Tuesday, May 18, 2010

The Media Is like High School For Science...

Ever have a rumor stat about you when you were in high school?

Infuriating, isn't it?

Sometimes it's a half truth, and sometimes it's just something taken out of context.  Either way, the efficiency of the rumor spreading is way higher than the efficiency of you explaining the whole story or putting what was actually said into context.  You converted one or two people, but everyone else you confronted just laughed in your face, called you a liar, and repeated the rumor to more people.

Many things in life are counter-intuitive.  We anticipate one thing but the other thing is true.  Less is sometimes more.  You really do, on average, hit the golf ball farther and straighter when you don't try to swing so hard, and so on.  These same types of things exist in food science.  Although, it's not that they're counter-intuitive, it's that the public has a bad intuition.

Let's look at BPA, for example.  Does everyone remember the statement put out by the FDA in regards to BPA in January?  What did it say?  Some remember it as an admission that the government was wrong about BPA and that we need to take a careful look at how it's regulated.  This is one of those rumors that start by something being taken out of context.  What the FDA said was that BPA was safe, but there haven't been studies on very subtle effects and that effects on the very young warranted further study.  But in several stories that came out since then, it was reported as if the government changed their mind on all the established science.

This is also a great example of how efficient rumors are vs. explaining the whole story.  For those who've really looked into BPA, you'll know that BPA(like most chemicals including some vitamins) have a level at which they cause harm.  So it has the potential to do harm, and the article(rumor) reports that it's in everyone's bodies, so it must be harming our bodies, right?  Hold on.  This is the type of rumor where you need to explain, and most people don't sit still for it, but let me have a crack at it.  BPA is established as a GRAS food additive.  It's not mixed directly in food, but rather it makes up part of the packaging that the food touches, so it must be labeled as an additive.  Anything GRAS has a limit on how it can be used and at what quantities.  That's per the Code of Federal Regulations.  That limit is based on the Lowest Observable Effect Level determined in lab tests.  Just to be sure it's safe enough to use, they take the quantity that is determined to cause the LOEL and divide it by as much as 1,000.  Now, the studies on BPA all say that it can't be proven that BPA is a definitive danger.  Furthermore, some studies show us how much BPA was found in a particular canned product.  I've mentioned this before but one of the highest BPA concentrations was found in canned fruit.  Here's the part you never get to read.  At those levels, a person would have to eat several hundred pounds of the fruit every day to reach the LOEL.  But the published rumor is titled, 'BPA in Cans Poses Health Threat, Report Claims'.

This has been going on for a while with different things.  I remember a similar story about Tab years ago and they figured you'd need to drink hundreds of cans a day to get ill from whatever the chemical was.

Salt is one of the relatively new bad guys.  The way you hear it told, the food industry is poisoning us with dangerous levels of sodium and they say it causes 100,000 deaths a year.  Wow, that's a lot of people.  The trouble is that BPA supposedly causes a lot of death's too, as does HFCS, as does trans-fat, but where do the numbers come from?


Correlation got you into fights in high school.  Correlation caused your parents to punish you when you did nothing wrong.  Correlation is basically this:  A is true and B contains A and B contains C, so C is true.  Put another way...  From statistics-help-online.com:
"Let's take some other ludicrous examples to explain the problem of correlation vs. causation. Define $T$ as the temperature of a day in Manhattan, and $I$ as the number of ice cream vendors out on that day. The correlation coefficient between these two is almost certainly quite positive. (How many vendors are out there in January?). Does this prove that ice cream vendors cause it to be hot? Obviously causation goes the other way. Common sense tells you that. Unless of course you believe in conspiracy theories."
 So all the salt people's numbers are based on correlation studies.  Is salt killing us? A clue can be found by observing other countries.  This is helpful in the HFCS argument since sugar is cheaper than HFCS outside of the US and we can compare with obesity and diabetic rates to see how much healthier they are(in the case of HFCS, non HFCS countries don't fare better).  

"When all the surveys in Britain are considered, there has been no consistent downward trend in salt consumption in recent years, said Dr. McCarron, who has been a longtime critic of the salt reformers. (For more on him and his foes, go to nytimes.com/tierneylab.) He said that the most notable feature of the data is how little variation there has been in salt consumption in Britain — and just about everywhere else, too.

Dr. McCarron and his colleagues analyzed surveys from 33 countries around the world and reported that, despite wide differences in diet and culture, people generally consumed about the same amount of salt. There were a few exceptions, like tribes isolated in the Amazon and Africa, but the vast majority of people ate more salt than recommended in the current American dietary guidelines."
 So if we all eat about the same amount of salt but we have varying amounts of high blood pressure and heart trouble, then what does that say about salt causing problems.

A couple rules of thumb for reading news releases about studies:
Scientists don't generally brag, so take note of the ones that do and what their intentions may be.
Regardless of your opinion of the matter, judge a study on it's science and methodology, not it's source of funding.
Look for numbers and look for them in context.  Beware of studies were they are vague aboiut the numbers or use them out of context, as in percentages.
Read the whole study if possible.  Sometimes a researcher or writer of the article will say the opposite of what the study actually proves.
Beware of Meta-analysis.  These are studies of studies and they're never as good as well designed original studies.
Learn statistics. This will make you smarter and less likely to be taken advantage of by con men or people with an agenda.
One study proves nothing!  Reproducibility is king.  A good study can be duplicated and if it's true, the results will be reproduced.
Correlations are a starting point in a scientific quandary, not the end.  You need causation, not correlation studies and a lot of percentages to be validated.

No comments:

Post a Comment

Put your comment here, kind sir/madame. Try to cite sources when stating facts and refrain from off topic comments or hateful/nasty rhetoric.