A sorry truth in faulty racism survey

PUBLISHED : Tuesday, 21 May, 2013, 12:00am
UPDATED : Tuesday, 21 May, 2013, 9:35am

A world map in The Washington Post giving a visual representation of racial tolerance around the world caused quite a splash last week. It was especially controversial as it made the following observation: "India, Jordan, Bangladesh and Hong Kong by far the least tolerant." It was reported that about 72 per cent of Bangladeshis and Hongkongers said they would not want to be neighbours with someone from a different race.

Several local newspapers including this paper reported on the map and I ran a scathing column yesterday about our racist attitude based on the reports and the map. We hacks now have our tails between our legs.

It turns out that someone "fat-fingered" the original data. As a result, in both cases, the World Values Survey erroneously posted the incorrect data on its website, which The Washington Post then used to create the map. The separate data from Hong Kong and Bangladesh would have to invert the yeses and nos for them to make sense. The survey which has polled attitudes around the world for decades covered 81 countries and territories, and most individual surveys were conducted in local languages.

So the real figures should have been 28.3 per cent of Bangladeshis and 26.8 per cent of Hongkongers who said they wouldn't want neighbours of a different race - not 71.7 per cent and 71.8 per cent! Moreover, 78 per cent of Hongkongers would have immigrants - regardless of race - as neighbours, as opposed to 20.5 per cent against.

Ng Chun-hung, a University of Hong Kong sociologist who conducted the Hong Kong part of the survey in 2005, said he was not aware of the problem until the news stories came out last week. He has now contacted the survey organisers to correct the errors. The Washington Post has also posted a correction. But numerous social scientists and critics have pointed out serious problems in the survey's methods, including the confusing options made available as answers to those polled. One glaring issue is that the values assigned by The Washington Post to different countries are not from the same period, but vary by years.

Apologies for not being more alert to the survey's faulty data and methods used.