Follow by Email

Tuesday, July 29, 2014

More Plagiarism in the News

I have a little bit of a fixation with plagiarism, it seems.  I've written about it twice:  here and here, about plagiarism generally (with specific mention of Jonah Lehrer, who was in the news at the time) and about Fareed Zakaria, who I hypothesized was willing to admit to possible plagiarism rather than admit that he has other people doing his writing for him.

And now the Washington Post has a big summary of spate of recent academic plagiarisms here, courtesy of the blog Retraction Watch, which reports on not just plagiarism, but falsified data.

Not that anybody asked me, but it seems to me that plagiarism and falsifying data are two separate things, morally and practically speaking.  Plagiarism typically means you are too lazy and too unoriginal to do your own work.  That's why it's such a common problem in high school and college -- the kids have other things to do.  But I think anyone with a modicum of self-respect outgrows it eventually. In the academic and writing world, my sense is still that underlying most cases of plagiarism is the fact that the "author" has been allowing someone else -- possibly a college student or a high school student -- to do his or her writing for him or her.  In other words, plagiarism can still be blamed mostly on the kids.  But as mentioned in connection with Zakaria, most of these academics/journalists would rather "admit" that they might have accidentally borrowed someone else's work than admit that they don't even do their own writing in the first place.

As for falsifying data, that's a different and more dangerous animal.  And what's scary is that the people who do it are the smarties -- they are the phD's and MDs who got great grades and are living their dreams -- doing important research.  The temptation to simply invent data is quite great -- you need to be doing exciting work to get recognition and grant money, and sometimes, the data you get in the real world is just not all that exciting.  This reminded me of  Diederik_Stapel, who I read about in Ben Goldacre's "Bad Science," a great book all about rampant falsification of data.

It would be very interesting to do brain scans (or some other kind of test) of people who have falsified data, just to find out if they are different from those who don't.  My guess is that for some people, falsifying data is simply unthinkable, but for others, it's quite thinkable.  It would be very interesting to know if there is some way to tell in advance who those others are.  And then we get into the very interesting moral question -- aren't the ones for whom it is "thinkable," but who don't do it, morally superior to those for whom it is simply unthinkable?  

Sunday, July 27, 2014

Polling Problems re Marijuana Legalization

Today's New York Times has an editorial -- by the editorial board -- in favor of legalizing marijuana; i.e. ending the 40-year federal ban on it, although continuing to prohibit sales to kids under 21.  All of the comments agree; tobacco and alcohol are clearly worse for you, and the costs of "fighting" the law against marijuana are enormous.  Not just the financial costs of catching people, trying them, and in many cases throwing them in jail, but also the social costs of ruining lives (disproportionately those of young black men) by criminalizing this kind of conduct, which, after all, is widely accepted socially.

Reading that motivated me to see how we are doing here in Virginia, and that led me to this article. It turns out the vast majority of Virginians support legalization for medical use, but it's a close call for legalization generally.  What's interesting to me is that they say that people 65 and older are 2-1 against legalization.  I could be wrong, but it sounds like they were pretty stupid in dividing up the "age groups" in the poll.  Presumably they have pre-designed polls where they record age groups, and one of the categories is simply 65 and older.  In general, that's a pretty logical category -- 65 is an important age re retirement, medicare eligibity, etc., so in general it's a good dividing line --, but for marijuana use, it's almost certainly really stupid, and it probably masks some interesting information.

Why do I say that?  Think about it -- people who are 65 today (in 2014) were born in 1949 or 1948, and thus came of age in the 1960s, and were at high risk of becoming hippies and potheads.  Many of them still smoke pot today.  For reference, Paul McCartney was born in 1942, John Lennon in 1940, and Mick Jagger in 1943.  The summer of love was 1967, when someone born in 1949 was 18.  These people were only 20 in 1969, when Woodstock happened.  In other words, 65 is a pretty dumb dividing line when it comes to dividing young from old in the pot legalization debate.  I have no doubt that people who came of age in the 50s -- people born e.g. a decade earlier; in the late 1930s -- would today be against marijuana.  By the time of Woodstock, they all had families and jobs and had basically missed the marijuana boat.  And there are a lot of them out there.

