If you want to smoke marijuana, then for all practical purposes you can't serve in the U.S. military, which has a rigorous drug-testing program.
I think the military should add tobacco to the list of drugs that military personnel are forbidden to use. Every military person is supposed to be combat-ready, and the effect of smoking on lung capacity alone would make them significantly less capable.
It's well-accepted that the tobacco industry, through the Tobacco Institute and other front groups, hired medical professionals to do studies to challenge every valid study that demonstrated the harm of cigarettes. They send a lot of their documents outside of the U.S., because they thought it couldn't be subpoenaed there. When the studies found that cigarettes were harmful, they didn't disclose them. When the studies cast doubt on the harm of cigarettes, they promoted them heavily. In science, this is now considered scientific misconduct.
The industry continued this process with second-hand smoke. There's pretty good evidence now that second-hand smoke causes lung disease, particularly in children who grow up in households with smokers. (I'll leave it to somebody else to cite the studies.)
You'll never have perfect evidence. You'll always have scientists who challenge the studies. They're paid to challenge the studies. There are a few scientists who honestly challenge the validity of the second-hand smoke studies without being on the payroll of the industry (now more often anti-regulatory groups), but their number is vanishingly small.
Unfortunately there are no *objective* methods (in your sense of the term) to deal with multi-variate results. You have to examine each study and determine how much of a weight to give it. If it has an obvious mistake, like traffic accident studies that don't correct for weather, then you give it a weight of zero.
It's tempting to think that there must be some information that you can squeeze out of a bad study if you could find the right mathematical method. But in the real world, where you can test your results, every attempt has failed. You can look at medical journals, and in just about every issue read about a treatment that people used, based on poor-quality studies that were analyzed in the way you describe. When they finally did a randomized controlled trial, the treatment failed (and sometimes did more harm than good).
Otherwise, the anti-helmet people would be saying, "Look! In this state, bicycle helmet laws went into effect in September. From October to March, they had 20 bicycle deaths. In the 6-month period before that, they only had 5 deaths. Bicycle helmet laws quadrupled the death rate. Include that in your meta-analysis." How much weight do you give that study?
Even in traffic studies, you have to evaluate the study before you include it in the meta-analysis.
For example, there are lots of quick-and-dirty studies of the effects of various traffic safety interventions. It's good for managers to do quick-and-dirty studies. They're better than nothing. But you have to know their limitations.
Someone may study the effects of seat belt laws by comparing the year before seat belts were required, with the year after, and find out that deaths from traffic accidents went up. However, sometimes it turns out that the year after had worse weather. Bad weather is strongly correlated with traffic deaths.
Do you include that study in the meta-analysis, and give it the same weight as a large study over several years and many geographic locations?
Do you say, "When we add this study, the statistical significance in the meta-analysis falls below.95, so seat belt laws have no effect"?
The Elvik study doesn't suggest that there was bias in the Cochrane study (unless you consider bias against poor-quality studies to be bias).
The Elvik study doesn't "strongly suggest" bias in the Cochrane study. More precisely, you are *speculating* that there is a bias in the Cochrane study. At a minimum, you'd have to read the 4 studies before you could say that it "strongly suggests" bias.
Maybe you could say that Elvik's study "weakly suggests" bias.
We do know which meta-analysis is more representative of reality: the one with the higher-quality studies. Low-quality studies are noise, which can give you a false-negative result.
For many years, drug companies used to do something like this until the medical journals decided not to put up with it any more. They would do poor-quality studies of their drug. If they got a good result, they would publish it, even if it was due to chance, selection bias, a sub-therapeutic dose of their competitor's comparison drug, etc.
Cochrane reviews the studies and throws out the poor-quality studies.
If a study shows a drug to be equivalent or superior to its competitor's drug, but it used a sub-therapeutic dose of the competitor's drug, and you're trying to figure out whether the new drug is equivalent or superior, you don't get results that are more representative of reality if you include that study in your meta-analysis.
Peer review means that experts in some field have reviewed it, and they certify that the manuscript as presented meets scholarly standards. It doesn't mean that it's true, or even that they think it's true.
Journals often publish contrarian articles, which go against the conventional wisdom. Even if you believe in the conventional wisdom, it's a good policy to challenge the conventional wisdom, and force its proponents to prove every step.
However, just because a contrarian publishes an article in a respectable journal, it doesn't follow that the contrarian is right and the conventional wisdom is wrong.
I'm not familiar with Accident Analysis & Prevention, but let's stipulate that it's a respectable journal.
I am familiar with the Cochrane Collaboration. I routinely read their studies, and I've interviewed some of their authors at medical conferences.
Cochrane is an international collaboration set up to provide objective analysis of medical evidence and decisions for doctors. They have a huge volunteer staff including the best epidemiologists and medical statisticians in the world. When Cochrane writes a report, literally every review article in every major medical journal (including BMJ, NEJM, Lancet and JAMA), will include Cochrane's conclusions as the best available evidence.
Medical epidemiology is one of the most difficult jobs in science. Medical journals have statisticians and epidemiologists to review papers. Often the experts can't agree. Often even the best statisticians come to conclusions that turn out to be wrong when they try to confirm it with the empirical data. (See Simpson's paradox on Wikipedia.) You seldom have perfect information to answer a question, so the interpretation is usually subjective. Somebody can always read a statistical analysis and find weaknesses or points of disagreement, (Just read the section that says, "Limitations of this study.")
