Monday, March 25, 2019

Junk Science of Both Kinds

Lately I've been noticing that I have two basic reactions to science stories that I see linked to around the net.  The first reaction is "that's obviously bullshit" and the second is really the same sort of reaction except for being positive: "I hope they didn't pay too much for that study". 

For an example of the first one, a story started making the rounds 10 days ago that this week eggs are bad for us again.  You may be one of the folks who feel like this old Sidney Harris cartoon:

Being a curious guy, I noticed there was next to nothing in the original reports.  It finally made it online a few days later so it was easy to take a quick look at the abstract for some details. 

Let me start with the disclaimer that I'm not a medical doctor or a statistician, but I'm a lot closer to being a statistician than an MD and that's all I need to find what's wrong with this study. 

Leading with the thing about the study that's good, they have a good number of subjects, nearly 30,000.  The big problem with the study is that it's a meta-analysis of six other studies.  Whenever a statistician gathers data from multiple experiments there's a mountain of hurdles they need to overcome, especially if the original studies were of something other than the effect being studied.  Practically, that means you'd want the risk ratio (the amount that the risk of bad outcome increases) to be large - like twice the risk or three times the risk of "non-treatment" group.  This study showed a relative risk increase of 17%.  While I know there's a whole crisis going on over statistical significance limits, I don't think a 17% relative risk increase is likely to be real.

Like the vast majority of these "he-who" studies, "he who eats N eggs per week" relies on food questionnaires.  The accuracy of these surveys has been widely criticized from the standpoint of memory accuracy (how many eggs did you eat per week last March?) and other things.  In this study, the six datasets being pooled might have handled things like recipes that use whole eggs differently with some counting them and some not counting.  Of course, no study measuring correlation can prove causation and that's another big problem with this study.  Finally, for a look at just how bad some of this can be, read my first post on the King of Junk Food Science.

(Table from Five Thirty Eight)

In the other class, "I hope they didn't pay too much for this", yesterday's Daily Mail is reporting that male and female brains are actually different!  Not only that, the differences are apparently observable while the baby humans are still in utero; during the second half of pregnancy.   I'll say it's interesting the difference is detectable, but is there really any doubt that there are differences between the brains of males and females? 

As you can imagine, the people saying that sexual differences are all learned behaviors are upset over this study, calling it ‘unfounded conclusions’. 

What I want to ask those people is if they've ever just sat back and observed male and female cats, dogs, or other animals.  We have a male and a female cat and they are so obviously different one would have to be extremely non-observant to not see it.  Furthermore, they're both neutered and were both neutered young, so it's not their current levels of testosterone or estrogen.  Influence on the brain before birth or up until they were neutered is completely logical.

I guarantee you our male, Mojo, has never played with toy trucks, hot wheels, or Nerf guns.  Likewise, our female, Aurora, has never played with Barbie dolls or tea sets - and while I think she'd use a tea set if she had one, she doesn't have a thing for shoes.  Their behaviors are absolutely not influenced by TV, advertising, or any of those cultural factors the people saying "they're learned behaviors" claim. 


  1. ACTUAL research costs money.....and is hard work. It's much easier and cheaper to take a bunch of numbers someone else came up with, massage them, squeeze them and
    trot them around for a while and then come to a conclusion then call it a result.
    ACTUAL science is a dying specialty being replaced with agenda driven pseudoscience.

  2. So you've gender-classified your dogs and cats? That could be considered animal cruelty in some parts of America.

  3. The same people who say "I was born this way" also say "all behavior is learned". 'Nuff said.

  4. Studies have shown that 5% of random data is statistically significant to P.05.

  5. I started to have my doubts about "X will kill you" studies a long while ago.

    When microwave ovens first became popular and cheap enough for domestic use, there was a study published saying that heating milk (particularly for babies) would produce toxins and everyone s going to DIE!!! DIE I tells ya!!! OK, I took a while to dig out the original report and the methodology employed. The researchers had sealed some milk in a closed pressure container and microwaved it for (if memory serves me correctly) about 1 1/2 hours.

