Posts filed under Risk (222)

At the Zadar Open chess competition last month, a player who had previously been at the low end of the chess master range did extraordinarily well, playing the level of the world’s very best. Or at the level of a good computer program. There was absolutely no physical evidence to suggest that he had been cheating, but his level of improvement, and the agreement between his moves and those produced by top computer programs are striking.  On the other hand, if you are going to allow accusations in the absence of any corroborating physical evidence, it’s also essentially impossible for an innocent person to mount a defense.

KW Regan, who is a computer scientist and chess master has analysed historical chess competition data, looking at agreement between actual moves and those the computer would recommend, and he claims the Zadar Open results should happen less often than once in a million matches. In his letter to the Association of Chess Professionals, he raises the questions

1.What procedures should be instituted for carrying out statistical tests for cheating with computers at chess and for disseminating their results? Under whose jurisdiction should they be maintained?
2. How should the results of such tests be valued? Under what conditions can they be regarded as primary evidence? What standards should there be for informing different stages of both investigative and judicial processes?

There’s a New York Times story, and Prof Regan also has a blog post. (via)

This weekend, Christchurch had its biggest aftershock for six months.  The moon was substantially further from the earth than average.

Stuff has a good story about a research study looking at smartphone apps for diagnosing melanoma.  These turn out not to be very accurate: they miss quite a lot of melanomas. The story doesn’t mention the false positives, but they are just as bad. Three of the four apps reported  more than 60% of non-melanomas as of concern, so you might as well just cut out the middleman and get checked properly from the start.

Nitpicking: the study was done at the University of Pittsburgh, which is in Pittsburgh, not in Chicago as Stuff seems to think.  Also, the name of the journal isn’t “Online First”, it’s JAMA Dermatology.

The Herald’s story  (via Medical Daily) about a report in the BMJ slightly misses the point. The lead says

A new report, published in the British Medical Journal, claims activities like having a couple of drinks, smoking, eating red meat and sitting in front of the TV can cut at least 30 minutes off a person’s life for every day that do it.

None of this is new. What’s new, and the point of the report, is the idea of quoting all these risks in terms of expected life lost, denominated in ‘microlives‘.   David Spiegelhalter, who is Professor for the Public Understanding of Risk, writes in the report:

We are bombarded by advice about the benefit and harms of our behaviours, but how do we decide what is important? I suggest a simple way of communicating the impact of a lifestyle or environmental risk factor, based on the associated daily pro rata effect on expected length of life. A daily loss or gain of 30 minutes can be termed a microlife, because 1 000 000 half hours (57 years) roughly corresponds to a lifetime of adult exposure. From recent epidemiological studies of long term habits the loss of a microlife can be associated, for example, with smoking two cigarettes, taking two extra alcoholic drinks, eating a portion of red meat, being 5 kg overweight, or watching two hours of television a day. Gains are associated with taking a statin daily (1 microlife), taking just one alcoholic drink a day (1 microlife), 20 minutes of moderate exercise daily (2 microlives), and a diet including fresh fruit and vegetables daily (4 microlives). Demographic associations can also be expressed in these units—for example, being female rather than male (4 microlives a day), being Swedish rather than Russian (21 a day for men) and living in 2010 rather than 1910 (15 a day). This form of communication allows a general, non-academic audience to make rough but fair comparisons between the sizes of chronic risks, and is based on a metaphor of “speed of ageing,” which has been effective in encouraging cessation of smoking.

There was a BBC documentary on this subject back in October (a 5-minute clip is available).

The Herald warns us

More than four in ten UK parents say that their children have been exposed to internet porn, an official survey reveals.

In the fifth paragraph we find

The Daily Mail is campaigning for an automatic block on online porn to protect children

which should make any reader sceptical about the numbers.  This is a scare story from a foreign source notorious for its creative use of numbers and its obsession with sex, and a story where they actually admit they are lobbying. It’s bad enough getting science stories from the Daily Mail; this sort of thing really suggests a news shortage.