[Here's an aside that I need to remember to ask next time I'm talking to someone about the science of polling.  I personally never have time for polls, and I would guess there are many other busy people like me who just don't take them.  But older people have more time, and also probably feel strongly about certain issues and want their views to be heard.  So my guess is that the age distribution of poll-respondents differs greatly from the distribution of the normal population.  Am I right?  And if I am, do polls correct for this?  It's not enough to simply report the overall result and then say, as the Virginia poll does, that a majority of young people favor legalization, while 2/3 of old people are against it.  That doesn't tell me if there has been a correction for age distribution in reporting the overall result.  In other words, one can imagine two worlds in the science of polling:   (1) the poll simply interviews 1000 people, notes their ages, but then does no correction, and (2) the poll somehow does a correction.  If 500 of those people are 65 and older, and they think differently on the polled issue than the rest of the population, you will get a highly misleading result absent a correction.  One can imagine a number of different ways to try to correct for this.  I.e. one would get the overall votes for each age group, and then "weight" the age group's contribution based on the true age distribution in the population.  I.e. even if a disproportionate number of old people respond to your poll, you take their response as representative of "old people" generally, and do the same for each age group.  Then you use what we know about the percentage of "old people" and other age groups in the country to guess what the true distribution is.  This then raises questions about how to ensure that the results in any given age group are accurate -- i.e. you need to poll above a certain number.  And then that presumably introduces a separate potential error for each age group.  I would guess that the "science of polling" has answers to these questions, but then again, most polls are so unreliable that maybe they just don't.]

Back to the marijuana poll.  Given the almost certain difference in views between people aged 65-70 or maybe 75, and those 75 and older, it seems to me the poll almost certainly should have divided up the age groups differently.  In fact, it's possible that those aged 65-70 would feel more strongly about legalizing pot than any other age group.  And yet they were simply averaged in to the older age group, which consists of many people who never even had the opportunity to try pot.

It's interesting also that the vote in Virginia was 84% in favor of legalizing medical marijuana.  One can readily see how all the anti-pothead prejudices of the 50's generation fell away with that question -- if there's a chance that they themselves would get to use it legally, and it might help them medically, then of course it should be legal for that purpose.

One thing that this tells us is that the anti-marijuana sentiment will die out soon enough.  I.e. really, it's only people currently over 75 or 80 that are strongly against legalization to begin with, and the ranks of those people are getting thinner every year.  These people are mostly out of touch on the marijuana issue, and are dying out anyway, but because they vote strongly against it, polls make it look like a close call here in Virginia.  Really, it isn't.  But politicians don't care -- they need to keep currying favor with this bloc.

from the rest of they simply report the results, which

.  Of course, the ]    

are trying tois that

Friday, July 18, 2014

True Cost of a Burger by Mark Bittman in NYT

NYT food writer Mark Bittman tries his hand at economics in The True Cost of a Burger in a recent NYT, which I just noticed today (dated July 15, 2014).  He has been working on it for a year, and wrote it with the help of an intern.
He starts off as follows:
"In 2005, the House of Representatives passed an act that forbade consumers to sue fast-food operators over weight gain."
I'll suspend judgment on his economics for now, but it looks like he's got a problem with civics.  An "Act" is legislation that has been enACTed -- i.e., a bill that has passed the House, the Senate, and been signed by the President.  The House "passes" stupid bills all the time for symbolic reasons, knowing full well they have no chance of success in the Senate.  Anyway, the "Cheeseburger" bill was not enacted, and thus is not an "act."

{Update:  having now read the whole article, I can say that his grasp of economics is awful.  See below}
The point of the article is to add up the "external costs" of cheeseburgers, and add them to the price of the cheeseburger, to arrive at a "true cost."  In his view,  "cheeseburgers are the coal of the food world, with externalities in spades."  I don't know enough about the coal industry or the rest of the food industry to judge this statement on any level, so I won't.

