My heart soared like a hawk this week after several readers sent in fascinating stories. Here are a few of them.
- Reader Sara C sent in a wonderful example of how easy it is to cheat with statistics. A Scottish city instituted a program to talk to “young people about sex in terms of relationships, not only mechanics.” They claimed that this caused a drop in the teenage pregnancy rates and presented graph to prove it. A brilliant instance of how choosing your end points selectively can flip a negative finding into a positive one. From the BBC: The bumps in a falling teenage pregnancy rate.
- Blogger and nuclear power expert Randy Brich sent in an interview he conducted with “Dr. Antone (Tony) Brooks, a grandfatherly figure who recently retired as the Technical Research Director of the US Department of Energy’s Low Dose Radiation Research Program.”
As a boy, Brooks lived downwind from a nuke test site and was once dusted. He survived and became interested in the health effects of low dose radiation, leading him to a PhD in radiation biology.
Exposure to low doses of radiation have been demonstrated to lower the number of free radicals in the body, protect the cells against transformation (changes that produce cancer), kill pre-cancer cells through a process of programmed cell death (apoptosis), activate immune responses and to lengthen the time between radiation exposure and the induction of cancer. It has also been established that exposure to the same amount of radiation (dose) over a short period of time is much more effective in producing biological changes than exposure of the same amount of radiation over a longer period of time (dose-rate).
Brooks rightly emphasizes that “It is important to recognize the difference between scientific data and radiation protection policy.” Learning how to effectively communicate subtle differences in measurement and research will be crucial if Mr Obama keeps his promise about increasing the use of nuclear power. The public is scared to death of radiation but also knows nothing about it.
Pay attention to the graphic Birch linked (it’s from the government). It’s confusing in spots (“Epidemiology?”) but worth looking at.
- Stalwart reader Katie sent in more evidence England has lost its mind. Ebay refused a listing of a Dad’s Army board game. Why? It had swastikas on the box. Dad’s Army was a harmless and silly Brit TV show about a crew of creaking soldiers who guarded the home front during WW II (Yes, I have seen it: somebody gave me the first season as a birthday present.) Ebay banned the sale because they claimed the game was “memorabilia associated with the Nazi Party” which is verboten!
- Reader John Emery found an utterly enthralling, peer-reviewed, IRB-approved research paper that claims “Democrats and Republicans Can Be Differentiated from Their Faces.” Although this is obviously another example of modern-day phrenology (the other is fMRI studies), it is impossible to dispute the conclusion.
Which is “Republicans were perceived as more powerful than Democrats.” Another obvious finding: They were also rated as more mature (somehow the copy editors misplaced the words “manly” and “virile”). Democrats were rated as more likable and trusty. All of which can be summed up with the question: Who wouldn’t want a pal rather than a father?
The study is so goofy that I might use it as an example of how bad statistics is easily published.
- This one has been making the rounds. “Climate chief was told of false glacier claims before Copenhagen.” Raise your hands if you’re shocked that yet another UN official turned out to be a bit of a fibber? Scandal is so endemic to that organization that it must be the result of planning. Perhaps incoming officials must take some kind of Oath of Graft, or be able to positively demonstrate to have at least one relative that lives in New Jersey.
Update I have corrected the spelling of Mr Brich’s last name. I beg his pardon.
Has any progress been made yet on the great Sir Terry Pratchett’s theory of retro-phrenology in which a dwarf fits a template on your head and taps various spots with a hammer in order to change your character?
“What’ll it be today Mr President? Perhaps a touch less arrogance?”
“A brilliant instance of how choosing your end points selectively can flip a negative finding into a positive one.”
I thought everybody knew that trick. It’s easy to demonstrate that if you solve the normal equations (least squares fit) to calculate the slope, the data points are weighted by their distance from the center. Just pick the end points with a few outliers in the proper direction and you have the trend you want.
Kevin B.,
The Soprano adaptation of Pratchett involves the substitution of a baseball bat. I hear it can be quite effective in correcting certain attitudes.