So we have Rune Elvik, who has 36 studies on PubMed, all but 1 in Accident Analysis and Prevention, who is apparently a political scientist who works for an economics institute. He is re-analyzing many important traffic safety studies, which is a commendable activity.
Then we have the Cochrane Collaboration, which is the world's largest collaboration of medical epidemiologists and statisticians who have published tens of thousands of papers, and whose work is widely accepted in the broad medical and scientific community.
The biggest practical problem with Cochrane is that they restrict themselves to the highest-quality studies, and if there are no high-quality studies, they don't come to a conclusion. So you often find the frustrating conclusion, "The evidence was not adequate for a recommendation." So when they do come to a conclusion, you know it's based on strong evidence.
Cochrane says that, based on the imperfect evidence, the best conclusion is that bicycle helmets save lives.
Elvik says in his abstract that helmets don't save as many lives as Cochrane says. According to the conclusion that you quote, when you include 4 additional new studies, helmets don't save any lives.
If I have to come to a conclusion about whether to wear a helmet, or whether to pass helmet laws, who am I going to believe -- Elvik or Cochrane? One political scientist in Norway, or the world's largest collaboration of epidemiologists who came to a conclusion based on the best available evidence?
The only way I would take Elvik's conclusions seriously would be if I could see him debate the Cochrane authors at a conference, and if a lot of the experts at the conference agreed that he had a good point. And I'd like to see them explain the reasons why they disagree in language that I can understand.
(BTW, I don't want to get into the woods of statistical analysis, but as a general rule, badly designed or underpowered studies are liable to give false negative conclusions. So if Elvik used 4 underpowered studies, his rejection of the Cochrane conclusions would have been unjustified.)
No, I'm talking about using terms like "trick" which was used to refer to a mathematical method, which non-scientists who were trying to discredit global warming used to claim that the East Anglian scientists were trying to deceive somebody.
Were the critics too stupid to understand what a mathematical "trick" is? Or did they actually know and were being deliberately deceptive? You decide.
(1) Scientific data has certain rules for collection and retention. Scientists have had papers retracted by the journals, and been found guilty of fraud, because they couldn't supply data to support their published results. (Back in the days of paper, you could go to the university book store and buy laboratory notebooks with numbered pages.) Having gaps in your records is itself suspicious. If you had a civil trial, the opposing lawyer would ask the scientist, "Isn't it customary in the profession to record this data?" If data is missing, the judge could rule that the jury should interpret it in the most unfavorable light.
(2) Destroying evidence because you know it could be harmful in court is called "despoilation of evidence," and it can be a crime (although Oliver North got away with it).
There were a lot of embarrassing moments in the early days of computerization when missing documents would turn up in the backup tapes. "Retention experts" advised companies to destroy backup tapes after a legally minimum time. But you never can be sure that something is completely gone. Documents get distributed, and somebody may have a copy left.
In a modern organization, it's hard to avoid putting something important into writing. Laywers on one side figure out ways of getting around it, lawyers on the other side figure out ways of catching them. The tobacco companies did get caught, although they unfortunately didn't go to jail.
Actually, GSK does have to share all their correspondence and preliminary analysis when they get sued. That's where we get a lot of the good stuff. Look up the tobacco industry documents online.
In the US, at least, a judge can order anyone -- even someone who isn't a party to the lawsuit -- to disclose any information that's "in the interests of justice."
I was once sitting through a drug patent lawsuit and they had admitted into evidence a guy's entire 4-drawer file cabinet. They digitized every page, put it in a database, and were projecting it onto a screen in the courtroom.
One of the problems with the East Anglia climate change emails was that people who didn't understand (or care about) the science took snippits out of context and used them in misleading and defamatory ways. For example, they seized on the term "trick", and claimed that it meant that he was trying to deceive people, when actually it was referring to a mathematical trick. Those scientists lost about 2 years defending themselves against baseless accusations.
Scientist don't want to spend hundreds of hours fending off phone calls and ambush journalists from Fox News. That's not the open process of science, it's just harassment by people who don't intend to give you a fair hearing in the first place, and don't understand or care about the science.
A lot of times, these people are working for corporations or industries that are trying to attack the science even when they know that the science is right.
A lot of times, as in the lead poisoning cases, these requests can lead to legal depositions, where in addition to hundreds of hours of time, they can spend hundreds of thousands of dollars in legal fees. And they don't get the legal fees back from the other side.
I was really annoyed when I read that NYT story because the reporter didn't talk to any scientists who studied traffic to find out how good the evidence is that helmets do prevent injury and death.
Instead she asked an economist. Duh -- when the science goes against you, find an economist.
It seems intuitively obvious that (1) bicycles have a lot more accidents than most everyday activities, like crossing the street (2) many people have been killed by head injuries in bicycle accidents without helmets, many people have survived similar accidents with helmets, and it seems as if helmets can make a lot of accidents survivable.
That's the intuition. What's the fucking evidence? The reason I read newspapers rather than blogs is that I expect a professional science journalist to look up the evidence. (Wikipedia is no help. They just have a big incoherent argument.)
