Survey translation: Cultural biases to look out for

Survey translation: Cultural biases to look out for

When planning to administer a survey in multiple languages, take these cultural issues into account before survey translation begins.  Even well-translated surveys can be biased by cultural issues. Responses can vary depending on the culture of the respondent. Here are some issues that might arise, and how to control for them.

Moderation or extremism?

A popular format for response categories is the Likert scale. The Likert scale asks a respondent to indicate on a scale of one through five how strongly they agree or disagree with a statement. Cross cultural experts warn of that some cultural groups are more comfortable choosing extreme answers (strongly agree, strongly disagree). Members of other groups are more likely, on average, to choose answers toward the middle (neither agree nor disagree). This variation occurs regardless of the content of the question. Some researchers suggest that the difference lies in cultural values about the acceptability of extreme positive or negative opinions.

Solution: To better capture degrees of individual variation, the survey designer might use a scale of 1-10 rather than 1-5. This allows for finer gradations of opinion.

Which end is up?

A second problem with Likert scales is that some cultures may be accustomed to seeing 1 as the most positive end of a scale and 5 as the most negative, while others may be used to the reverse. In the United States, some researchers have noticed that respondents tend to choose answers nearer to the far left of the scale regardless of content, possibly because of patterns of perception; one wonders whether cultures whose language reads right-to-left (Arabic, Hebrew) might favor the far right end of the scale?

Solution:  To control for this tendency, the survey designer might survey half the respondents using a scale that runs from left to right, and the other half with a scale that runs the opposite direction.

Can they ask me that?

Another source of confusion can come from demographics measures. Asking about income levels and ethnicity might feel intrusive and offensive in some cultures.

Solution:  When creating and localizing survey instruments, work with cultural consultants who can help you ask these questions in acceptable ways.

Apples v. oranges

Be careful about assuming cross-cultural equivalence. For example, measuring levels of education can present difficulties. It is not clear whether the O-levels in Pakistan are the educational equivalent of the GCSE in England and the GED in the United States. Furthermore, the assessment of the value of a University degree versus a technical school certificate might be different for residents of one country than for another.

Solution:  Cultural consulting can be helpful in this instance as well.

What do you want me to say?

Finally, a survey developer might consider whether respondents could be biased toward answering questions in ways that are socially acceptable. For example, if deference to authority is a widely held cultural value, invitations to rate or evaluate managers may trend toward favorable ratings. People may feel similarly regarding their peers. Even if you assure the respondents that individual answers are going be kept anonymous, faith in the preservation of “privacy” and “anonymity” vary widely across cultures. You might not know whether your survey measures what people actually think or what they know they are supposed to be thinking. Here is where a cultural assessment is crucial to design questions that are specific enough to gain useful data, but that are non-threatening to core social values.

Be respectful

When you ask people to participate in a survey, you are asking for a favor. An employee engagement survey, for example, does more than measure attitudes and beliefs. It also communicates to employees what management thinks is important. Although there are lots of things that might be “nice to know,” you must be careful not to include questions about issues or problems that you have no intention of addressing. Survey participants may resent giving answers to questions that were asked “in bad faith,” and this may reinforce feelings of mistrust.

Finally, in global corporations, employees may speak a different language than management. If you plan to share it, be sure to translate the report of the survey findings to a language that employees understand. Finally, always rely on an experienced language service partner to get the most out of your survey translation project.