The problem concerning “Don’t know” (DK) responses:
There are times when the researcher comes across few difficult situations to handle like a DK response. This response of DK does not hold much importance till the time it remains small. But when it becomes big, it becomes a matter of concern. The respondent might not actually know the answer or fail in getting the right information.
The best way to deal with DK response is to design correct and more appropriate questions. Even the rapport with respondents can help in minimizing DK responses. There could be times when the DK response has already occurred. What should e done in such a case? An estimate should be made about the allocation of DK answers from other data in the questionnaire. The other could be to keep them as altogether a separate category if DK responses are legal, otherwise the reader can make his own decision.
Another way to minimize DK response is to assume that DK responses can occur more or less randomly and as such could be distributed them among the other answers in the ratio in which the latter may have occurred. The same response could be achieved if all DK responses are excluded from tabulation purpose and that too without inflating the actual and right numbers of other answers.