In the absence of rigorous scientific evidence --
Here's my data point. I knew a doctor in my neighborhood who was riding his bike along the West Side Highway bicycle/pedestrian path, which was laid out in such a way as to not inconvenience drivers who wanted to cut across it. A truck cut across him, sent him flying, and he hit his head against a lamp post. He died. He wasn't wearing a helmet.
(Sure the road design was wrong. Streets are dangerous. But given that, helmets will save your life a lot of times.)
You say personal choice. I say tragic waste of life. Sometimes the results are so tragic the government should tell you what to do.
Somebody asked me the same question when some Con Ed workers were drilling up the street in front of her window, and this was the answer I gave her (although I like the noise-cancelling idea):
I used to work in a factory in the Bronx that made acoustic soundproofing panels for radio stations, music studios, etc.
These panels were about 8 feet wide, 4 feet wide, and 5 inches thick. They were made like a sandwich. The filling was 3/4-inch sheetrock glued to a layer of about 4 inches of fiberglass. On the outside of the filling, they had 3/16-inch aluminum panels. The outside aluminum panels were normal aluminum sheet. The inside aluminum panels were punched with lots of tiny holes to absorb sound better. They were spot-welded together, which was a lot of fun.
They were very effective. The factory was very noisy, with grinders, pneumatic drills and stuff. If you stood the panels up into a little closet on 3 sides of you, the sound would be dramatically reduced. They were pretty heavy -- it took 2 guys to move them onto a truck, and I'd guess they weighed 150 pounds apiece.
You know how the walls of a normal apartment is made of sheetrock (wallboard) panels. To make a recording studio, they would install these acoustic panels instead.
As I write this I am sitting on 43rd St. and 10th Ave. 10th Ave. is a heavily-trafficked main route for fire engines, and drivers love to blow their horns. Con Edison has jackhammers in the street all the time. If I leave my window open (as I like to do), it's pretty noisy. If I close the double-glass panel window, it's a lot quieter (although it's stuffier and I have to turn on the air conditioner).
If the noise were continuous, and it were really bothering me, I would consider making my own acoustic panels. Instead of aluminum and sheetrock, I'd make them out of cardboard and fiberglass. Fiberglass is used for heat insulation, so you can easily find it in Home Depot or local lumber yards. It comes in standard sizes, either in rolls or panels. I'd make the panels 4 feet wide and just high enough to clear my ceiling. The thicker the better -- I'd make them 4 or 6 inches thick.
The best way to make them would be to find cardboard cartons that somebody had thrown out that were just the right size -- say, 4 feet by 7 feet. They ship mattresses in cartons like that. I'd check in the mattress stores, or maybe hotels, that get mattresses delivered in boxes like that and throw them out. Or you can find big sheets of cardboard that you can cut and fold to size. (It's not hard to cut and fold cardboard. I learned how to do that in fourth grade). You can use glue and tape to hold the cardboard together after you fold it. The best heavy-duty cardboard is from the boxes that refrigerators come in. Some apartment buildings regularly replace refrigerators so they'd have a lot of boxes.
So you make for example 4 cardboard boxes, each about 4 feet by 8 feet by 5 inches (inside measurement -- that's important, so the fiberglass will fit). Fold the boxes. Place the 4 foot by 8 foot by 5 inch fiberglass batting into the boxes. Seal the edges of the boxes around the fiberglass with glue and/or packing tape to keep the fiberglass inside and prevent it from spilling out. Put those panels in a line in front of your windows and they should dramatically reduce the sound.
(Fiberglass can be messy, so you have to be careful, maybe working outdoors, wearing old clothes, etc., but people use it all the time. You can get instructions on the Internet.)
One minor problem is that the cardboard reflects sound. You preferably want absorbent material on the side towards you (which is why we used perforated aluminum sheet).
One of the best absorbent materials is egg crates. (This is not styrofoam egg crates, but fluffy cardboard, which has much better acoustics.) People have actually made sound studios using egg crates on the walls and ceiling for sound absorption. You used to be able to get egg crates from big restaurants that used lots of eggs. (These are not t
Tissue-culture bioreactors, as far as I know, usually grow tissue on fetal calf serum, which come from slaughtered cattle. (Sometimes they use mice, fertilized chicken eggs, etc.)
So you have to slaughter even more cattle to create leather in bioreactors.
The vaguer message of Occupy was that the Democratic Party in the US has utterly ignored the liberals in their base in an effort to pander to Wall St and the right wing. And why should people like Obama do that, when all they need to do to get their votes is scare the bejeesus out of them by threatening them with the prospect of President Romney?
I heard Amy Goodman of Democracy Now give a good answer to that, when she introduced Ralph Nader in the 2000 election. The Republican Party has been moving further and further to the right. The Democratic Party has been moving further to the right to match them. On domestic policy, the Democratic Party is further to the right now than Richard Nixon (don't forget, Nixon's secretary of HEW was Pat Moynihan). If we continue to vote for the Democratic Party, they will continue to move to the right until there's no meaningful difference between them. We have to vote for third party candidates to tell the Democrats that they can't take us for granted.
Since that time, Obama gave us a health care plan that was literally written by the Heritage Foundation. His chief of staff, Rahm Emanuel, told progressives that they were "fucking retarded" for wanting the single payer system Obama promised us. He appointed Wall Street financiers to run his White House. They tossed Acorn under the bus, which was one of the best campaign organizing tools they had.
