Stats Chat

Team Ratings for Round 12

The basic method is described on my Department home page.

Here are the team ratings prior to this week’s games, along with the ratings at the start of the season.

Performance So Far

So far there have been 84 matches played, 47 of which were correctly predicted, a success rate of 56%.

Here are the predictions for last week’s games.

Predictions for Round 12

Here are the predictions for Round 12. The prediction is my estimated expected points difference with a positive margin being a win to the home team, and a negative margin a win to the away team.

Team Ratings for Round 16

The basic method is described on my Department home page.

Here are the team ratings prior to this week’s games, along with the ratings at the start of the season.

Performance So Far

So far there have been 99 matches played, 66 of which were correctly predicted, a success rate of 66.7%.

Here are the predictions for last week’s games.

Predictions for Round 16

Here are the predictions for Round 16. The prediction is my estimated expected points difference with a positive margin being a win to the home team, and a negative margin a win to the away team.

In an interesting sting on the world of science journalism, John Bohannon and two colleagues, plus a German medical doctor, ran a small randomised experiment on the effects of chocolate consumption, and found better weight loss in those given chocolate. The experiment was real and the measurements were real, but the medical journal  was the sort that published their paper two weeks after submission, with no changes.

Here’s a dirty little science secret: If you measure a large number of things about a small number of people, you are almost guaranteed to get a “statistically significant” result. Our study included 18 different measurements—weight, cholesterol, sodium, blood protein levels, sleep quality, well-being, etc.—from 15 people. (One subject was dropped.) That study design is a recipe for false positives.

Think of the measurements as lottery tickets. Each one has a small chance of paying off in the form of a “significant” result that we can spin a story around and sell to the media. The more tickets you buy, the more likely you are to win. We didn’t know exactly what would pan out—the headline could have been that chocolate improves sleep or lowers blood pressure—but we knew our chances of getting at least one “statistically significant” result were pretty good.

Bohannon and his conspirators were doing this deliberately, but lots of people do it accidentally. Their study was (deliberately) crappier than average, but since the journalists didn’t ask, that didn’t matter. You should go read the whole thing.

Finally, two answers for obvious concerns: first, the participants were told the research was for a documentary on dieting, not that it was in any sense real scientific research. Second: no, neither Stuff nor the Herald fell for it.

 [Update: Although there was participant consent, there wasn’t ethics committee review. An ethics committee probably wouldn’t have allowed it. Hilda Bastian on Twitter]

“In Australia road deaths are going down but in New Zealand the number has shot up“, says the Herald, giving depressing-looking international comparisons from newly-announced OECD data. The percentage increase was highest in New Zealand The story does go on to point out that the increase reverses a decrease the previous year, suggesting that it might be that 2013 was especially good, and says

“An ITF spokesman said New Zealand’s relatively small size made percentage movements more dramatic.”

Overall, it’s a good piece. Two things I want to add: first, it’s almost always useful to see more context in a time series if it’s available. I took the International Road Traffic Accident Database and picked out a group of countries with similar road toll to New Zealand in 2000: all those between 200 and 1000. The list is Austria, Denmark, Finland, Ireland, Israel, New Zealand, Norway, Slovenia, Sweden, Switzerland. Here are the data for 2000 and for 2010-2014; New Zealand is in red.

There’s a general downward trend, but quite a bit of bouncing around due to random variation. As we keep pointing out, there are lots of mistakes made when driving, and it takes bad luck to make one of these fatal, so there is a lot of chance involved. It’s clear from the graph that the increase is not much larger than random variation.

Calculations using the Poisson distribution (the simplest reasonable mathematical model, and the one with the smallest random variation) are, likewise, borderline. There’s only weak evidence that road risk was higher last year than in 2013. The right reference level, though, isn’t ‘no change’, it’s the sort of decrease that other countries are seeing.  The median change in this group of 10 countries was a 5% decrease, and there’s pretty good evidence that New Zealand’s risk did not decrease 5%.  Also, the increase is still present this year, making it more convincing.

What we can’t really do is explain why. As the Herald story says, some of the international decrease is economic: driving costs money, so people do less of it in recessions. Since New Zealand was less badly hit by recession, you’d expect less decrease in driving here, and so less decrease in road deaths. Maybe.

One thing we do know: while it’s tempting and would be poetic justice, it’s not valid to use the increase as evidence that recent road-safety rule changes have been ineffective. That would be just as dishonest as the claims for visible success of the speed tolerance rules in the past.

This isn’t precisely statistics, more applied probability, but that still counts.  First, an interactive from Lewis Lehe, a PhD student in Transport Engineering at UC Berkeley. It shows why buses always clump together.

You might also like his simulations of bottlenecks/gridlock and of congestion waves in traffic (via @flowingdata)

And second, a video from the New York subway system. When a train gets delayed, it holds up all the trains behind it. More surprisingly, the system is set up to delay the train in front of it, to keep the maximum gap between trains smaller.

The Herald has a story with data from the General Social Survey. Respondents were asked if they would feel comfortable with a neighbour who was from a religious minority, LGBT, from an ethnic or racial minority, with mental illness, or a new migrant.  The point of the story was that the figure was about 50% for mental illness, compared to about 75% for the other groups. It’s a good story; you can go read it.

What I want to do here is look at how the 75% varies across the population, using the detailed tables that StatsNZ provides. Trends across time would have been most interesting, but this question is new, so we can’t get them. As a surrogate for time trends, I first looked at age groups, with these results [as usual, click to embiggen]

There’s remarkably little variation by age: just a slight downturn for LGBT acceptance in the oldest group. I had expected an initial increase then a decrease: a combination of a real age effect due to teenagers growing up, then a cohort effect where people born a long time ago have old-fashioned views. I’d also expected more difference between the four questions over age group.

After that, I wasn’t sure what to expect looking at the data by region. Again, there’s relatively little variation.

For gender and education at least the expected relationships held: women and men were fairly similar except that men were less comfortable with LGBT neighbours, and comfort went up with education.

Dividing people up by ethnicity and migrant status was a mixture of expected and surprising. It’s not a surprise that migrants are happier with migrants as neighbours, or, since they are more likely to be members of religious minorities, that they are more comfortable with them. I was expecting migrants and people of Pacific or Asian ethnicity to be less comfortable with LGBT neighbours, and they were. I wasn’t expecting Pacific people to be the least comfortable with neighbours from an ethnic or racial minority.

As always with this sort of data it’s important to remember these responses aren’t really level of comfort with different types of neighbours. They aren’t even really what people think their level of comfort would be with different types of neighbours, just whether they say they would be comfortable. The similarity across the four questions makes me suspect there’s a lot of social conformity bias creeping in.

Has the cure for cancer been with us all along? A vaccine in development. The details, tonight at seven. (Yes we are still here!)

— Campbell Live (@CampbellLiveNZ) May 25, 2015

The segment was about research from the Malaghan Institute, who are working on ways to encourage a patient’s immune system to attack tumours. They say, in a press release

While the research will focus specifically on targeting melanoma, it is anticipated that the methodology being developed could be applied to other cancers in the future.

The therapeutic vaccine approach differs from the preventative vaccines used to protect against diseases such as measles or the flu because the cancer vaccine is designed to be given to an individual after they have already shown signs of disease.

“It is known that white blood cells called T cells can kill tumour cells,” says Dr Hermans. “The cancer vaccines, which are custom-made for each cancer patient, are designed to stimulate the activity of these cancer-fighting immune cells.”

As the press release makes clear, the term ‘vaccine’ is technically correct, but liable to mislead: these are customised immune-system treatments specific to one tumour in one individual. It’s nothing like the measles vaccine that you get as an infant for lifetime protection.

Like other research groups, the Malaghan Institute are starting with melanoma. There are at least two reasons melanoma is a good place to start. The simple reason: until very recently, metastatic melanoma was completely untreatable, so anything would be an improvement. There’s also a complex reason: melanoma occasionally shrinks or vanishes of its own accord, apparently more often than other tumours do. The spontaneous regressions are presumably thanks to the immune system waking up and realising the tumour is a problem, so melanoma is a good starting point if you want to find out how this happens and encourage it to happen more often.

The basic problem is that the immune system tends to see cancer cells as part of the patient, since, fundamentally, they are. The Malaghan Institute has a innovative addition to the treatment, a chemical that scares specific parts of the immune system into action. They expect that combining this with the existing tumour vaccine approaches will give a more reliable result.

