Stats Chat

This graph, from Public Radio International, may help illustrate why public-health people have such a different attitude to business and economics people on cigarette plain packaging

Economists worry about whether plain packaging is effective in reducing smoking, and whether the costs it imposes in New Zealand are out of proportion to the benefits. Other businesses may worry that the idea will spread.

Public health advocates are genuinely mystified by this. They regard tobacco as unique, and tobacco companies as among the common enemies of humankind, to be wiped out as expeditiously as is possible without too much collateral damage.

• For journalists, an excerpt on statistics from The Science Writers Handbook, via a post at Scientific American
•  For everyone, a new book called Naked Statistics, with an interview with the author on NPR

While distracted by a conference in late January, I missed the next exciting installment of the Edinburgh pomegranate saga.

As you will recall, a research group in Edinburgh have put out press releases in recent years about the impact of pomegranate juice on blood pressure, cortisol (a stress hormone), and testosterone.  They haven’t published any scientific papers about these findings, though they have produced a presentation at a scientific conference.

The most recent installment claims that pomegranate extract reduces hunger and food consumption. This study seems to be better designed than the previous ones: participants were randomised to pomegranate extract tablets or placebo for three weeks.  They were then given a glass of pomegranate juice and a meal.  Those who had been taking the pomegranate extract reported feeling less hungry and ate less — 22% less. It’s a pity the study didn’t measure weight, because if the 22% reduction in food consumption generalised beyond the one experimental meal it would have led to measurable weight loss over three weeks.

Again, this was a premature and unsubstantiated press release, and the experiment has not been published in a peer-reviewed journal, although the researchers do at least say they will be presenting it at a conference later in the year.

Two small gems from the Twitter feed

• It’s not you, it’s Powerpoint (via @adzebill)
• Conway’s Life cellular automaton running in your browser (proving that computers are a lot faster than they used to be)

Congratulations to Tony Cooper for his nomination for Stat of the Week:

Statistic: New Zealand has been placed 26th out of 68 countries in a global ranking of life expectancy in retirement.

The Bloomberg data is wrong. They have incorrectly calculated “average number of years residents of 68 countries around the world can look forward to after they stop working.”

It’s a Bloomberg story but the data is being reproduced in NZ for Kiwisaver investors. For example http://media.gmi.co.nz/mailouts/morganOnline/2013/MorganOnlineApr2013.htm.

What Bloomberg have done is subtract the average age of retirement from the life expectancy at birth to calculate “sunset years.” For example, for NZ males this is 78.8 – 65 = 13.8.

But according to NZ Stats Dept figures released last week the life expectancy for NZ males at 65 is 18.8 – a big difference.

The error that Bloomberg made is that you can’t calculate life expectancy at age 65 from the life expectancy at birth. 65 year-olds have a better life expectancy than newborns since they have survived 65 years whereas newborns have a chance of dying before 65.

A long piece in the Washington Post: by Ezra Klein, recommended by Atul Gawande

Brenner puts it more vividly. “There is a bias in medicine against talking to people and for cutting, scanning and chopping into them. If this was a pill or or a machine with these results it would be front-page news in the Wall Street Journal. If we could get these results for your grandmother, you’d say, ‘Of course I want that.’ But then you’d say, what are the risks? Does she need to have chemotherapy? Does she need to be put in a scanner? Is it a surgery? And you’d say, no, you just have to have a nurse come visit her every week.”

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 3 2013.
• Statistics can be bad, exemplary or fascinating.
• The statistic must be in the NZ media during the period of April 27 – May 3 2013 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…)

We heard today that the Christchurch rebuild will be more expensive, by sagans of dollars, than previously computed.

3News attempted to provide some context for the cost of the rebuild, saying it was about the “same as the GDP of Kuwait or Vietnam.“(1:20)  The problem is that most of us have less understanding of the GDP of Kuwait or Vietnam than of $40 billion, so this isn’t terribly helpful.

In fact, the lack of understanding appears to extend  to 3News, since in fact the GDP of Kuwait and Vietnam are about US$175 billion and US$125 billion, not NZ$40 billion.  The government budget (not GDP) in Vietnam is roughly NZ$40 billon, but I don’t know where the Kuwait figure comes from.

• BBC news report on the Amanda Knox case and misuse of statistics: judge did not believe that repeating a test could increase its accuracy (via Mark Wilson)
• A tool for finding stories based on light makeover of press releases. Unfortunately, it often doesn’t detect recycling of stories from sites such as Medical Daily and Science Daily, which do enough rewriting to mask the sources.
• New York Times opinion piece on evidence and science reporting — yes, distinguish experimental and observational studies, but also distinguish small exploratory studies from larger confirmatory ones. (via @brettkeller)
• Using anecdotes rather than data to convince patients. On one hand, speaking to people in language they understand is good; on the other hand, you can use anecdotes to support anything.

From the Stephen Wolfram blog, lots of analysis of Facebook friend data with well-designed graphs.  For example, this graph shows how the mean age of your `friends’ is related to your age.

Those under 40 have Facebook friends of about the same age, but after than the age distribution levels off and becomes much more variable.