What I don't understand is why the Democrats didn't learn from 2000 that if you tell the left wing of your party to go fuck themselves, you can lose an election. Maybe it's like training a mule -- you have to hit them on the head with a sledgehammer -- again.
I think it's like a strike. You don't want to go on strike, you don't want to lose weeks or months of salary, you don't want to take a chance on having your employer move to China. But if we hadn't gone on strike over the last 100 years, we would be making Chinese wages right now, and if we never go on strike, we will be making Chinese wages.
Somebody tell the Democrats. If you tell us to fuck off one more time, we'll fuck up your election, just like we did in 1968 and in 2000.
They didn't know because the Bloomberg Administration kicked them out of Zucotti Park and kept chasing them away from any other place they tried to set up.
Hell, the Bloomberg Administration wouldn't even let them use microphones. They refused to let them rent portable toilets, and then complained when they used restaurant bathrooms.
How much would the Americans have accomplished if the Continental Congress had to disband after 2 months?
OWS accomplished one good, important thing: They said that the richest 1% of the country had a wildly disproportionate amount of wealth and power, didn't deserve it, and was using it to fuck the rest of us. A lot of people believed that, but that wasn't in the public discussion. OWS put it into the public discussion.
What do we do about it? Well, as Bertold Brecht said, you have to answer that question yourself.
One thing we don't do is make compromises, write a platform, collect $100 million, and try to beat the 2 big parties at their own game.
Another thing we don't do -- I don't think -- is to organize a movement and deliver ourselves to the service of the Democratic party, who will tell us again after the next election, "Thank you very much, now go fuck yourselves as we make some deals with our corporate masters, the Blue Dogs and Republicans."
But you have to think for yourself and make your own decisions. OWS did enough thinking for you. You don't like it? Go start your own organization.
I sympathize with the advertising industry, and the companies like Google that have done so many kewl things with the money they made out of Internet ads.
But the Internet killed off the newspaper and magazine business, where I used to work. At one time, newspapers in every city would hire a lot of reporters to spend a lot of time following the issues and informing their readers what was going on. Now they've been laying off their reporters, they're down to skeleton crews, and some of the great newspapers went out of business. The advertising, much as I hated it, was paying the bills. Now it's decimated. Blogs are nice, but they really don't replace the newspaper model that served pretty well for the last 2 or 3 centuries.
Technology improves, we got what we wished for, the market changes, and there are winners and losers. (There are a lot more losers than most of us expected.) If people delete their cookies every morning, it might make it harder for web advertisers to make money, but we've all got problems. Let the market rule, it will anyway. This isn't tragic enough to give these guys a government favor.
First, I'm not sure that the story is true. A lot of stories get "improved" with the telling. It sounds like the dialog of some porn movies that I prefer to watch with the sound off. There might be a germ of truth to it.
Second, if it is true, that girl is one obnoxious bitch. It's one thing to be flirtatious; it's another thing to sleep with everybody in the office for the joy of playing them off against each other. This is a good way to disrupt the office and get fired. If true.
And I don't see how it's particularly an issue in programming. The (exaggerated) stereotype is that male programmers are socially inept nerds. This sounds like one socially inept woman. If true.
There are 2 issues here: (1) How safe is it to drive at 85mph and (2) How safe is it to have people on a highway ignoring the posted signs and driving at different speeds?
The rational way to set speed limits would be for society to decide what your cost/benefit ratio is -- how many deaths and injuries are you willing to accept in exchange for what level of driving convenience? Then you post the speed limits (and minimums) and stick to it. The best way to enforce laws is to have small penalties and strict enforcement. The worst way is to have severe penalties and arbitrary enforcement.
We have a culture especially in some parts of the country where some people just don't want to obey the speed laws. It's very unpopular to have the police enforce the speed limits effectively, so they just don't do it. People just drive as fast as they want. And the enforcement is arbitrary. Why stop some cars and not others? They've turned traffic safety into a game.
Maybe it's a libertarian culture, maybe it's a macho culture, maybe it's just people who say, "I don't give a fuck what anybody else thinks, I'm going to do whatever I want."
The price you pay is that people get killed and severely injured. There are so many vehicle accidents in the U.S. that almost everyone knows somebody who was killed or severely injured. I had one friend who died, and another who went through the windshield, had most of her face smashed up, lost her front teeth, and spent 6 months in the hospital. Do you think it's worth it in order to drive 75mph? I don't.
It's safer to have everybody driving at the same speed. But other things being equal, when you drive faster, the fatalities go up.
Slashdot Mirror
If you want to smoke marijuana, then for all practical purposes you can't serve in the U.S. military, which has a rigorous drug-testing program.
I think the military should add tobacco to the list of drugs that military personnel are forbidden to use. Every military person is supposed to be combat-ready, and the effect of smoking on lung capacity alone would make them significantly less capable.
It's well-accepted that the tobacco industry, through the Tobacco Institute and other front groups, hired medical professionals to do studies to challenge every valid study that demonstrated the harm of cigarettes. They send a lot of their documents outside of the U.S., because they thought it couldn't be subpoenaed there. When the studies found that cigarettes were harmful, they didn't disclose them. When the studies cast doubt on the harm of cigarettes, they promoted them heavily. In science, this is now considered scientific misconduct.
