There are advantages and disadvantages to both. Open ended questions may elicit more detail but are harder to analyse and may take longer to administer. Clearly, response categories for closed questions need to be fully comprehensive and require careful consideration (Polgar et al. 1995).

In the DHS, questions that need adapting include words in parentheses.  How the question is phrased will depend on the respondent’s specific circumstances.  In the question below the interviewer must choose the appropriate term depending on whether the respondent is married or not.

In the question below the capitalised word in parentheses is an instructions to the interviewer, and should not be read out.  The interviewer must substitute the instruction for the respondent’s child’s name.

Biased questions elicit a response that does not accurately reflect the respondent’s position.  A question can be biased in 3 ways:

• If it implies the respondent should be engaged in a particular behaviour. 
• If the response categories are unequal or loaded in a particular direction.
• If words with strong positive or negative connotations are used.

For example, "What side effects did you experience?" assumes side effects were experienced. First ask "Did you experience any side-effects?"

For example studies show that recall of behaviour, such as condom use, is easier if it is associated with a specific event, for example last sexual event, rather than a specific reference period, such as last 4 weeks. Questions associated with specific events may also be less prone to social desirability bias. Moreover, evidence suggests that use of condoms at last sex correlates strongly with measures of consistent use (Cleland et al. 2004).

For example, the meaning of terms may vary by age, sex and locality. In the National Survey of Sexual Attitudes and Lifestyles (NATSAL) conducted in the UK, the term "sexual partner" was carefully explained to all participants because in some cases it was considered too "casual" to refer to a married partner, while among single respondents it was considered appropriate to describe a steady relationship but not a casual encounter (Fenton et al. 2001).

"How much did your last visit to the clinic cost?" requires the respondent to recall the 3 bits of information - last visit, the travel cost and the clinic costs. A series of more simple questions would be preferable.

• For example: "When did you last visit the clinic?
• How did you travel to the clinic?
• How much did you pay for the transport?
• Did you have to pay any money to the clinic?"

Breaking the questions down helps the respondent recall relevant details and that should be more accurate than a single, inclusive question.

Attitudinal questions are challenging (what is your attitude to condoms?). It is often better to pose a series of statements such as:

"If I loved someone I would rather not use a condom than risk losing this person", "It is not ‘cool’ to use condoms", or "It is very embarrassing to obtain condoms" can be combined into a score for a multi-item scale. The individual statements of such a scale could be answered by "Yes" or "No", or by "Likert scale" response options ranging from strongly agree to strongly disagree. By combining the response to all statements, a score is computed and an attitude towards condoms estimated.

Likert scale is a widely used method to rank responses and typically takes the following format.

1. Strongly disagree
2. Disagree
3. Neither agree nor disagree
4. Agree
5. Strongly agree

Source: adapted from Collumbien et al. 2012.