We don’t get told what actual questions were asked, how the sample was gathered, or any of the other basic survey details.  More importantly, we don’t even get told anything about the age range of the children, which makes a big difference in this case.  We do learn

Almost a third say their sons or daughters have received sexually explicit emails or texts and a quarter say they have been bullied online or on their phones.

Neither of these issues would be affected by the proposed internet filtering, and both are very different from internet porn in that they are almost exclusively between kids who know each other in real life.

Perhaps the journalists are too young to know that porn existed before the Internet, and teenagers were occasionally exposed to it. You can get a more useful perspective from danah boyd and from a report from Harvard’s cyberlaw clinic that contains actual research.

I’ve been away or busy for a couple of weeks, so here are some collected links on statistics, graphics, the media, and risk

• Wired has a list of top scientific graphics for 2012; Andrew Gelman doesn’t like some of them (more)
• Mathematics in elevator (you know, lift) scheduling, in the Wall Street Journal
• Neuroscience Fiction, in the New Yorker (what brain waves do and don’t tell us)
• Why iPad-native journalism doesn’t work
• A bloke called Tim Gowers writes about how he decided to have an operation, and about the difficulty in getting an intuitive feeling for risks.  There’s nothing particularly unusual about the story, except that it shows risk is difficult even for really smart people

Stuff has a story about an increase in sleeping-pill prescriptions in young people.

The increase in prescriptions is real. What’s more dubious is the explanation that it reflects an increase in sleeping difficulties is being caused by electronic devices, rather than trends in treatment.  It’s not that it’s implausible for gadgets to affect sleep — the mechanisms are fairly clear — it’s more that there isn’t any evidence supplied that sleeping problems are becoming hugely more common.

With the help of the Google and PubMed, I found a few papers looking at time trends in sleep. A recent US paper looked at time-use studies from 1975 to 2006, and found that

Unadjusted percentages of short sleepers ranged from 7.6% in 1975 to 9.3% in 2006.

A Finnish study  found about a 4% decrease in average sleep duration from 1972 to 2005, about half a minute per year.

Other research in both kids and adults seems to agree that sleep duration is decreasing slowly, but not by anything like enough to justify Stuff’s lead:

Your tablet computer, smartphone or other mobile device could be the reason you are not sleeping – and the ubiquitous devices are being cited as a possible cause for a 50 per cent jump in the number of young people scoffing sleeping pills.

It doesn’t make matters better that the “50 per cent jump” is really just for one region in NZ (the Waikato). Or that taking a single 165mg tablet per night is described as “scoffing sleeping pills”.

With the summer holiday season approaching we’ve had requests for a post on the relative safety of driving and flying.

To a large extent this depends on where you are going: if you’re heading from Auckland to the Coromandel then I’d recommend driving, but if you want to spend some time on a beach in the Cook Islands your chances of getting there safely by car are distressingly low.

Clearly we need to rephrase the question.  Two possibilities are:

• for a destination where either flying or driving makes sense, which one is safer?
• if you compare a typical holiday road-trip to a typical holiday flight, which is safer?

We should also think about what risks to include: for a long plane flight the chance of a pulmonary embolism is higher than a crash, possibly much higher depending on your other risk factors.

The risk of a `fatal incident’ on a flight is largely independent of the length of the flight, and based on US data is about eight deaths per hundred million flights.  The risk is probably lower in NZ, since the figure includes the September 11 terrorist attacks.

The risk of death from car crash when driving in the US is about 4 per billion kilometers.  I don’t have good figures for NZ, but it’s a bit higher here. On the other hand, there’s a lot of variation depending on how you drive.