Regarding the interview with Dr. Antone (Tony) Brooks, I remember reading something a few years back about a problem that the US Navy was having with their nuclear subs. The crews were showing a statistically higher incidence of cancer than the general population. After much study the Navy determined that their crews were not receiving enough low dose radiation. So they reduced some shielding and, like magic, the incidence of cancer amongst the nuclear submarine crews went back to the general population averages.
Modern-Day Phrenology–
Anyone who categorizes themselves as republican has a face different from a democrat… Ridiculous premise. And then what the hell are they doing among the different studies. Yuck. Yet they achieve significance. It goes to show how powerful statistics can be. I wish I was a participant. I would had been discarded as an outlier: ALL politicians look the same to me and they ALL have different faces–subtract Palin, she only gives me Warmth ;o)
Actually one the finest pieces of genuine statistical analysis was done in the UK about whether living close to a nuclear power station led to a rise in the incidence of leukemia in those residents.
Now leukemia is an odd disease which is not well understood in terms to of it’s spread and transmission, but it is known to appear and disappear in clusters in local populations from time to time: and it is suspected that excess radiation exposure can induce it particularly in children.
This was what was alleged was occurring near nuclear power plants.
What was done was to consider all the sites in the UK where the building a nuclear plant had been proposed and divide them into two groups, which conveniently turned out to almost equal numbers, those where a reactor had been built and those where it hadn’t.
The result of this elegant notion showed that there was no difference between incidence of leukemia between the two groups, so the clusters were not due to the nuclear reactors.
But it also showed that the incidence of these clusters was slightly higher than in the country as a whole: nobody knows why although there has been some speculation about it.
By the way
I know something about nuclear boats and was surprised at the comment above because I had never heard of any such suggestion, nor had any of my ex RN colleagues. So I also checked with a number of retired USN counterparts whom I know and they had never heard of it either.
I strongly suspect somebody was extracting the Michael over that one.
Kindest Regards
fMRI studies indeed. I swear I read new nonsense every week in the mainstream press coming from fMRI studies. Interesting technology badly abused.
What’s the problem with the radiation story? Basically, the interviewee is saying that at “low doses” the adverse effects are much lower consistent with a non-linear trend response, and, that some positive benefits are also noted. This is very old news, in general (e.g. http://discovermagazine.com/2002/dec/featradiation ), and the article/interview does not go into details of any significance (undoubtedly due to the fact that there’s not a whole not of objective measureable data available, yet). The observation of positive health benefits on various life forms exposed to low doses of toxins, including radiation, has been noted for a very very long time — its called hormesis — which fell out of favor early on due to its abuse via homeopaths.
The item about party affiliation perceived by facial characteristics has no business in this blog item — that’s an article about a particular area of psychology, not statistics. There’s a lot to it and comparable studies have been done with photos and the ability of executives to recognize other, real, senior executives distinct from actors groomed to look like part of that club. Also, there’s recent studies in which, solely using silent video footage of business meetings, the winning teams/persons could be accurately (well over 80 percent) predicted based solely on non-verbal mannerisms occurring in the first few minutes (as I recall this was published in just the last year or two & the person responsible is making a brisk business from it). Basically these sorts of studies are just showing in the mainstream what a few government intelligence agencies refined over 30 years ago as part of their recruitment & counter-intel toolbag (a former Soviet military intelligence officer reported how their psychologists would be consulted to review silent films taken of prospects to develop profiles in one of his books written in the 70s or 80s). As it turns out, people with sterotypically “geaky” temperments (e.g. engineers, etc. who are typically found to be the antithesis of “people persons”), in general, have extremely poor capacity to appreciate/see/observe the nuances that are readily apparent to others. Partly for this reason, as it turns out, many of these are particularly ripe targets for intelligence recruitment (usually without their overt knowledge), including modern competitive intelligence gathering.
Thanks for a hugely interesting selection of articles. Your recent ones included of course.