Refer to previous surveys from the same sector to see how questions were
phrased and the lists of coded responses provided. If possible, discuss with
staff who participated in the survey which questions worked well and which
didnot. It is important to build on past experience and avoid repeating the
same mistakes.
Refer to international guidance for developing survey questions. Many
sectors, including health, nutrition, education and agriculture, have extensive
guidance on developing internationally recognized indicators and survey
questions. Refer to the Food and Nutrition Technical Assistance III Project
(FANTA) for guidance.
31
Make questions specificso all respondents will understand them in the same
way. Include details and ask that the enumerators read the questions word for
word during data collection.
Note that some indicators may require multiple questions. For example, you
need to first ask, ―did you attend a health center in the last six months?‖
before asking, ―how many times did you attend a health center in the last six
months?‖
Ensure that the questions are culturally appropriate by getting input from
experienced staff with a good understanding of the local context.
Limit questions to one piece of information. If questions include multiple
pieces of information (such as ―do you limit your number of meals and the
number of items in your diet during the hungry season‖), it will be difficult to
interpret the responses. Ask these questions separately.
Use appropriate language that will be understood by respondents. Develop
or translate the questionnaire into the language in which you will conduct it.
There should be no translation in the field. Work with field staff to determine
which words and terms will be best understood by targeted communities or
households. Wording of the question should be simple and clear and not open
to multiple interpretations. If you translate the questionnaire after it is
developed, thoroughly review the quality of the translation or translate the
questionnaire back into the original language and compare this retranslation
with the original draft to identify any gaps or discrepancies.
Ensure that questions are neutral, not biased, and that they are not leading
participants toward a particular answer. Think about any assumption that the
question might contain.
Avoid emotionally charged or overly personal questions that maydraw out
a heated response or make the respondent feel uncomfortable. Either can
jeopardize the remainder of data collection with this respondent.
Ask questions about the respondent’s own knowledge, attitude and
practice. Do not ask respondents about other people‘s practices as these data
would not be reliable and would potentially be subject to bias.
Specify whether the enumerator should read the list of possible responses
or if the respondent should provide the answer without a list to choose from.
Include this information just after the question itself. It is rare that the
enumerator should read the list before the participant has a chance to
respond. Consider the type of information you would like to collect when
deciding whether to read the list or not.
Specify whether the enumerator should record one or multiple answers.
Following questions that could solicit multiple responses, provide a note to
the enumerator stating either ―circle only one answer‖ or ―circle all that
apply.‖ If you are hoping for multiple answers, include a note to the
enumerator to prompt the respondent by saying ―any others?‖ or ―anything
else?‖ so respondents will know to provide multiple answers.