The industry continued this process with second-hand smoke. There's pretty good evidence now that second-hand smoke causes lung disease, particularly in children who grow up in households with smokers. (I'll leave it to somebody else to cite the studies.)
You'll never have perfect evidence. You'll always have scientists who challenge the studies. They're paid to challenge the studies. There are a few scientists who honestly challenge the validity of the second-hand smoke studies without being on the payroll of the industry (now more often anti-regulatory groups), but their number is vanishingly small.
Unfortunately there are no *objective* methods (in your sense of the term) to deal with multi-variate results. You have to examine each study and determine how much of a weight to give it. If it has an obvious mistake, like traffic accident studies that don't correct for weather, then you give it a weight of zero.
It's tempting to think that there must be some information that you can squeeze out of a bad study if you could find the right mathematical method. But in the real world, where you can test your results, every attempt has failed. You can look at medical journals, and in just about every issue read about a treatment that people used, based on poor-quality studies that were analyzed in the way you describe. When they finally did a randomized controlled trial, the treatment failed (and sometimes did more harm than good).
Otherwise, the anti-helmet people would be saying, "Look! In this state, bicycle helmet laws went into effect in September. From October to March, they had 20 bicycle deaths. In the 6-month period before that, they only had 5 deaths. Bicycle helmet laws quadrupled the death rate. Include that in your meta-analysis." How much weight do you give that study?
Even in traffic studies, you have to evaluate the study before you include it in the meta-analysis.
For example, there are lots of quick-and-dirty studies of the effects of various traffic safety interventions. It's good for managers to do quick-and-dirty studies. They're better than nothing. But you have to know their limitations.
Someone may study the effects of seat belt laws by comparing the year before seat belts were required, with the year after, and find out that deaths from traffic accidents went up. However, sometimes it turns out that the year after had worse weather. Bad weather is strongly correlated with traffic deaths.
Do you include that study in the meta-analysis, and give it the same weight as a large study over several years and many geographic locations?
Do you say, "When we add this study, the statistical significance in the meta-analysis falls below .95, so seat belt laws have no effect"?
The Elvik study doesn't suggest that there was bias in the Cochrane study (unless you consider bias against poor-quality studies to be bias).
The Elvik study doesn't "strongly suggest" bias in the Cochrane study. More precisely, you are *speculating* that there is a bias in the Cochrane study. At a minimum, you'd have to read the 4 studies before you could say that it "strongly suggests" bias.
Maybe you could say that Elvik's study "weakly suggests" bias.
We do know which meta-analysis is more representative of reality: the one with the higher-quality studies. Low-quality studies are noise, which can give you a false-negative result.
For many years, drug companies used to do something like this until the medical journals decided not to put up with it any more. They would do poor-quality studies of their drug. If they got a good result, they would publish it, even if it was due to chance, selection bias, a sub-therapeutic dose of their competitor's comparison drug, etc.
Cochrane reviews the studies and throws out the poor-quality studies.
If a study shows a drug to be equivalent or superior to its competitor's drug, but it used a sub-therapeutic dose of the competitor's drug, and you're trying to figure out whether the new drug is equivalent or superior, you don't get results that are more representative of reality if you include that study in your meta-analysis.
Less is more.
So you should know that RDW is correct.
I agree that the Elvik meta-study is interesting to compare to the Cochrane study. But it doesn't reject the Cochrane study.
Peer review means that experts in some field have reviewed it, and they certify that the manuscript as presented meets scholarly standards. It doesn't mean that it's true, or even that they think it's true.
Journals often publish contrarian articles, which go against the conventional wisdom. Even if you believe in the conventional wisdom, it's a good policy to challenge the conventional wisdom, and force its proponents to prove every step.
However, just because a contrarian publishes an article in a respectable journal, it doesn't follow that the contrarian is right and the conventional wisdom is wrong.
I'm not familiar with Accident Analysis & Prevention, but let's stipulate that it's a respectable journal.
I am familiar with the Cochrane Collaboration. I routinely read their studies, and I've interviewed some of their authors at medical conferences.
Cochrane is an international collaboration set up to provide objective analysis of medical evidence and decisions for doctors. They have a huge volunteer staff including the best epidemiologists and medical statisticians in the world. When Cochrane writes a report, literally every review article in every major medical journal (including BMJ, NEJM, Lancet and JAMA), will include Cochrane's conclusions as the best available evidence.
Medical epidemiology is one of the most difficult jobs in science. Medical journals have statisticians and epidemiologists to review papers. Often the experts can't agree. Often even the best statisticians come to conclusions that turn out to be wrong when they try to confirm it with the empirical data. (See Simpson's paradox on Wikipedia.) You seldom have perfect information to answer a question, so the interpretation is usually subjective. Somebody can always read a statistical analysis and find weaknesses or points of disagreement, (Just read the section that says, "Limitations of this study.")
So we have Rune Elvik, who has 36 studies on PubMed, all but 1 in Accident Analysis and Prevention, who is apparently a political scientist who works for an economics institute. He is re-analyzing many important traffic safety studies, which is a commendable activity.
Then we have the Cochrane Collaboration, which is the world's largest collaboration of medical epidemiologists and statisticians who have published tens of thousands of papers, and whose work is widely accepted in the broad medical and scientific community.
The biggest practical problem with Cochrane is that they restrict themselves to the highest-quality studies, and if there are no high-quality studies, they don't come to a conclusion. So you often find the frustrating conclusion, "The evidence was not adequate for a recommendation." So when they do come to a conclusion, you know it's based on strong evidence.
Cochrane says that, based on the imperfect evidence, the best conclusion is that bicycle helmets save lives.
Elvik says in his abstract that helmets don't save as many lives as Cochrane says. According to the conclusion that you quote, when you include 4 additional new studies, helmets don't save any lives.
If I have to come to a conclusion about whether to wear a helmet, or whether to pass helmet laws, who am I going to believe -- Elvik or Cochrane? One political scientist in Norway, or the world's largest collaboration of epidemiologists who came to a conclusion based on the best available evidence?
The only way I would take Elvik's conclusions seriously would be if I could see him debate the Cochrane authors at a conference, and if a lot of the experts at the conference agreed that he had a good point. And I'd like to see them explain the reasons why they disagree in language that I can understand.
(BTW, I don't want to get into the woods of statistical analysis, but as a general rule, badly designed or underpowered studies are liable to give false negative conclusions. So if Elvik used 4 underpowered studies, his rejection of the Cochrane conclusions would have been unjustified.)
No, I'm talking about using terms like "trick" which was used to refer to a mathematical method, which non-scientists who were trying to discredit global warming used to claim that the East Anglian scientists were trying to deceive somebody.
Were the critics too stupid to understand what a mathematical "trick" is? Or did they actually know and were being deliberately deceptive? You decide.
That's a good point, but
(1) Scientific data has certain rules for collection and retention. Scientists have had papers retracted by the journals, and been found guilty of fraud, because they couldn't supply data to support their published results. (Back in the days of paper, you could go to the university book store and buy laboratory notebooks with numbered pages.) Having gaps in your records is itself suspicious. If you had a civil trial, the opposing lawyer would ask the scientist, "Isn't it customary in the profession to record this data?" If data is missing, the judge could rule that the jury should interpret it in the most unfavorable light.
(2) Destroying evidence because you know it could be harmful in court is called "despoilation of evidence," and it can be a crime (although Oliver North got away with it).
There were a lot of embarrassing moments in the early days of computerization when missing documents would turn up in the backup tapes. "Retention experts" advised companies to destroy backup tapes after a legally minimum time. But you never can be sure that something is completely gone. Documents get distributed, and somebody may have a copy left.
In a modern organization, it's hard to avoid putting something important into writing. Laywers on one side figure out ways of getting around it, lawyers on the other side figure out ways of catching them. The tobacco companies did get caught, although they unfortunately didn't go to jail.
Actually, GSK does have to share all their correspondence and preliminary analysis when they get sued. That's where we get a lot of the good stuff. Look up the tobacco industry documents online.
In the US, at least, a judge can order anyone -- even someone who isn't a party to the lawsuit -- to disclose any information that's "in the interests of justice."
I was once sitting through a drug patent lawsuit and they had admitted into evidence a guy's entire 4-drawer file cabinet. They digitized every page, put it in a database, and were projecting it onto a screen in the courtroom.
Because there are people like James O'Keefe around http://en.wikipedia.org/wiki/ACORN_2009_undercover_videos_controversy who aren't interested in science, and don't even understand the science, but want to use information to damage their opponents in elections.
One of the problems with the East Anglia climate change emails was that people who didn't understand (or care about) the science took snippits out of context and used them in misleading and defamatory ways. For example, they seized on the term "trick", and claimed that it meant that he was trying to deceive people, when actually it was referring to a mathematical trick. Those scientists lost about 2 years defending themselves against baseless accusations.
Scientist don't want to spend hundreds of hours fending off phone calls and ambush journalists from Fox News. That's not the open process of science, it's just harassment by people who don't intend to give you a fair hearing in the first place, and don't understand or care about the science.
A lot of times, these people are working for corporations or industries that are trying to attack the science even when they know that the science is right.
A lot of times, as in the lead poisoning cases, these requests can lead to legal depositions, where in addition to hundreds of hours of time, they can spend hundreds of thousands of dollars in legal fees. And they don't get the legal fees back from the other side.
The abstract doesn't say that there is no significant mitigation. It says that the mitigation is smaller than the last meta-analysis.
And this study seems to have been done by an economist. Sometimes economists come up with results that go against the conclusions of scientists.
I was really annoyed when I read that NYT story because the reporter didn't talk to any scientists who studied traffic to find out how good the evidence is that helmets do prevent injury and death.
Instead she asked an economist. Duh -- when the science goes against you, find an economist.
It seems intuitively obvious that (1) bicycles have a lot more accidents than most everyday activities, like crossing the street (2) many people have been killed by head injuries in bicycle accidents without helmets, many people have survived similar accidents with helmets, and it seems as if helmets can make a lot of accidents survivable.
That's the intuition. What's the fucking evidence? The reason I read newspapers rather than blogs is that I expect a professional science journalist to look up the evidence. (Wikipedia is no help. They just have a big incoherent argument.)
