Congratulations to Brendon Blue for his nomination of a the “10,000,000,000,000,000 times likely” statistic in last week’s Stat of the Week competition.
For a discussion on this, please refer to James Curran’s post: “10 quadrillion times more likely to have done it”.
Congratulations Brendon! Please add your nominations to this week’s competition.
Thank you for the fantastic nominations in last week’s competition.
We’re awarding the prize to Lindsay Mitchell, for her nomination of Stuff’s mixup in the reporting of a statistic:
25 in every 1000 ex-nuptial births occur to women aged between 15 and 19.
Because it is untrue. The claim is a misrepresentation of the following statistic:
25 in 1000 unmarried females aged 15-19 have an ex-nuptial birth (2011)
Sometimes rewording a statistic ends up with a totally different statistic. For example, the proportion of men who are left handed is not the same as the proportion of left handers who are men.
Congratulations again to Nick Iverson for his nomination on newspaper headlines about cancer, which is worth a post all by itself. It’s no wonder people are feeling confused about medical advice when most days there’s new headlines with isolated findings:
“Oily fish twice a week is best breast cancer defence.”
Someone should have told Angela Jolie. She thought the best defence was total removal.
Anyway, the Herald previously told us “Biggest breast cancer risks – obesity, alcohol and lack of exercise.”
So make up your mind, Herald, is the biggest risk lack of fish or obesity, alcohol and lack of exercise?
By the way, salted fish is another Group 1 carcinogen. So presumably your oily fish mustn’t be salted. And it must be oily not fatty because “Fatty food linked to breast cancer”.
Other big breast cancer risks: “HRT doubles danger of breast cancer” and your risk is also doubled by NOT having a teen pregnancy.
Also being a professional almost doubles your risk “Professional women more susceptible to breast cancer”, and living in NZ doubles your risk “Breast cancer: Geographic links analysed”.
Not finished yet.
Air fresheners double your risk “Breast cancer link to air fresheners”; being left handed doubles your risk “Left-handed women’s risk of breast cancer higher”.
Taking vitamin pills adds 20% to your risk “Major research finds link between multi-vitamin pills and breast cancer”. Whew! That’s a relief. For a minute there I thought it was going to double my risk.
Congratulations to Nick Iversen for his nomination of this stat of the week from the NZ Herald:
This story states that “Public opinion has turned …” To justify this they quote statistics from surveys done in 1012 and 2103.
But the surveys are not comparable. They asked different questions.
The 2012 survey questions and results:
I disapprove 40.3%
I approve 19.6%
I conditionally approve 37.7%The 2013 survey questions and results:
I disapprove 61.5%
I approve 33.8%When the condition (that the number of pokie machines drops across the city) was removed from the 2013 survey those respondents had to switch to a new answer.
My guess is that most would have switched to “I disapprove” because their approval was conditional and the condition has now been removed. If so then the 2012 survey would have had results:
I disapprove 78.0%
I approve 19.6%In that case public opinion has moved in the opposite direction to that in the headline.
Even if only half would have switched the figures would have been
I disapprove 59.2%
I approve 38.5%and it still looks like the opposite to the headline.
The headline is pure fiction. The numbers don’t support it. The two surveys can’t be compared.
For more, see Thomas’ post on this.
Congratulations to Simon Connell for his nomination of this stat of the week from the Otago Daily Times:
“In a three-month trial in Manukau [of an automatic number plate recognition unit], police recovered 15 stolen vehicles, took 180 disqualified, forbidden or suspended drivers off the road, and recovered stolen property from various offenders, [Southern District acting road policing manager Senior Sergeant Steve Larking] said.
”So the benefits are clear.””
Simon makes an excellent point about evidence:
“I’m not sure if this is strictly a statistic or not, but this story (which reads like a copy/paste from a press release) seems to accept without question the benefits of the automatic number plate recognition unit.
We can’t actually make any assessment of the benefits of the unit without some information about how many stolen cards would have been recovered etc. without the unit installed. It does seem like the unit is useful, but it’s hard to judge whether or not it’s worth the $40-$50k price tag without further information.”
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.
Congratulations to Steve Black for his highly topical nomination for Stat of the Week:
Statistic: A bogus txt in poll on Campbell Life shows 22% support for the same sex marriage bill, 78% opposed.
Source: TV3 Campbell Live
Date: 17 April 2013A bogus txt in poll on Campbell Life shows 22% support for the same sex marriage bill, 78% opposed. Nice illustration of just how far off bogus polls can be from general population results done by proper sampling techniques. Ironically, it came on the same day as the legislation passed 77 to 44. I can’t find a trace of it left on the TV3 web page, but I’m still collecting references. Note that the conservative blog I referenced considers it “The best indication yet, short of a referendum, of what the public actually think about same-sex marriage.” thus compounding the mistake of bogus polls taken as meaningful.
This bogus poll would be a perfect example for our first year introductory statistics course on non-sampling errors causing biased results.
Special mention also to Nick’s nomination of Statistics New Zealand’s latest New Zealand Period Life Tables.
Congratulations to Nick Iversen for his humorous nomination of “Sex basically doubles your life span” winning this week’s Stat of the Week competition:
“The anti-ageing benefits [of frequent sexual activity] are amazing. It basically doubles your life span once you get to your 60s and 70s.”
The article obviously get correlation confused with causation but that’s not why I’m posting it here.
The reason for the post is that the statement about life span is so obviously blatantly wrong that I don’t see how any intelligent person could make it.
The average life span of people in their 60s and 70s is (funnily enough) at least 60 or 70. Doubling this means that sex extends your lifespan to 120 or 140.
OK so she really means “remaining lifespan.” For those in their 60s remaining lifespan is about 20 to 25 years. So frequent sex doubles that to 40 to 50 years.
Just doesn’t sound at all reasonable to me. If sex makes that much difference to life expectancy it far outweighs any contributions due to diet, exercise etc and would have been noticed centuries ago.
Congratulations to David Farrar for his nomination last week in our Stat of the Week competition.
He nominated this table on the NZ Herald (click to enlarge the screenshot):
We liked that the Herald were quick to fix the mistake after being alerted of it, but it would have been helpful to also include a note that it had been updated since published.
Thomas Lumley commented that while the totals were just a mistake, not a premeditated misuse, the whole table was a bad idea.
Thank you for all the fantastic Stat of the Summer nominations.
Congratulations to Jonathan Goodman for winning a copy of Tufte’s book “Beautiful Evidence” whose nomination explained and neatly summarised the NZ Herald article “Women with more sex partners turn to drink and drugs”:
“The title and the first paragraphs make the statements that the more promiscuous you have been in the past, the more likely you are now to be dependent on drugs and alcohol. That being promiscuous causes dependency.
“Women who averaged more than 2.5 sexual partners a year in the years leading up to each interview with the researchers were 10 times as likely as women who had only one or no sexual partners a year to be clinically dependent on alcohol or drugs at age 21. They were seven times as likely by age 26 and 17 times as likely by age 32, even after allowing for all other factors in their lives.”
The increase in likelihood of dependence as the number of years of promiscuity increases is seen as evidence that promiscuity causes alcohol and drug dependence.
However the article then paraphrases the researches by saying “The researchers said the link between the number of partners and later substance dependence might be due to the “shared context” of drinking and meeting people in bars, or to both behaviours being related to underlying risk-taking attitudes.”
While the author of the article claims in the title that researchers have found that promiscuous activities drives people to drink and drugs, all the researchers have found out is that there is a correlation between the two. As we all know Correlation does not imply Causation.”
Thank you to everyone who took part and we are now resuming our weekly competition, so please keep your nominations coming in!