Recently, Australia’s ABC News profiled Erin Zhao, who did a good linguistic deed for Australia’s Chinese community. But criticism reveals troubling underlying truths about the nature of census and survey translation. Despite being home to a number of groups with limited English proficiency, Australia’s census is only available in English. While family members often help translate questions, the pandemic has complicated things, with lockdown measures in place in many areas that prevent families from working together on the same form. This moved Zhao to take action. A Chinese immigrant herself, she spent 10 hours translating the English form into Chinese. At first, Zhao only shared her translation with family and friends. But word got out and Australia’s Chinese community began to use it widely. Zhao’s action is as understandable as it is commendable, but the Australian Bureau of Statistics (ABS) isn’t a fan, telling the news site that amateur translations like Zhao’s pose a data risk. They’re not wrong. Survey and census translation is a task full of pitfalls that an amateur translator, or even a single professional translator, may not be aware of. For instance, the ABS points out that when you translate a word into another language, you must be aware of all of its connotations and meanings. Another issue is culture. Although Zhao is likely very familiar with both Chinese and Australian culture, this may not be the case for all translators. Cultural differences can affect everything from questions about respondents’ habits to even less obvious things, like the very way they respond. For example, this study on healthcare survey translation notes that even the wording typically used for survey responses (“agree”, “disagree”, “strongly agree”, “not sure”, etc.) may not exist in certain cultures. The issue can even go beyond words. A Pew Research article suggests that the “X” symbol that’s typically used to indicate a correct/chosen response on the US Census form should ideally be replaced in an Arabic translation, since it’s not used this way in that language. Survey translation can get even more complex when it comes to healthcare and pharma. In addition to linguistic, cultural, and formatting concerns, specific problems may arise around the accurate way to translate the name of a particular device. In this case, professional translators will have to reach out to the company for the official terminology. None of what I’ve written here is meant to chastise Zhao; she’s done something admirable and necessary. Although the ABS claims it offers support to non-English speakers, this wasn’t sufficient for members of the Chinese community, at least in part due to lockdown restrictions. So, Zhao stepped up and took action. But in an ideal situation, current studies show that census and survey translations are best performed by a group of professional translators who will translate, back-translate, question, test, and verify the translations for the best possible accuracy. You can get a behind-the-scenes glimpse at how a multilingual questionnaire is translated by professionals here. You can also find a concise description of how the US Census Bureau creates translations via committee, in this article. The fact that not all survey and census material may be translated under ideal circumstances is a harsh reminder that not all data we gather from translations may be 100% accurate. Still, if you find yourself in a situation similar to Zhao’s or if you’re a small business owner who may not want or be able to pay for a professional translation team, there are some ways to check that your translation will be as accurate as possible. This article, for instance, advises getting to know the broad rules of your target language. For example, if you’re translating into Chinese (or if you’re only able to hire a single translator to do this for you), understanding that there are two forms of Chinese writing, traditional or simplified, and then choosing the appropriate one for your audience, is a good place to start. Another strategy is to reach out to native speakers of your target language through crowd sourcing or local organizations, to see what they think of your translation. Is it easy to understand and grammatically correct? Does it raise any questions or cause confusion? Hopefully, these strategies will help amateur translators who have no other choice but to translate a survey or census themselves. That said, a group of professional translators is always the best option for this very important, surprisingly complex type of translation, whose accuracy could literally determine how we see the world.