The 2015 publication of what has become known as the the "Chinese-sounding-names" story on Auckland home ownership was, says Harkanwal Singh, "a really pivotal moment for me, working in a New Zealand newsroom. Because that's when I realised that things don't have to be true to be published."
Singh was working as the New Zealand Herald's first dedicated data journalist and was at the meeting where Labour Party MP Phil Twyford and party researcher Rob Salmond brought in their data – which they said showed a hitherto unsuspected level of Chinese foreign ownership in Auckland housing.
"They said 'we're not being racist' as they handed over the data set," he told Jogai Bhatt and I at last Sunday's Orcon IRL.
Singh's questions over the data delayed publication by a week. During that week he contacted Auckland University's Thomas Lumley and Edward Abraham of Dragonfly Data Science ("the best statisticians in the country").
"And I went back to my editors and I said, look, you should publish it, but you should say that Labour is saying this – and the statisticians are saying that it's not true."
His suggestion was not taken up by his editors.
"The story ran with the headline 'We have Chinese buyers' and and all I did was add some bullet points which said 'this data is wrong'. But they were published on the fifth page, inside, in a little box, so no one really saw them.
"It was hugely problematic and as a immigrant and as a person of colour, I saw a huge problem with it. But no one else in the newsroom saw any problem with it. And when I approached senior journalists I was told 'it's a great story'.
"I think it's still not been addressed and no one's really addressed how they went about doing it. And it's a huge issue of data literacy if you're just going to publish analysis done by political parties for their own goals."
The whole 11 minutes of the IRL discussion is, I think, worth your time watching. Harkanwal is unfailingly polite and pleasant and occasionally pretty brutal on issues of data and diversity: