As a rule respondents don't like open-ended questions. Compared to questions with pre-defined answer lists respondents are forced to think more and, for online surveys, they also have to type their answers. These things take time. Open-ended questions take 2 or 3 times longer to respond to than a question with pre-coded answers.
Bearing this in mind resist the temptation to bombard respondents with open-ended questions. Some survey designers are nervous about pre-coded answer lists covering all eventualities. Better, they think, to make the question completely open-ended so that nothing is missed. As long as your pre-coded answer list is well constructed you shouldn't miss anything. Make sure your answer list covers the most common responses. Try to find similar questions that have been used in the past through internet searches. Better still ask friends and colleagues your question as an open-ended but use their answers to construct your pre-coded list.
An example of a question with a pre-coded answer list is shown below:
Use an open-ended "other" category to mop up any answers that might not be covered by your answer list. And don't forget to include "None of these" or "Not sure" answers if they are valid responses.
Pre-coded questions are also much easier and quicker to deal with in analysis. Open-ended questions produce streams of text that need to be manipulated (coded) to be cross-analysed while data from pre-coded questions is analysis ready.