Is the question format to be open or closed? Closed format is where the response is one or more selections from a list, or a number within a pre-defined range. Open format is where the response cannot be pre-determined, e.g., free-text, such as an opinion or "other comments". For a computerised interview, the questions need to be closed format. For a paper form, it needs to be borne in mind that open format questions are more difficult to handle in terms of coding. It is therefore advisable to use closed format questions wherever possible.

The questionnaire needs to contain an instruction as to how it should be completed, together with an example. On a paper form, the clearest method is to ask subjects to ring the appropriate answer, for example, in questions concerning the frequency of consumption of a range of foods. There are potential problems with other methods of indicating an answer from a list. For example, if there are only two alternative answers to a question, a tick or cross may indicate that this is the intended answer, or it may indicate a deleted response such that the true answer is the unmarked one. If subjects are asked to write a number in a box to indicate say, the frequency of consumption, they need to remember which number relates to each frequency. This may be straightforward if the response "three days/week" is to be indicated by "3" and "everyday" is to be indicated by "7". However, it may be confusing if "three servings/week " is to be indicated by "3" but "one serving/day" is to be indicated by "5".

The range of responses for each question type should be consistent throughout the questionnaire. For example, if frequency responses used for one question are 1,2,3,4,5,6,7 to indicate the number of days per week, F to indicate once every 2-3 weeks and R to indicate rarely or never, then responses to other frequency questions should be the same. The range of responses for each question needs to be adequate to differentiate between individuals. For example, in questions about attitudes, four alternatives are better than five - if five alternatives are given subjects may be most likely to choose the middle (least informative) response. The type of response may also need to be different for different questions. For example, frequency responses may be appropriate for consumption of foods such as meats, fish, pasta and rice, but not for sugar in tea and coffee, as the range of responses would need to be much greater. In the latter case, it would be better to ask how many cups of each drink are usually consumed per day and how many teaspoons of sugar are added to each cup.