Survey and Questionnaire Translation

Why surveys are different

A contract translation succeeds if it says the same thing. A survey translation succeeds only if people answer it the same way. "How satisfied are you?" with a five-point scale seems simple until the scale labels meet a language with no neat word between "satisfied" and "very satisfied", or a culture where the top box is rarely chosen. Items that use idioms ("feeling blue"), double negatives or US-specific examples (a "zip code", "Thanksgiving") need adapting, and the adaptation has to be documented so the analyst knows what changed.

Team translation and back-translation

For research that will be published or compared across countries, we recommend the team model used in cross-cultural survey research: two independent translations, a review meeting that settles the differences, an adjudicator's decision on each disputed item, and a pretest with a few real respondents. Back-translation, where a second translator turns the translation back into English, is widely requested by ethics boards and sponsors. It is a useful check for errors of meaning, but it rewards literal translations, so we use it alongside review, not instead of it.

Response scales and fixed elements

• Scale labels are translated as a set and checked for even spacing, not one by one.
• Answer codes, skip logic and variable names stay in English so the data file is unchanged.
• Programmed surveys (Qualtrics, SurveyMonkey and similar) are translated from the platform's translation export, so piping and display logic survive.
• Open-ended responses are translated back into English for coding, with the respondent ID kept on every line.

Demographic questions need adapting, not just translating

Background questions are where literal translation fails most often. Education levels do not map one to one (a German Abitur is neither a US high-school diploma nor a degree), so offer categories that exist in the respondent's country and map them to your analysis codes. Income bands need local currency and local ranges. US race and ethnicity categories follow federal standards that make little sense to respondents abroad; keep them for US respondents and use the country's own categories elsewhere, coded separately. Addresses, phone formats and name order also differ, so form fields need their validation rules checked for each language.

Schedule and cost

A one-translator job is priced per word like any document. Team translation costs more because two translators work on each item and a reviewer settles the differences, and back-translation adds a third pass. Build the time into the project plan: questionnaires that change after translation starts cost more than ones frozen first, so lock the English source before sending it.

Consent forms

If the study is subject to the Common Rule or FDA regulations, the consent form is the one document that must be translated fully, and IRBs usually want the translation and a certificate of accuracy in the file. We supply both. A short-form consent process is an alternative the regulations allow; your IRB decides which you use.