    Now, if you heat any organic stuff in a closed container under pressure (just like a pressure cooker) at likely 140 degrees C for that length of time, then naturally you will get cross polymerisation of the molecules and something not particularly good for you - caesin glue is made from milk - but just HOW bad is it for your health? I don't know but to tell mothers not to microwave their babies milk for 15 to 20 seconds based on that experiment is ridiculous.

    Similarly, tea is bad for you. There is enough tannin in 12 gallons of tea to kill a fly, so stop drinking it immediately. Or perhaps not.

    Only Kalifornia has taken this to heart and extremes where everything is harmful to human health in a sufficiently large quantity.

    However, it is not just eggs. I can never remember if red wine is good for you this week/month or not ... And so it goes.

    One fo my friends is as fit as a fiddle and is literally piano wire and whipcord, exercising and watching his diet. He's 64 and was recently taken to hospital for a mild heart attack. I'm convinced that genetics plays a much higher role in your health than worrying over the number of eggs that you eat in a month.

    But I don't rely on research grants or special interest lobbies for my income.

    Phil B

  6. "ACTUAL science is a dying specialty being replaced with agenda driven pseudoscience."

    "I don't rely on research grants or special interest lobbies for my income." Bingo!!!

    When I read about these "studies" that claim something is now "bad" for you, I try to find the source of the funding for the study. In quite a few instances, you'll find that the funding was provided by a competing entity or industry. I think back to the saccharin debacle from back in 60's?/70's. Not only did the saccharin experiment feed the rats 1000's of times more saccharin than they should have gotten based on the rat's body weight compared to a human, but the "study" was funded by BIG SUGAR.

    Low and behold 50 or so years after the initial "bad" result study on saccharin, a new study, from the early 00's if memory serves, proclaimed saccharin was not harmful to humans ... or rats.


    1. You're both right on.

      One of the things that shocked me was when I found out that there is a "big vegan" or "big vegetarian" out there and they have tremendous influence over these studies. How much of what we think is bad and good comes from Kellogg's, or the grain industry? (big agra?) A lot.

      I started paying more attention to this in the mid-90s. We were totally indoctrinated into the lowfat, almost vegetarian lifestyle. Then my wife got cancer. Along with the starter kit, we get a dietary advise brochure from the Cancer Society that says things like, "hard boiled eggs make a good snack that keep your protein up", "remember, your immune system is almost 100% protein" and "too dry? Add a tablespoon of butter" It made me start researching more and believing less of those studies. Then I started reading the actual studies and applying what I know of statistics.

      If any company had fudged the data like Ancel Keys, who started the whole diet/heart thing, fudged his, the main people would be doing hard time in Leavenworth.

  7. The whole egg/butter/trans fats (male or female? And how do you dress your salad?) nonsense is, I believe, all based on the agro _and_ pharma corporations that have been funding much of the scientific and medical research to "prove" a specific result, either in favor of their product, or to make a competitor's product look bad.

    I raised chickens and milk goats years ago, and discovered back then (late '80s, early '90s) that some people respond to dietary cholesterol and some (the majority, IIRC) don't. I ate eggs everyday, as you can imagine, even though I did sell many dozens to a local co-op. I had my cholesterol/lipids checked after a two month period without eating any eggs, and my numbers were the same as when eating them every day.

    Now, thanks to the whole global warming/climate change scam, scientists can no longer be trusted to actually use the scientific method, let alone search for the truth, as opposed to a desired outcome. Yes, there are still many true scientists who will publish the truth, even if it wasn't what they had hoped it would be, but far too many of them prostitute their names and their results for monetary benefit, with research funded by companies with a wallet in the game.

    The current use of "digital modeling" instead of real world observations and measurements has led to a new industry where "scientists" use "CGI" (Computer Generated Integers) to produce the results they desire. Some of them have even admitted that they know the numbers are false, but make the statement that it would be better for the environment and the world if the numbers were accepted as being true.

    In almost every case the "cui bono" question provides the real answers.