On Saturday, the New Zealand Herald released data provided by the Labour Party that attempted to provide figures around the impact of foreign buyers in the New Zealand property market.

This has caused quite a stir on left-wing blogs and on Twitter. Some people with foreign sounding names have stood up and said that the use of this data is racist or possibly poorly-conceived. I tend to agree with them. Some white folk have said that it’s not racist at all because numbers don’t lie. They’ve got a point too.

I thought I should weigh in here, since as a middle-aged white bloke I don’t really know anything about racism.

Anyway, the problem I’m seeing here is that the two sides of this argument aren’t even arguing the same thing.

On one side, Rob Salmond and the Labour Party have taken ethnicity data from the electoral roll, used that to correlate surnames with ethnicity, then from that worked out the likely ethnicity of property buyers in Auckland. This has used 3 months of data from one real estate company that apparently comprises 45% of sales in Auckland during that period.

This data shows that of those buyers, 39.5% have names that are likely to be of Chinese origin whilst the electoral roll has only 9% of people with Chinese names. Ergo, foreign buyers are a problem.

Rob Salmond has written at length to describe the effort made to investigate other possible causes for the disparity between the electoral roll and property buyers. Do Chinese earn more? Do they have more assets? He can’t find anything, so it must be foreign buyers.

So, what’s the problem?

There are numerous problems that I see with this analysis.

Firstly, the 39.5% figure is treated as an absolute fact. It is rather precise, presented as a percentage to 1 decimal place. But it’s not precise, it’s an estimate based Bayesian inference. Labour are inferring ethnicity based on names on the electoral roll. It’s unclear to me whether the source for this is solely the electoral roll or the electoral roll in conjunction with census data. If it is the latter then it is clearly less reliable.

Then they are taking real estate data of unknown provenance. We are told that it comprises 45% of the data for Auckland during the relevant period. This is a pretty big sample and it should be reasonable to use it. But we know nothing about it, other than that it came from one firm. It is not a random sample of data so that may skew the outcome of analysis. It could make the final figure of 39.5% either higher or lower than in the whole market.

From the ethnicity data and the real estate transactions, they are inferring again the ethnicity of buyers. This seems like a school project or data that might be useful within the Labour Party. It is so unreliable that it should never be publish. Even if it’s accurate, we cannot possible infer the country of origin of buyers.

But my biggest problem is that the data is letting New Zealanders draw the conclusion that 39.5% of buyers are foreign Chinese. Labour isn’t coming out directly and saying this because it’s not true, but in the immortal words of John Key, “At the end of the day, New Zealanders will make up their own mind about this”.

The data only refers to likely ethnicity. You cannot infer with any certainty whether these buyers are New Zealanders or not. It is highly unlikely that they are all NZ Chinese but we have no idea. Absolutely no idea.

Labour has been very careful not to make that inference themselves but it is hard not to unconsciously make that mental leap yourself, especially when the media are constantly referring to “Chinese buyers” and “ethnic Chinese”. Who wouldn’t assume that means Chinese people, from China? “You might very well think that but I couldn’t possible comment.”

Finally, this data analysis appears incomplete because we have no idea how accurate this is. In an opinion poll, we have a margin of error. In other forms of statistical analysis, we have the concept of statistical significance. I’ve seen no documentation anywhere of whether these results are statistically significant. This is very important in making these assertions.

So, to summarise, the Labour Party is saying that the number of buyers (based on name) that are of Chinese ethnicity differs markedly from the number of residents (based on name) that are of Chinese ethnicity. There is some merit there. As I understand it, those accusing the Labour Party of racist tactics are saying that the “good” Chinese (NZ residents) are being lumped in with the “bad” Chinese (foreigners) and the data analysis is incomplete. There is also some merit there.

If I were a Chinese New Zealander, I would be pretty pissed off about the release of this data.

Chinese buyers (by country) are allegedly the enemy, but Chinese buyers (by name) are the proxy. The use of this proxy works for creating dodgy statistics, but it also works for creating a target: all those Chinese faces you see at auctions, whether they are Chinese by citizenship or just happen to look Chinese and maybe have a Chinese name. They are now the target.

The inference that I take from Labour’s use of this data is that the Labour Party is accusing Chinese-New Zealanders of being part of the 39.5% of the housing market that is the problem. And that’s not acceptable to me.

Note: in all of this I have expressed no opinion about the state of the property market or whether foreigners should be allowed to buy property here. This might be an issue, or it might not be. Vilifying New Zealanders with Chinese names is also an issue.