DHS6_Survey_Org_Manual_7Dec2012_DHSM10 ... - Measure DHS

D. TRANSLATING QUESTIONNAIRES
The main aim of a good survey instrument is to minimize the number of errors that are
made when obtaining the desired information. Interviewers can obtain answers that are both valid
and reliable only if they are using well-designed questionnaires. Validity is achieved if the question
elicits a response that is true and accurate, and measures what is intended to be measured. A good
questionnaire should produce valid measures of the indicators intended to be measured, by helping
to ensure that the respondent understands what information is being sought. Reliability is achieved if
the question elicits a consistent response, no matter who asks the question or where and when it is
asked. Using a well-designed questionnaire, each question is asked in the same way by every
interviewer and differences between interviewers will be kept to a minimum.
In order to maximize validity and reliability, DHS policy is to use questionnaires that have
been translated into and printed in all the major local languages in which interviews are expected to
take place. Studies have shown that when interviewers themselves translate from the language in the
printed questionnaire into a different language as they ask the questions, they tend to introduce
errors that can often substantially change the meaning of the question. Even small differences of
interpretation can destroy the validity and reliability of the final data. As a rough rule, any language
group that constitutes 10 percent or more of the sample should have its own translated questionnaire.
The need for on-the-spot translation by the interviewer or someone else often cannot be
avoided totally, as there may be no adequate language version of a questionnaire for some respondents
who fall in the sample. However, the need for on-the-spot translation should be minimized.
Translation is not an easy task and requires both strong linguistic skills as well as an
understanding of terms and expressions used in the questionnaires. The recommended approach is
that someone who understands both the technical terms as well as both languages translates the
questions into the required local language. DHS discourages the use of language institutes or local
language radio program hosts to make the translations as this often leads to wording that is too
formal or stilted or too imprecise. If there has been an earlier DHS or similar survey that was
translated, that translation should be taken into account. The translated questionnaire should be
back translated into its original language by an independent translator—preferably someone who is
not acquainted with the technical subject matter—without reference to the original questionnaire.
The back translation should be compared to the original questions and any discrepancies should be
resolved by the two translators in collaboration with survey technical staff.
This process should result in questionnaires that are well understood by respondents who
are interviewed in the respective language. It is important to remember that the purpose of the
translation is to ensure that every respondent is asked the same question, that they understand the
question, and that they feel more comfortable being interviewed in that language. This does not
mean that translation should be literal. A good translation will transmit the intended meaning,
although it may not be a word-for-word translation.
Survey documentation such as interviewers’ and supervisors’ manuals should be translated
into a language that is understood by all the field staff.
18 | DHS Survey Organization Manual