His conclusion is that given the "average" cost of a burger of $4.49, the easily-calculable external costs range from 68 cents to $2.90 per burger.)
He mentions and in some cases quantifies the following:
Litter -- the social cost of cheeseburger wrappers that are not put in the trash. 
Carbon generation -- According to those who know (see “Meat Eater’s Guide” (2011)), the carbon footprint of beef cattle is 27 pounds of CO2 equivalent produced per pound of beef, but when you subtract the savings used by adding ground-up dairy cows into your cheeseburgers, it's down to 25 pounds of CO2 emissions.  Cheese produces 13.5 pounds of CO2 equivalent per pound; he notes but does not quantify the carbon footprint of bread. Estimated cost:  53 cents per burger, based on an average of three competing monetary valuations of greenhouse gas pollution (ranging from $37 per metric ton [the government's official rate, which works out to 15 cents per burger] to an estimate that's almost 10 times as great.

Obviously, when two estimates are off by a factor of ten that means that someone is very wrong, but Bittman can be forgiven for reporting them all and then just averaging.  I'm a big believer in requiring producers to bear the costs of the externalities that they cause, but the problem is in quantifying those externalities.  It sounds like the carbon footprint externalities here are at least 15 cents per burger, so that's a start.  But this is the only place where Bittman's analysis of externalities makes any sense.
Chronic disease costs: 
-- He acknowledges evidence that red meat may increase risk of cardiovascular disease, but because the evidence is not certain, he leaves that cost out.
-- He states that "a main factor in the rise of obesity has been an increase in the availability of calorie-dense foods, and burgers played a big role in this process."  This sounds logical, but when he tries to quantify it, his point loses all logical force.

Here's the process:

--  "Between 1970 and 2000, per capita calorie intake increased by 24 percent . . . ."

-- During the same period, "the 'food-away-from-home sector' grew to nearly half of all food we eat."

-- Restaurants, of course, are the source of most burger consumption.

I suppose this is true, based of the amounts of hamburger buns and meat available at every supermarket, it's a fair bet that a lot of people make them at home, too.  What's interesting is that in the end, his analysis of health-related externalities is limited to externalities caused by fast food burgers, when those same externalities are caused by home-cooked burgers as well.  
-- "Between 2007 and 2010, 11.3 percent of adult Americans’ daily caloric intake came from fast food."

It's not clear how this squares with the apparent fact that half of the food we eat is from the "food away from home sector."  If he's only focusing on fast food, why did he give us the other statistic?

-- "Correlation is not causation, of course, and it seems likely that foods high in sugar and other hyperprocessed carbohydrates are most responsible for high obesity rates, but burgers certainly played a role in rising caloric intake."

The last two bullet points are successive sentences in the article.  I still don't get it.  All we know from the first sentence is that 11.3 percent of caloric intake during a particular 3 or 4 year period came from fast food.  It's hard to relate this to anything else we know.  It's also hard to understand how it relates to his earlier statistic that nearly half of our food intake is from the "food away from home sector," by which I guess he means restaurants (not clear where e.g. workplace cafeterias fall into this; but his stat that 11.3 percent of our calories come from fast food suggest that fast food is only 11.3/50 of our away-from-home food intake).  And there's an acknowledgment that the main cause of obesity is probably something other than burgers (sugar and hyperprocessed carbs).  But hold on, there's more:
"To estimate the share of obesity-related costs resulting from burger consumption, we estimated the share of calories coming from burgers in fast-food restaurants, where the majority are eaten."  

I have to do this one sentence at a time; here I think he is saying we will determine what percent of fast-food restaurant calories come from burgers, and that this will somehow help us understand the overall share of obesity costs attributable to burgers. 

"Assuming that the 11.3 percent of calories is proportional to the incidence rate of obesity (it may be higher), its associated health risks, and its treatment costs, up to 15 percent of fast food’s share of direct and indirect costs arising from obesity (about 1.65 percent of the whole) are attributable to burgers."

