NewsHong Kong

How dangerous data can cast global shadow

Social scientists warn about use of openly available research without questioning how it was collected and what it really means

PUBLISHED : Friday, 24 May, 2013, 12:00am
UPDATED : Friday, 24 May, 2013, 3:51am

In the fallout from The Washington Post publishing an incorrect map on global racial tolerance, Hong Kong's social scientists are issuing a caution: data is useful, but watch how you use it.

Last week, the US newspaper published the map purporting to show nations' racial tolerance using openly available information from the World Values Survey. The map made Hong Kong and Bangladesh out to be the most racist places on earth, with 71.8 per cent of Hongkongers and 71.7 per cent of Bangladeshis opposed to living next to people of a different race.

The real figures were closer to 27 per cent and 28 per cent respectively. The Washington Post wrote in its correction that it had used erroneous data posted on the World Values website.

"Open data in itself is very good," said Susanne Choi Yuk-ping, a sociologist with the Chinese University of Hong Kong. "It increases transparency and the capacity of citizens and civil society to monitor what's going on.

"Statistics are useful, but also dangerous. This [racial tolerance] was a seriously sensitive issue, [ The Washington Post] should have been more careful.

"NGOs and media organisations may not have first-hand knowledge about the data collection and the questionnaire design. They often don't have sufficient professional knowledge to judge [if the analysis is correct]."

She added: "This mistake is a good thing. Making data openly available has only become very common in Asia over the past 10 years. This will make researchers more careful with their data."

Choi also suggested researchers highlight limitations and guidelines for use when making their data available.

The Washington Post's decision to use the data has also been questioned. The information dated back to the early 1990s; the idea of race is different in each country; some of the responses are confused; and a country's tolerance levels were unlikely to be gauged by one question.

Journalist Siddhartha Mitter pointed out that in 2000, only 0.9 per cent of respondents from Iran objected to having a homosexual neighbour, whereas in 2007, 92.4 per cent objected .

"It could be that something happened that year in that country to skew results," said Ng Chun-hung, the Hong Kong University professor who conducted research on behalf of the World Values Survey in Hong Kong in 2005. He said it was vital people using open data speak to the original researchers to understand the methodologies.

The Washington Post example was not the only questionable use of data from surveys. The Democratic Alliance for the Betterment and Progress of Hong Kong's poll in April reports that 70 per cent of Hongkongers were not supportive of a planned blockade of Central roads to demand genuine universal suffrage.

"I believe most of the people they interviewed were over 50 years old," Choi said. "It's not representative of the whole of Hong Kong. The political motivation there was pretty clear."

She added that surveys on abuse of the elderly could have similar problems. "If you're only interviewing, say, 50 old people from a few homes for the elderly and you find instances of abuse, you can't say that this is the case across Hong Kong," Choi said.




Send to a friend

To forward this article using your default email client (e.g. Outlook), click here.

Enter multiple addresses separated by commas(,)

For unlimited access to: SCMP Tablet Edition SCMP Mobile Edition 10-year news archive