So, for a trip of 500km (eg, Auckland-Wellington), we’re looking at an average figure of about eight deaths in crashes per hundred million flights and about 200 deaths in crashes per hundred million car trips. Flying wins by a huge margin

University of Otago research estimates the risk of pulmonary embolism at about 0.5 per million short flights and about 1.3 per million long flights.  Estimates of the risk of death with pulmonary embolism in modern times seem to be around 10-20%, giving death rates of about 50-100 5-10 per hundred million short flights or 120-250 12-25 per hundred million long flights.  Flying still wins for the Auckland-Wellington route, even if driving doesn’t increase pulmonary embolism risk at all (it probably increases it but by less than driving)

If you compare a 500km drive with a long-haul intercontinental flight the numbers get less clear.  Flying to London could possibly be more dangerous than driving to Wellington, especially if you are a safe driver but at relatively high risk of blood clots.

After all these calculations it’s important to keep a sense of perspective. Driving is pretty safe. Flying is even safer.

According to the Herald, a West Island eye expert says that ‘up to’  5% of people who watched the solar eclipse will have permanent eye damage in the form of a blind spot or black spot in the center of their vision.  That could easily be hundred thousand people in New Zealand, which seems (a) excessive and (b) rather light on supporting data for such an important public health claim.

Auckland eye doctor Sarah Welsh is quoted as being a bit more realistic

… anyone who watched the event with the naked eye could have damaged their retina.

She had seen at least one patient today who believed they damaged their eyes yesterday….

Yet Welsh said it was “unlikely” five per cent of people suffered such burns.

“I’m not sure where he got that number from,” she said.

A brief session with the internets reveals that after a 1999 eclipse in Britain, there were 14 confirmed cases of permanent eye damage. The same eclipse was also seen in Stockolm, Sweden, where there were 15 cases recorded.  And in 1995, an eclipse in Pakistan led to 36 cases at the Abbottabad Hospital, 26 of whom recovered completely.  There will be some under-reporting in all these examples, but it’s hard to imagine that only one in a hundred or one in a thousand of the people with eye damage reports it.

So where did the 5% number come from? It probably sounded plausible.  That is, he pulled it out of his hat. Or somewhere else round and inappropriate.

Overselling the significance of medical research is a familiar problem in the media, with scientists and journalists both being culpable.  There’s a dramatic example going on at the moment, where Pluristem Therapeutics, who made available their in-development treatment on compassionate grounds to three patients, issued press releases with titles such as “Compassionate Use of Pluristem’s PLX Cells Saves the Life of a Child After Bone Marrow Transplantation Failure.” The child died six months later.  One of the other two patients has also died.

Bloomberg News is the main source for this story, and they focused on the impact on share prices and whether the reporting might have violated stock-market regulations, which an expert quoted in the story says is a grey area, and the company denies:

The Haifa, Israel-based company doesn’t follow patients after they are released from the hospital and wasn’t obligated to report the girl’s death, he said.

“What counts legally is whether there is an improvement in the physical condition,” Aberman said in a telephone interview with Bloomberg News. “When we saw significant improvement in the blood count, we declared a successful treatment.”

Even assuming that excitement with the progress of their new treatment was the only factor in the press releases, the whole miracle-cure thing is irresponsible.  Leigh Turner, a bioethicist from Minnesota, (in a blog interview) said it well:

There are some valuable lessons here both for reporters as well as for consumers of news coverage of stem cell research and other areas of science that are routinely overhyped. Whose interests are served when press releases are issued? Is there something significant about timing of press releases? Are independent voices included in news coverage or do all quoted individuals have vested interests in promoting positive account of study results? Why is a press release being issued if results are based upon a single research participant or merely a few research subjects? And finally, what financial interests are swirling beneath the language of “miracles”, “cures”, and “lifesaving” interventions? These are questions reporters should ask and they are also questions those of us who read and otherwise consume the news should consider whenever we are confronted with dramatic claims floated on anecdotes, testimonials, and research “findings” based on meaningless sample sizes.

Pluristem’s work, with placental stem cells, is promising, and they have actual clinical trials starting and planned. That’s how we will be able to tell whether the treatments help patients, and that is what might be worth reporting. (via)