Here's where he simply loses me.  The reported 11.3 percent of calories is just our daily percentage of fast-food intake (in calories).  Why should that necessarily be proportional to the incidence rate of obesity?  I guess he is saying that our overall food intake is responsible for our obesity.  But by saying that, isn't he giving away the whole game?  Doesn't this mean that because 88.7 percent of our calories come from non-fast food, we can look at each of those items and determine their contribution to obesity?  Or put another way, if bread is 20% of our diet and carrots are 1% of our diet, doesn't his logic lead to the conclusion that blame bread for something like $50 billion in diet-related health care costs, and carrots for something like $2 billion?  He seems to have calculated that about 1.65% of our total calories come from fast-food burgers, and from this he is saying that fast-food burgers are to blame for 1.65% of obesity's costs.  You could isolate any other part of our diet -- e.g., as above, bread or carrots -- and come up with a similar, or much larger, percentage.  

"The link between obesity and a handful of deadly chronic diseases — arthritis, cardiovascular disease, hypertension, Type 2 diabetes and some cancers, among others — is well documented, as is their enormous economic burden. Direct medical diet-related costs are currently pegged at about $231 billion annually." 

I have to keep quoting, just to be fair; I don't want anyone to think I'm mischaracterizing.  The bizarre thing is that by not making any attempt to distinguish between kinds of foods, he is almost certainly UNDERESTIMATING the effect of burgers, without realizing it.  But I say that just because I happen to believe that there is a causative correlation between excessive burger intake and obesity.  His data does nothing to prove this.

Of course, it's possible to read the data and say we all know that burgers are worse for you than other constituents of your diet.  But we will be generous and treat them as being no more or less linked to diet-related health care costs than any other food item, and that's how we know our numbers are conservative.  But that doesn't seem to be what he is saying.  And the bottom line is that if some Americans are eating too much food, that doesn't mean that each particular food item (which doubtless are eaten by many thin and healthy people as well) should bear some amount of the externalities caused by that.  One needs to figure out WHY those particular Americans are eating so much, and deal with the problem there. 
"These numbers above would mean that this cost of burgers is about $4 billion per year (from fast food burgers only!), which averages out to 48 cents per burger. (Some put these costs five or six times as high, and there are indirect costs as well; again, we’re being conservative.) And between 2010 and 2030, the combined costs arising directly from diseases related to obesity could increase by an additional $52 to $71 billion each year. This could double the cost per burger in additional health costs alone."

Again, what's missing from this article is any evidence that burgers contribute to obesity more than other foods in our diet.   It's intuitive, but the data and analysis presented here almost point the other way.
Other "vaguely calculable" non-zero costs:

-- cost of dealing with "elevated nitrates in water supplies resulting from the chemical fertilizers used to grow corn to feed cattle" [true, but to be fair one would have to compare this sort of thing with pollution etc caused by other production of other foods]

-- "cost of food stamps and other public welfare programs made necessary in part by the ultralow wages paid at most fast-food operations" 

This one seems crazy and reveals an uninformed liberal agenda (again, I'm not particular conservative or liberal -- I just have trouble with uninformed partisan agendas).  It's equally possible (and maybe even more plausible) that the availability of these "ultralow"-wage jobs reduces the overall need for foodstamps, and that the availability of inexpensive fast food saves money for poor people.]

-- "the beef industry’s role in increasing antibiotic resistance, which costs, according to the Centers for Disease Control and Prevention, something like $55 billion a year; some measure of E. coli illnesses; and land erosion, pesticide residues, direct corn subsidies, injury rates at slaughterhouses, and so on."  

I absolutely agree that this is a problem.  But this is where he should have real statistics.  Otherwise, there's just no way to say that burgers are any worse than bread.  And to all this I would add the simple moral cruelty of raising animals for food; although of course that doesn't apply just to beef.

"Even more difficult to calculate":

-- "the 'cost' of a shortened life"

-- "the value of loss of biodiversity that results from the destruction of rain forests to provide land for cattle or their feed."

-- "an emerging body of research linking decreased male sperm quality to mothers’ beef consumption."

He concludes with more numbers that don't seem to add up:

"Last year, burger chains grossed about $70 billion in sales. So it’s not a stretch to say that the external costs of burgers may be as high as, or even outweigh, the “benefits” (if indeed there are any other than profits)."

