Reading time: 4 minutes
In 2001, in the USA, the polls for the general election of the previous year were analysed, to confront them with the real results, and an incongruency arose. More than 20% of who claimed that they would have abstained had instead voted for one of the two candidates. A similar situation is present in another kind of poll, conducted in 1971: in this case, 30-70% of candidates claimed that they had never assumed illegal substances, while testing positively to urine drug tests.
Why is there such a discrepancy between what someone claims in a poll and what someone actually does? It is probably because a certain kind of answer could appear socially undesirable to the eyes of other people. This tendency to lie if someone has to answer certain questions, often called sensitive questions, is called the social desirability bias, and it is a cognitive error that could bring negative consequences to scientific studies and polls.
The sensitive questions
In a work from 2007, a metanalysis conducted by Tourangeu and Yan, several studies that dealt with the effect of this bias were analysed, looking for all the possible variables involved. Above all, what these sensitive questions are and in what way they can influence the answer; then, the intrinsic factors to the subjects, or the methods of asking the questions; to conclude with the possible use of various techniques that could limit the effect of this cognitive error.
Going step by step, the sensitive questions are those questions that can be perceived as intrusive by the subjects, because they can deal with both personal topics, such as alcohol assumption, being a smoker, their own health or emotions, and social matters, going from the ecology and the environment to political beliefs. In addition, there are those questions that some fear, worrying about the fact that the answers could be revealed to others. In these cases, what is feared is the so-called threat of disclosure.
Asking this kind of questions can lead to answers that are considered too socially acceptable, that represent a world far better than the real one. So much so that, in some cases, this kind of results is discarded from the start, if they surpass a certain limit. In other cases, instead, these questions can lead up to a 8% rate of non-answer.
The extrinsic factors
Among the elements that can influence the choice in a poll, there are the administration method and the context in which it is conducted. It’s been proven that it is much easier to answer sincerely to a written or virtual questionnaire, rather than a face-to-face interview with a researcher. Even answering by phone to sensitive questions is simpler.
If the poll is conducted in person, the context is extremely important, and the presence of other subjects can influence the answers. Just imagine a student that participates in a study about smoking at school: with their parents present, they would probably answer in the most socially acceptable way, avoiding telling the truth because they fear being judged.
There are several strategies to make this bias have the smallest possible effect. The more the anonymity of the questionnaire is guaranteed, the more sincere the answers will be. As it is when the interview is conducted by a complete stranger. Another funny but clever trick is connecting the subjects to a fake lie detector, telling them that the system can see when the person is lying. This increases the rate of sincere answers during the poll, even if the machine is turned off or it is just an empty box.
Know how to ignore the opinion of others
Therefore, how someone answers to a questionnaire is influenced by the fear of social disapproval, the unwillingness to show their own “weaknesses” or the fear of disagreeing with the dominant collective opinion. However, according to a research from Park and Lessing in 2007, this effect diminishes with the increase in age, maybe because the subjects are more confident in their opinions or, simply, they just don’t care about what others think.
Choosing the most socially acceptable answer is for sure the least risky choice. It is more self-preserving, but it does not turn out to be the right decision in the context of a scientific research. Knowing the importance that the answers in a study or research can have, there is the necessity to answer in most sincere way possible, in order not to distort possible scientific results.
- Belli, Robert & Traugott, Michael & Beckmann, M.N. (2001). What leads to voting overreports? Contrasts of overreporters to validated voters and admitted nonvoters in the American National Election Studies. Journal of Official Statistics. 17. 479-498.
- Wish ED, Hoffman JA, Nemes S. (1997). The validity of self-reports of drug use at treatment admission and at followup: comparisons with urinalysis and hair assays. NIDA Res Monogr.;167:200-26. PMID: 9243563.
- Tourangeau, R., & Yan, T. (2007). Sensitive questions in surveys. Psychological Bulletin, 133(5), 859–883. https://doi.org/10.1037/0033-2909.133.5.859
- C. Whan Park, V. Parker Lessig (1977). Students and Housewives: Differences in Susceptibility to Reference Group Influence, Journal of Consumer Research, Volume 4, Issue 2, September 1977, Pages 102–110, https://doi.org/10.1086/208685