In the absence of rigorous scientific evidence --
Here's my data point. I knew a doctor in my neighborhood who was riding his bike along the West Side Highway bicycle/pedestrian path, which was laid out in such a way as to not inconvenience drivers who wanted to cut across it. A truck cut across him, sent him flying, and he hit his head against a lamp post. He died. He wasn't wearing a helmet.
(Sure the road design was wrong. Streets are dangerous. But given that, helmets will save your life a lot of times.)
You say personal choice. I say tragic waste of life. Sometimes the results are so tragic the government should tell you what to do.
Somebody asked me the same question when some Con Ed workers were drilling up the street in front of her window, and this was the answer I gave her (although I like the noise-cancelling idea):
I used to work in a factory in the Bronx that made acoustic soundproofing panels for radio stations, music studios, etc.
These panels were about 8 feet wide, 4 feet wide, and 5 inches thick. They were made like a sandwich. The filling was 3/4-inch sheetrock glued to a layer of about 4 inches of fiberglass. On the outside of the filling, they had 3/16-inch aluminum panels. The outside aluminum panels were normal aluminum sheet. The inside aluminum panels were punched with lots of tiny holes to absorb sound better. They were spot-welded together, which was a lot of fun.
They were very effective. The factory was very noisy, with grinders, pneumatic drills and stuff. If you stood the panels up into a little closet on 3 sides of you, the sound would be dramatically reduced. They were pretty heavy -- it took 2 guys to move them onto a truck, and I'd guess they weighed 150 pounds apiece.
You know how the walls of a normal apartment is made of sheetrock (wallboard) panels. To make a recording studio, they would install these acoustic panels instead.
As I write this I am sitting on 43rd St. and 10th Ave. 10th Ave. is a heavily-trafficked main route for fire engines, and drivers love to blow their horns. Con Edison has jackhammers in the street all the time. If I leave my window open (as I like to do), it's pretty noisy. If I close the double-glass panel window, it's a lot quieter (although it's stuffier and I have to turn on the air conditioner).
If the noise were continuous, and it were really bothering me, I would consider making my own acoustic panels. Instead of aluminum and sheetrock, I'd make them out of cardboard and fiberglass. Fiberglass is used for heat insulation, so you can easily find it in Home Depot or local lumber yards. It comes in standard sizes, either in rolls or panels. I'd make the panels 4 feet wide and just high enough to clear my ceiling. The thicker the better -- I'd make them 4 or 6 inches thick.
The best way to make them would be to find cardboard cartons that somebody had thrown out that were just the right size -- say, 4 feet by 7 feet. They ship mattresses in cartons like that. I'd check in the mattress stores, or maybe hotels, that get mattresses delivered in boxes like that and throw them out. Or you can find big sheets of cardboard that you can cut and fold to size. (It's not hard to cut and fold cardboard. I learned how to do that in fourth grade). You can use glue and tape to hold the cardboard together after you fold it. The best heavy-duty cardboard is from the boxes that refrigerators come in. Some apartment buildings regularly replace refrigerators so they'd have a lot of boxes.
So you make for example 4 cardboard boxes, each about 4 feet by 8 feet by 5 inches (inside measurement -- that's important, so the fiberglass will fit). Fold the boxes. Place the 4 foot by 8 foot by 5 inch fiberglass batting into the boxes. Seal the edges of the boxes around the fiberglass with glue and/or packing tape to keep the fiberglass inside and prevent it from spilling out. Put those panels in a line in front of your windows and they should dramatically reduce the sound.
(Fiberglass can be messy, so you have to be careful, maybe working outdoors, wearing old clothes, etc., but people use it all the time. You can get instructions on the Internet.)
One minor problem is that the cardboard reflects sound. You preferably want absorbent material on the side towards you (which is why we used perforated aluminum sheet).
One of the best absorbent materials is egg crates. (This is not styrofoam egg crates, but fluffy cardboard, which has much better acoustics.) People have actually made sound studios using egg crates on the walls and ceiling for sound absorption. You used to be able to get egg crates from big restaurants that used lots of eggs. (These are not t
Tissue-culture bioreactors, as far as I know, usually grow tissue on fetal calf serum, which come from slaughtered cattle. (Sometimes they use mice, fertilized chicken eggs, etc.)
So you have to slaughter even more cattle to create leather in bioreactors.
Right?
The vaguer message of Occupy was that the Democratic Party in the US has utterly ignored the liberals in their base in an effort to pander to Wall St and the right wing. And why should people like Obama do that, when all they need to do to get their votes is scare the bejeesus out of them by threatening them with the prospect of President Romney?
I heard Amy Goodman of Democracy Now give a good answer to that, when she introduced Ralph Nader in the 2000 election. The Republican Party has been moving further and further to the right. The Democratic Party has been moving further to the right to match them. On domestic policy, the Democratic Party is further to the right now than Richard Nixon (don't forget, Nixon's secretary of HEW was Pat Moynihan). If we continue to vote for the Democratic Party, they will continue to move to the right until there's no meaningful difference between them. We have to vote for third party candidates to tell the Democrats that they can't take us for granted.
Since that time, Obama gave us a health care plan that was literally written by the Heritage Foundation. His chief of staff, Rahm Emanuel, told progressives that they were "fucking retarded" for wanting the single payer system Obama promised us. He appointed Wall Street financiers to run his White House. They tossed Acorn under the bus, which was one of the best campaign organizing tools they had.
