

If we assume that the possible error is +/- 2.5% there are no parties that are out of this margin.

Where do this small differences come from – my speculation:

- SPS voter are solid and this party basically there is no error.

- DS and DJB – it appears that the part of their voters was usually in that group: I’m voting, I’m not saying anything. That was the socially accepted answer provided by the surveyed because of the negative campaign against these two parties.

Other parties didn’t manage to take their potential voters out on the election day…

And as I say on the lectures: you have to believe in the public opinion researches because you believe in the blood testing. In both cases we take the sample. Nobody is taking all of your blood out to see if you are ill. It is the same here – researches are usually experienced enough and they care about their profession enough. Still, same as you double check with doctors you double check with the researching agencies. Do not rely just on one.