A minute ago he said that he was conservatively estimating the externalities due to burgers at $4 billion (48 cents per burger) for health care, plus 53 cents per burger (say another $4 billion) for carbon.  So that's about $8 billion in external costs.  Note that he could do a similar calculation for all the non-burger fast food that is sold -- by his reasoning, if fast food burgers make up only 15% of 11.3% of our calories, then 85% of our fast food calories -- or 9.6% of our total calories, create another health-care related externality that dwarfs that caused by the burgers (i.e. over $20 billion).  So that means the $70 billion business is causing about $30 billion harm, by his reasoning. 

Economically speaking, it's also annoying that he only concludes that the costs "may" outweigh the benefits (to which he snarkily adds, "if indeed there are any other than profits").  The word "may" renders the statement completely meaningless.  And the lack of any actual attempt to quantify the benefits (and only a very poor attempt to quantify the externalities) makes this a very sloppy statement.  The benefits may well include the availability of jobs for unskilled labor, the availability of cheap and convenient food, the added productivity that comes as a result of access to cheap and convenient food, etc.  Fast food, eaten in moderation, may well be a very good thing for those who eat it that way.

 If those externalities were borne by their producers rather than by consumers and society at large, the industry would be a highly unprofitable, even silly one. It would either cease to exist or be forced to raise its prices significantly.  

By his own statistics, maybe 68 cents per $4.49 burger, which doesn't seem like all that much.  But as above, he'd have to do a lot more to prove that fast food is a "silly" business.   And of course, if you make them internalize their externalities, you'd have to do the same to all their competitors (grocery food, restaurant food, etc), so "fast food" might still remain comparatively cheaper.

And now for the grand sum-up:

"In this discussion, the cheeseburger is simply a symbol of a food system gone awry. Industrial food has manipulated cheap prices for excess profit at excess cost to everyone; low prices do not indicate “savings” or true inexpensiveness but deception. And all the products of industrial food consumption have externalities that would be lessened by a system that makes as its primary goal the links among nutrition, fairness and sustainability."

Perhaps this is his admission that, using his methods, we could use bread as an even more compelling symbol.  Sadly, I actually agree with the overriding sentiment of this sentence.  The problem is that his data and methodology do nothing to support his conclusion. 

To summarize my view: there's no question that our diet contributes to health problems, and mere diet-related health care dollar costs vastly underestimate the cost of this problem -- it's not just the dollars that we pay for treatment when our diet makes us sick, it's also all the lost productivity and unhappiness that comes from being sick in the first place.  And intuitively burgers could be part of the problem.  But as he acknowledges, all the studies on the harms due to red meat are speculative, and the same could be said for fat generally.  And of course, different people are affected differently by different types of food.  

There's no question that the carbon footprint caused by raising beef is an externality that should be internalized somehow.  But I feel that way about almost all externalities.

And this is a horribly misguided way of placing blame for obesity on one particular food.  A much better way to do that is by studying obese people (especially obese people with health problems) and figuring out what is causing their obesity from there. 

Since we're talking about fast food, I'm going to make one point, that is related only to the question of "benefits" of fast food, above.  In all my years of eating fast food, I don't recall EVER having gotten sick as a result of doing so.  On the other hand, I get sick an appreciable number of times after eating at mid-to-high priced restaurants.  Maybe around 10% of the time.  And sometimes very sick -- bad colds, or food poisoning.  I think the reason is that the fast food is processed, and the workers don't contact it nearly as much.  And for many restaurants, workers are reluctant to call in sick and show up for work sick.  This, plus their closer contact to the food -- which is fresher, but also more liable to spoil or have contaminants in it already -- puts customers at far greater risk than fast food.

Friday, July 4, 2014

Insurance for contraceptives and abortions?