What I don't understand is why the Democrats didn't learn from 2000 that if you tell the left wing of your party to go fuck themselves, you can lose an election. Maybe it's like training a mule -- you have to hit them on the head with a sledgehammer -- again.
I think it's like a strike. You don't want to go on strike, you don't want to lose weeks or months of salary, you don't want to take a chance on having your employer move to China. But if we hadn't gone on strike over the last 100 years, we would be making Chinese wages right now, and if we never go on strike, we will be making Chinese wages.
Somebody tell the Democrats. If you tell us to fuck off one more time, we'll fuck up your election, just like we did in 1968 and in 2000.
They didn't know because the Bloomberg Administration kicked them out of Zucotti Park and kept chasing them away from any other place they tried to set up.
Hell, the Bloomberg Administration wouldn't even let them use microphones. They refused to let them rent portable toilets, and then complained when they used restaurant bathrooms.
How much would the Americans have accomplished if the Continental Congress had to disband after 2 months?
OWS accomplished one good, important thing: They said that the richest 1% of the country had a wildly disproportionate amount of wealth and power, didn't deserve it, and was using it to fuck the rest of us. A lot of people believed that, but that wasn't in the public discussion. OWS put it into the public discussion.
What do we do about it? Well, as Bertold Brecht said, you have to answer that question yourself.
One thing we don't do is make compromises, write a platform, collect $100 million, and try to beat the 2 big parties at their own game.
Another thing we don't do -- I don't think -- is to organize a movement and deliver ourselves to the service of the Democratic party, who will tell us again after the next election, "Thank you very much, now go fuck yourselves as we make some deals with our corporate masters, the Blue Dogs and Republicans."
But you have to think for yourself and make your own decisions. OWS did enough thinking for you. You don't like it? Go start your own organization.
I shudder to think what kind of environment one must live in to suffer from such illusions.
The Bush White House? Commentary magazine?
0123456789ABCDEF
I sympathize with the advertising industry, and the companies like Google that have done so many kewl things with the money they made out of Internet ads.
But the Internet killed off the newspaper and magazine business, where I used to work. At one time, newspapers in every city would hire a lot of reporters to spend a lot of time following the issues and informing their readers what was going on. Now they've been laying off their reporters, they're down to skeleton crews, and some of the great newspapers went out of business. The advertising, much as I hated it, was paying the bills. Now it's decimated. Blogs are nice, but they really don't replace the newspaper model that served pretty well for the last 2 or 3 centuries.
Technology improves, we got what we wished for, the market changes, and there are winners and losers. (There are a lot more losers than most of us expected.) If people delete their cookies every morning, it might make it harder for web advertisers to make money, but we've all got problems. Let the market rule, it will anyway. This isn't tragic enough to give these guys a government favor.
White-outs? To learn typing?
Misogynistic. That's the word.
First, I'm not sure that the story is true. A lot of stories get "improved" with the telling. It sounds like the dialog of some porn movies that I prefer to watch with the sound off. There might be a germ of truth to it.
Second, if it is true, that girl is one obnoxious bitch. It's one thing to be flirtatious; it's another thing to sleep with everybody in the office for the joy of playing them off against each other. This is a good way to disrupt the office and get fired. If true.
And I don't see how it's particularly an issue in programming. The (exaggerated) stereotype is that male programmers are socially inept nerds. This sounds like one socially inept woman. If true.
Michigan's speed limit is a maximum of 70mph, not 75. http://en.wikipedia.org/wiki/Speed_limits_in_the_United_States#Michigan So they don't have any experience with raising the speed to 75mph.
I think this is what you mean. http://www.mlive.com/news/index.ssf/2012/06/should_michigan_up_its_70_mph.html
There are 2 issues here: (1) How safe is it to drive at 85mph and (2) How safe is it to have people on a highway ignoring the posted signs and driving at different speeds?
The rational way to set speed limits would be for society to decide what your cost/benefit ratio is -- how many deaths and injuries are you willing to accept in exchange for what level of driving convenience? Then you post the speed limits (and minimums) and stick to it. The best way to enforce laws is to have small penalties and strict enforcement. The worst way is to have severe penalties and arbitrary enforcement.
We have a culture especially in some parts of the country where some people just don't want to obey the speed laws. It's very unpopular to have the police enforce the speed limits effectively, so they just don't do it. People just drive as fast as they want. And the enforcement is arbitrary. Why stop some cars and not others? They've turned traffic safety into a game.
Maybe it's a libertarian culture, maybe it's a macho culture, maybe it's just people who say, "I don't give a fuck what anybody else thinks, I'm going to do whatever I want."
The price you pay is that people get killed and severely injured. There are so many
vehicle accidents in the U.S. that almost everyone knows somebody who was killed or severely injured. I had one friend who died, and another who went through the windshield, had most of her face smashed up, lost her front teeth, and spent 6 months in the hospital. Do you think it's worth it in order to drive 75mph? I don't.
It's safer to have everybody driving at the same speed. But other things being equal, when you drive faster, the fatalities go up.
If you read and understood the literature you'd know what I was talking about. I have nothing more to say to abusive people like you.