Posts written by Thomas Lumley (2645)

If you have time, there’s an interesting event that will be streamed from New York University this (NZ) morning (10:30am today NZ time, 5:30pm yesterday NY time)

..the Data & Society Research Institute, the White House Office of Science and Technology Policy, and New York University’s Information Law Institute will be co-hosting a public event entitled The Social, Cultural, & Ethical Dimensions of “Big Data.” The purpose of this event is to convene key stakeholders and thought leaders from across academia, government, industry, and civil society to examine the social, cultural, and ethical implications of “big data,” with an eye to both the challenges and opportunities presented by the phenomenon.

The event is being organised by danah boyd, who we’ve mentioned a few times and whose new book I plan to write about soon.

The latest BNZ-REINZ Residential Market Survey is out, and the Herald has even embedded the full document in their online story, which is a very promising change.

According to the report 6.4% of homes sales in March are  to off-shore buyers, 25% of whom were Chinese. 25% of 6.4% is 1.6%.

If you look at real estate statistics (eg, here) for last month you find 6125 residential sales through agents across NZ. 25% of 6.4% of 6125 is 98. That’s not a very big number.  For context, in the most recent month available, about 1500 new dwellings were consented.

You also find, looking at the real estate statistics, that last month was February, not March.  The  BNZ-REINZ Residential Market Survey is not an actual measurement, the estimates are averages of round numbers based on the opinion of real-estate agents across the country.  Even if we assume the agents know which buyers are offshore investors as opposed to recent or near-future immigrants (they estimate 41% of the foreign buyers will move here), it’s pretty rough data. To make it worse, the question on this topic just changed, so trends are even harder to establish.

That’s probably why the report said in the front-page summary “one would struggle, statistically-speaking, to conclude there is a lift or decline in foreign buying of NZ houses.”

The Herald  boldly took up that struggle.

The cosmologists are excited today, and there is data visualisation all over my Twitter feed

RT @Resonaances: Parameter space of inflation as of today: pic.twitter.com/i45NLQX9DI Honestly didn’t think I was going to see > 0 for a while!

— Renee Hlozek (@reneehlozek) March 17, 2014

That’s a nice display of uncertainty at different levels of evidence, before (red) and after (blue) adding new data.  To get some idea of what is greater than zero and why they care, read the post by our upstairs neighbour Richard Easther (head of the Physics department)

Q: Did you see the story about aphrodisiacs on Stuff this weekend?

A: Yes

Q: How did they find out which ones worked?

A: It says “Richard Cornish investigates the only way he knows how.”

Q: Randomised n-of-1 trials with independent evaluation by someone who doesn’t know what he’s eaten?

A: Sadly, no.

Q: Allocating different foods, and some control foods, to a large group of people and collecting their reports?

A: No

Q: Getting a librarian to help him review the scientific research on the topic? Or the traditional knowledge?

A: Not really, though there are some biochemical or historical anecdotes for many of the items.

Q: Um. Did he just try each food as you would if you wanted to use it as an aphrodisiac?

A: Not that, either.

Q: I give up. What did he do?

A: ” It was my task to consume them in a bland environment, with no chance of any stimulation or excitement.”

Q: What a waste. But aren’t you being a bit harsh?  He’s a food writer and TV producer. He does sustainability and Spanish food. He’s not a science journalist or an investigative reporter.  They didn’t expect anyone to take it seriously.

A: Ok, but some of the nutrition stories and sex stories they run are supposed to be taken seriously. It should be easier to tell which is which online.

Q: Wait, isn’t it March now?

A: Yes.

Q: That sounds more like a Valentine’s Day column

A: An interesting point. You thought of that faster than I did.

Q: Well?

A: It is a Valentine’s Day column. From the Southland Times. Except they took out the foie gras and truffles to make it suitable for the national audience. Reruns aren’t just for The Simpsons, you know.

• Your high school transcript will haunt you forever? Partly reasonable, but partly the recurring claim that knowing what works in a given situation will take away your freedom.

• Both men and women think women are bad at math. “Men and women employers alike revealed their prejudice against women for a perceived lack of mathematical ability. When the only information that the employers had was a photograph of the candidate, men were twice as likely to be hired for the simple math job, no matter whether it was a man or woman doing the hiring”. Could do with a lot more scare quotes around words like `employer’, `candidate’, `hired’, but a good report.

• Why Google Trends can’t track the flu (yet) There are lots of differences between flu season and the rest of the year, and the ones most relevant to Google’s ordinary activities don’t have much to do with influenza

The Herald said, on Friday “Fewer workers stoned on the job”

Information from the New Zealand Drug Detection Agency showed 81,410 on-site drug-screening tests were carried out last year, 16 per cent up from the previous year.

