Open-ended questions – especially if the sample is large – demand a great deal of coordination (without it we do not know if "apple" and "apples" apply to the same response when performing automated analyses).
Therefore, we do not ask "Which is your favourite fruit?" using an open-ended question, but instead present all main candidates (apples, bananas, etc.) and include the option "other" for rare responses. For similar reasons, open-ended questions should not be used to ask about the year of birth, as that would mean a wide range of responses and an unnecessary need for coding; dropdown menus should be implemented instead.
Close-ended question types should be used wherever possible and whenever we have extensive knowledge about the observed issue. That, of course, does not mean that we should not use open-ended questions at all, as they are extremely valuable because they enable collecting important responses that we did not predict and did not understand enough to structure them sufficiently, or we even forgot to include them.
As a rule, questions such as "do you have any additional comments …" should be included at the end of the questionnaire. This may also be done in a more structured way, e.g. "can you specify two of the most critical aspects …", "how do you in general rate …", and also "how do you rate the questionnaire …". Textbox size is also important – a large box encourages more extensive comments.
That does not only enable valuable insight into the issue but also contributes to the respondent's satisfaction. Today, one of the main reasons for participating in a survey is the need to express one's opinion.