Malaghan got $4.5 million from the Health Research Council to work on this, which is pretty impressive given the HRC budget and competition, but the melanoma vaccine is still in the initial stages of testing. The majority of promising treatments going in to Phase I clinical trials don’t end up being useful. Even if this one does, that’s no guarantee it will work for other types of cancer, and while it’s a vaccine in the sense that it works by stimulating an immune reaction, it’s nothing like the vaccines we give to kids.

It’s not a good time to criticise Campbell Live, but although the Malaghan’s research is truly impressive, it’s not a general-purpose cure for cancer anytime this decade. And there’s a basic principle that you shouldn’t say “cure” in the headline unless there’s a cure.

New York Times columnist Richard Friedman is writing about hormones, genetics, and infidelity.  This paragraph is about recently-published research by Brendan Zietsch and colleagues (the NYT tries to link, but the URL is wrong)

His study, published last year in Evolution and Human Behavior, found a significant association between five different variants of the vasopressin gene and infidelity in women only and no relationship between the oxytocin genes and sexual behavior for either sex. That was impressive: Forty percent of the variation in promiscuous behavior in women could be attributed to genes.

If you didn’t read carefully you might think this was a claim that the  vasopressin gene association explained the “Forty percent” and that the percentage was lower in men. In fact, the vasopressin gene associations are rather weaker than that, and the variation attributed by the researchers to genes is 62% in men.

But it gets worse. The correlation with genetics was only seen in identical twins. That is, pairs of identical twins had fairly similar cheating behaviour , but there was no similarity at all between pairs of non-identical twins (of any gender combination) or between non-twin siblings.  If that’s not due to chance (which it could be), it’s very surprising. It doesn’t rule out a genetic explanation — but it means the genetics would have to be weird.  You’d need either a variant that had opposite effects with one versus two copies, or a lot of variants that only had effects with two copies and no effect with one, or an effect that switched on only when you had variant copies of multiple genes, or an effect driven by new mutations not inherited from parents.  The results for the vasopressin gene don’t have this kind of weird.

The story is all “yes, it’s surprising that you’d get this sort of effect in a complex social behaviour, but genetics! And voles!”. I’ll give him the voles, but if anything, the strong correlation between identical twins (only) argues against vasopressin gene variants being a major driver in humans, and the research paper is much more cautious on this point.

Each week, we would like to invite readers of Stats Chat to submit nominations for our Stat of the Week competition and be in with the chance to win an iTunes voucher.

Here’s how it works:

• Anyone may add a comment on this post to nominate their Stat of the Week candidate before midday Friday May 29 2015.
• Statistics can be bad, exemplary or fascinating.
• The statistic must be in the NZ media during the period of May 23 – 29 2015 inclusive.
• Quote the statistic, when and where it was published and tell us why it should be our Stat of the Week.

Next Monday at midday we’ll announce the winner of this week’s Stat of the Week competition, and start a new one.

(more…)

• Stat of the weekend:  1,201,607 to 734,300.

• Flickr has an automatic photo-tagging algorithm. Users don’t like it, because tagging is their business. And also because it “labelled images of black people with tags such as “ape” and “animal” as well as tagging pictures of concentration camps with “sport” or “jungle gym.”

• “Analysis of samples from 55- to 73-year-olds found more than 100 DNA mutations linked to cancer in every 1 sq cm (0.1 sq in) of skin.“ That’s not telling us anything about risk, since we already know pretty accurately what the risk is. But if other cancers are at all similar, it’s a complication for drugs that target the cancer-causing mutations: the friendly-fire damage could be a real problem.

• A basic medical-reporting rule is that you don’t say “cure” in the headline if it isn’t a cure. The recent study of mutations in prostate cancer, reported in the Herald (from the Telegraph), is headlined “Breakthrough offers hope for prostate cancer cure“. About two-thirds of the tumours “had mutations in a molecule that interacts with the male hormone androgen which can already be targeted by current drugs.”  That’s true, but it’s not only “can already be targeted”, but also “is already targeted as standard practice“. Many of other commonly-found mutations are in genes like in BRCA1 and BRCA2, where we don’t have anything like curative treatment. In the medium term, yes, this could be very useful, but the headline is over the top.