Sociology of Polls: Interview with a Troll

Opinion polls show that many Americans claim to have voted for Donald Duck in the last presidential election, others pretended to be Martians, and still others have reportedly been able to convince themselves not only that Lady Diana was in fact assassinated, but also on the same level of evidence that the crown princess is still alive;

of course nobody believes all this.

But when the same survey technique finds for large numbers of respondents that they believe Barack Obama is a Muslim, Hillary Clinton the patron of a pedophile association, and Donald Trump the legitimate winner of the last presidential election, then the proliferation of such misconceptions becomes a disturbing book written after the other.

The horror of the gullibility of many of their contemporaries would only be appropriate if one were to assume that they are sincere in polls.

, shows that this assumption would be quite gullible on your part

.

A prominent source of error is that Donald Trump's supporters assume that every social scientist who surveys them or has them surveyed is a supporter of the opposite party who wants to trick them.

They would therefore rather bear the stigma of sharing the imaginative views of their political hero than read the newspaper a little later as punishment for their own sincerity, not even they would believe him.

In other surveys, participants were offered a small sum of money for honest answers, and here the Republicans judged completely realistically and little differently from the Democrats who were also surveyed.

Problematic online surveys

Of course, the political defiance does not explain why there are respondents from both political camps who are equally convinced of right-wing and left-wing nonsense.

The second major source of error is therefore the troll.

He doesn't want to fool his political opponents, he wants to fool scientists.

After he had been doing his mischief in secret for a long time, research has long been on his trail.

In a youth study, a surprisingly high proportion of those questioned said they were adopted children, gang members, struck blind or wear prostheses.

Since such a concentration of extremely rare characteristics in a sample is highly improbable, it was considered one of the indicators that the respondents had allowed themselves a joke.

Other indicators such as the frank admission of one's own insincerity in the follow-up interview or the information obtained from the alleged adoptive parents could confirm the suspicion.

By the way, only one of the hundred prosthesis wearers of a first data set remained after these corrections.

A few years ago, the technique of identifying trolls was used in a self-critical opinion poll to examine how their lack of seriousness contributed to the high levels of support for outrageous factual claims.

The result is impressive: while the convicted trolls accounted for only 21 percent of those questioned, they were responsible for a much larger proportion of the misconceptions reported.

Without their involvement, for example, the legend of the pedophile association would not have been approved by twelve but only a good six percent of those questioned.

One of the reasons for the trolling is that more and more surveys are taking place online, in which the interviewer is not present.

But who knows whether his authority might not be overestimated.

Elementary rules of conversational behavior presuppose that there are those present who will refer to them if they are violated.

But the interviewer is absolutely not allowed to defend himself against the imposition of nonsense, to ask for clarification in the face of obvious self-contradictions or to show his own doubts about the sincerity of the interviewee.

This uncritical attitude is intended to prevent the interlocutor from seeking praise and recognition of the questioner or from trying to avoid his disapproval, both at the expense of the reliability of the data thus collected.

But it has the side effect