Exit polls may be inaccurate, but they are not useless
The results of the parliamentary elections indicated in the exit polls have become the subject of discussions in recent days about whether such measurements make sense. Scientists from the Slovak Academy of Sciences claim that measurements in the social sciences are almost never completely accurate, which, of course, also applies to surveys of political preferences. We took a closer look at the inaccuracy in this area.
"Even if we do not make fundamental methodological mistakes already during the measurement (e.g. non-standardized data collection procedure) and we have a sufficient number of respondents, we still have to expect that certain errors will occur. My research focuses on the so-called distorting response styles in which psychological variables such as the level of affability, self-evaluation, meaning of life, fulfilment of basic psychological needs and others, are reflected," says Patrik Havan of the Institute of Experimental Psychology of the Centre of Social and Psychological Sciences SAS.
The psychologist analyses the extreme and non-extreme response style more closely. "Respondents who are less sure can choose extreme answers to a lesser extent on the answer scale (these can be, for example, "I slightly disagree" or "I slightly agree"). On the other hand, other respondents, who may actually be less affable but may be assertive and self-confident, may rather choose extreme answers (for example, "I strongly disagree" or "I strongly agree")," explains the psychologist.
According to him, the results in exit polls are generally relevant, but may contain some inaccuracy. This can be created, for example, by a tendency to answer in a way that is most socially desirable.
"For example, when we ask respondents how kindly and respectfully they treat other people, their answers will generally be more in the affirmative. After all, no one wants to look like an ill-mannered brat. However, we cannot easily distinguish who from the research group agreed because they are really friendly and who only responded so due to social desirability," said the psychologist.
Choosing a particular political party can also have a certain level of undesirability. There may also be people who are close to the rhetoric of one member of the party, but other close people do not like this particular party, so they do not want to admit it. However, according to the psychologist, with such a measurement, it is very difficult to distinguish who answered honestly and who only answered so because it was socially desirable.
"An easy counterargument can be that the voters did not answer the interviewers verbally, but wrote their answers on printed sheets - thus, their anonymity was also ensured. Research shows, however, that respondents tend to distort their results even in cases where anonymity is ensured, for example, when filling out questionnaires online," claims Patrik Havan.
The psychologist summarized that the differences between the results of exit polls and the results of elections were also caused by social desirability, which is confirmed by research from other scientific disciplines. He points out that we should not have naive ideas about measurements in general because the measurement of preferences is not perfect and infallible, but at the same time, it is not completely imprecise and useless. According to him, measurements make sense, even if they contain a certain degree of inaccuracy (measurement error). And we have to take into account this possibility.
More information on this topic can be found HERE.
Source: ÚEP CSPV SAV, v. v. i.
Prepared by: Monika Tináková
Illustration photo: Unsplash/Element5 Digital