Validating an existing questionnaire into another language

Question

I am doing research and trying to use a questionnaire that is available in English in another language.

1. Is there a systematic approach for validating a questionnaire in another language?
2. What statistical test should I be undertake and why?

Answer 1

I don't know what your questionnaire aims to assess. In Health-related Quality-of-Life studies, for example, there are a certain number of recommendations for translation issues that were discussed in the following papers (among others):

1. Marquis et al., Translating and evaluating questionnaires: Cultural issues for international research, in Fayers & Hays (eds.), Assessing Quality of Life in Clinical Trials (2nd ed.), Oxford, pp. 77-93 (2005).
2. Hui and Triandis, Measurement in cross-cultural psychology: A review and comparison of strategies. Journal of Cross-Cultural Psychology, 16(2), 131-152 (1985).
3. Mathias et al., Rapid translation of quality of life measures for international clinical trials: Avoiding errors in the minimalist approach. Quality of Life Research, 3, 403-412 (1994).

Translation can be done simultaneously in several languages, as was the case for the WHOQOL questionnaire, or in a primary language (e.g., English) for the SF-36 followed by translation in other target languages. There were many papers related to translation issues in either case that you will probably find on Pubmed. Various procedures have been developed to ensure consistent translation, but forward/backward translation is most commonly found in the above studies.

In any case, most common issues when translating items (as single entities, and as a whole, that is at the level of the questionnaire) have to do with the equivalence of the hypothetical concept(s) that is/are supposed to be covered.

Otherwise, the very first things I would look at would be:

• Am I measuring the same concepts? -- this is a question related to validity;
• Are the scores delivered in the foreign language reliable enough? -- this has to do with scores reliability;
• Are the items behaving as expected for everybody, i.e. without differential effects depending on country or native language? -- this is merely related to differential item functioning (DIF), which is said to occur when the probability of endorsing a particular item differs according to a subject-specific covariate (e.g., age, gender, country), when holding subject trait constant.

Some of the common techniques used to assess those properties were discussed in an earlier related question, Validating questionnaires.

About DIF specifically, here are some examples of subtle variation across subject-specific characteristics:

• In psychiatric studies, "I feel sad" / "Able to enjoy life" have been shown to highlight gender-related DIF (Crane et al., 2007).
• In personality assessment, there are well-known age and gender-effect on the NEO-PI questionnaire (reviewed in Kulas et al., 2008).
• In Health-related Quality-of-Life, items like "Did you worry?" / "Did you feel depressed?" have been shown to exhibit country-related DIF effect (Petersen et al. 2003).

A good starting point is this review paper by Jeanne Teresi in 2004: Differential Item Functioning and Health Assessment.

Answer 2

I found some good ideas in the short Sage book, Translating Questionnaires and Other Research Instruments, at http://www.uk.sagepub.com/books/Book5861 . I wouldn't call it the most systematic or the most entertaining read, but it was helpful and it's fairly up-to-date and inexpensive.

Answer 3

Wakening up an old question I just wanted to add two sources concerning guidelines for translating questionnaires. The first one would be EORTC's Guidelines and the second PROMIS's Guidelines. Reading these will help you grasp the concept of translating according to every possible detail. From there, you can make your own decisions about how precise to be (although I hope you're already done since your question was asked some years ago...)

Answer 4

The procedure i have normally seen followed is to translate the questionnaire from english, and then have it back translated by someone else. If the two english translations match up, then you are good to go, otherwise repeat until they do.