Respondents generally answers questions based on their perception of social desirability of those answers. The answer that seems to be most desirable is endorsed the most. Therefore, questions should be framed in a way to enable the respondents to distance themselves from their responses.
The respondents can imagine what their other peers would answer rather than what they would. It is expected that this can reduce the possibility of response in a way that can anticipate would e more acceptable.
In cases if these errors go undetected, they can represent sources of error in basically measurement of concepts. There are few writers who have out rightly condemned social research on the basis of evidence of response sets, but it becomes important to not get swayed away by such findings.
One cannot be sure of how prevalent these effects could be. To some extent awareness of them has also led to measures to restrict their effect on data or by letting interviewers to limit the probable effect of the social desirability effect by not becoming too friendly with respondents and by not becoming prejudiced about their responses.