But only 5.5 per cent of tests showed the presence of drugs, down from 6.4 per cent in 2013

As usual, there’s no mention of the fact that NZDDA is just one of the private companies offering drug testing services. It took me a long time to realise this, until I was tipped off by a news story advertising one of their competitors.

Presumably NZDDA don’t think their customers choose them at random, and with no real reason for wanting testing. If customers were behaving even a little rationally you’d expect an expansion of drug testing to pull in lower-risk employees. If we look at the actual number of positive tests, using the quoted figures, it was about 4480 last year and about 4490 in the previous year. Given no change in the number of positive tests and a 1 percentage point change in the proportion of positive tests, from a single company, there’s not a lot of numerical evidence for an increase in number of workers with detectable cannabis in their systems.

More importantly, there’s no evidence whatsoever for the ‘stoned on the job’ headline: absolutely no information is given about this. One of the big problems with cannabis testing is that there is no useful biochemical assay for being stoned. Detectable levels persist long after impairment is over, and even when you’re actually stoned there is not a good relationship between drug concentration and impairment.  This is a real problem for Washington and Colorado, which have legalised cannabis and need to set driving laws. In contrast to alcohol, if you actually care about safe driving and cannabis, it’s really hard to get a useful and objective test.

The story ends with two examples of disasters. In one, cannabis was definitely ruled out as a contributing factor; in the other, the conclusion was only that “it could not be excluded”. The NZDDA  press release is at Scoop, and despite how the story reads, there is surprisingly little text in common.

The Sunlight Foundation has released its internal data visualisation style guidelines (via)

Some of it is just one organisation’s house style, but it’s an attractive style. Some of it is generally useful advice. The main thing they are missing is any advice on representing uncertainty.

• Ok, it’s not statistics as such, but you have to see the mash-up of the Cyclone Lusi animation and Van Gogh’s Starry Night, by @bobsyouruncle

• Cathy at Mathbabe.org on survey questions, pointing out that the sun does too go around the earth.

• Buzzfeed says “According To Pornhub, The South Watches More Gay Porn Than Any Other Part Of The U.S.” The point, for those of you not up on US sociogeography, is that the South is religiously conservative.  It turns out that’s not really what the data say. The figures aren’t % of men who watch gay porn, they are % of porn that is gay male. The data are equally consistent with straight guys in religiously conservative states watching very slightly less porn than those in other states. Data on total porn consumption are mixed.

• ProPublica, the non-profit, public-interest journalism foundation in the US, are setting up a data shop. Data that they could just download, they’ll make available for free, but the data that took a lot of effort will cost. Interesting to see this as a data journalism funding model.

• From ProPublica, a good example of simple arithmetic applied to unreasonable claims.

Since 2009, Dagogo-Jack has been paid at least $257,000 by Glaxo, Lilly and Merck.

“If you actually prorate that by the hours put in, it is barely more than minimum wage,” he said. (A person earning the federal minimum wage of $7.25 would have to work 24 hours a day, seven days a week for more than four years to earn Dagogo-Jack’s fees.)

Cyclone Lusi, from the earth wind animation

And coloured by total precipitable water (orange: dry, light blue: very wet)

Keep safe.

The Herald has a story about a potential blood test for dementia, which gives the opportunity to talk about an important statistical issue. The research seems to be good, and the results are plausible, though they need to be confirmed in a separate, larger sample before they can really be believed. Also, the predictions so far are just for mild cognitive impairment, not actual dementia. But it’s the description of the accuracy of the test that might be misleading.

The test had 90% sensitivity — 90% of who developed cognitive impairment tested positive. It had 90% specificity — 90% of those who who did not develop cognitive impairment tested negative.  That’s what is described in the story as 90% accuracy.  What a user would care about is the positive predictive value: if you test positive, how likely are you to get cognitive impairment?

In the study 451 people started out cognitively normal; 28 of these developed impairment, the other 423 did not. The test would be correctly positive for about 25 of the 28, and correctly negative for about 381 of the 423. So, of the 25+42=67 who test positive, less than 40% will develop impairment. That’s reasonable for a  diagnostic test but a bit low for a screening test in healthy people.

Where the test is more immediately relevant is in designing clinical trials. So far, attempts to affect Alzheimer’s Disease progression have failed, though there are some modestly effective symptomatic treatments. It’s possible that the treatments are doing the right thing but that clinical illness is too late, so there’s a lot of interest in testing treatments very early in the process. A test like the new one could be very useful