Honestly, I might just be out of touch.  But today I happened to read two different articles, both of which rubbed me the wrong way in the same way.  First there was this by Elias Isquith in Salon (followed soon thereafter by this, touting Ruth Bader Ginsburg as the greatest justice since sliced bread).  Then there was this by Loren Clark-Moe in the Washington Post.  I know that abortion and contraception are sensitive topics and I sense that they are particularly sensitive for women, which is why I am trying to be sensitive here.  I'd like to ask the question wholly apart from the normative issue of whether one is pro-contraception (I am) or pro-choice (I am, sort of, but I'm also disturbed by abortion).  Isquith seems furious that Hobby Lobby doesn't have to pay for contraception for its employees, just because of some made-up test about closely-held corporations, courtesy of the Supreme Court.  I get (and agree with) some of the complaints about that case -- yes, the Court has told us that corporations are people, but really, do they have religious beliefs and sensibilities too?  But Isquith is really mad at something else -- just the fact that those poor Hobby Lobby employees have to pay for their own contraception.  And that brings me to poor Ms. Clark-Moe -- who considers it a "punishment" that she had to pay for her own $480 abortion.

My question is WHY HAVE WE GOTTEN USED TO THINKING THAT INSURANCE IS SUPPOSED TO PAY FOR EVERYTHING??  Don't we realize that just means the cost of insurance will go up?  Wouldn't it be more efficient to just let us pay for the small -- and routine and predictable -- stuff ourselves?

But that's the problem with things like ObamaCare -- everybody, including the insurance companies, now is motivated to make insurance as expansive as possible, in the sense that the more transactions are covered by insurance, the more the insurance companies stand to make.  It would  have been so much smarter to provide a system where the "uninsureds" got access to insurance for catastrophic problems.  And if we want to pay for their doctors visits too, then give them tax breaks.

[btw I just had a random unrelated idea -- why don't emergency rooms have a bus service that will drive uninsured, non-urgent patients to nearby clinics?  Sometimes it can take forever to get help in an emergency room (I know, I've been there), whereas a clinic is often much faster.  In fact, it could simply be a service paid for by the clinic, since they would get more business as a result.  On check-in at the emergency room lobby, patients would be told how long they might have to wait and how much they'd have to pay for emergency room treatment, and let them know that a van is waiting to take them to the clinic which will treat them much faster and cheaper]

Again, I understand the basic complaint -- since ObamaCare is a law that was duly enacted by the legislature, and since it apparently requires employers above a certain size to pay for contraception of employees, then that's a legislative judgment that can't be overturned unless it's unconstitutional.  And it's pretty obnoxious of all those Catholic men on the Supreme Court to call it unconstitutional the way they did.  But my point is simpler -- since when can't we pay for our own contraception?  And that goes double for Ms. Clark-Moe.

UPDATE (July 5, 2014) ( the next day):  The NYT reports that the Obama administration is trying to figure out ways to work around the Supreme Court's ruling and deliver contraceptives to these women.  The rationale is that without easy access to contraceptives, there are unwanted pregnancies and overall bad effects on women's health.  This is true enough.  But it's still a distribution question.  ObamaCare (from what I understand from the Hobby Lobby ruling) only requires that employers with greater than 50 employees have to provide contraceptive "insurance."  Isn't that a somewhat skewed distribution?  The only people who get access are the people who happen to work for bigger companies.  Often, the people working for smaller businesses need the financial assistance (and that's what free contraceptives are) the most.  If simply giving all women free contraception is a good idea -- and I honestly believe that maybe it is -- then just do that, and stop creating weird and arbitrary arrangements trying to get other people to pay for it.  In the end WE (the consumers and taxpayers) pay for it, in the form of higher taxes and higher prices.

And now I see this from MSN:  This piece echoes some of what I've said above -- that this is just a "bonus" of Obamacare to the already-insured (like the provision allowing kids up to age 26 to stay on their parents' plans).  It says that the free birth control is worth about $269 a year to the women who get it, but interestingly, the article mentions that it does not appear that more women are going on birth control because it's free.  If that's the case, then it's just a simple wealth transfer to a somewhat arbitrarily selected group of women, which specifically excludes the neediest women.

There's a caption here that makes the point ("No, it doesn't cover hip replacements, but you're eligible for free birth control!")