Sunday, November 9, 2014

association and causation -- a rant

Oh, the hypocrisy!  [sigh]

I had never heard the expression "association does not prove causation" till I started reading nutrition-centered websites, but it made perfect sense to me.  Since that time, I probably see people pointing it out several times per week.  I respect the concept ... and use it often myself, especially when refuting claims of silly people on Facebook.  ;-)

But when I read someone stating it, and then the very next day they reel off all sorts of epidemiological observations as support for their pet notions, it really leaves a bad taste in my mouth.  They bloody well know better -- there's no excuse for it.

Sometimes observational studies are the best we can hope for -- an RCT is completely impractical for some hypotheses, and totally impossible for others.  We HAVE to depend on animal studies for some things, and it's a real pity to "do that" to them as well, but how else can we learn certain things?  I understand that it's easier on the consciences of experimenters to torture mice than dogs, but mice make piss-poor surrogates for human subjects....

In the absence of proper human experimentation, we need to use open-minded INTELLIGENCE in choosing supportable hypotheses about what we should do concerning this-or-that problem.  Do we see a pattern in rat studies which doesn't pan out in clinical experience?  Then it's GARBAGE -- chuck it out and move on to a different approach.  Are there clear relationships in a study among healthy young athletes which aren't seen in the aging population?  Then it's not universal, and it shouldn't be promoted as such. 

It would make things SO much easier if bad science were spelled out clearly in titles and headlines:  "Test-tube Science Proves Spontaneous Generation" or "According to Questionnaire Recalling Food Intake of Eight Years Ago, Pineapples Cause Acne"....  But no, a lot of "science" exists nowadays with the mere goal of courting future grants, not increasing the sum of human knowledge.  :-(  Journalists want readership, not wisdom.  Brainless, money-wasting studies have to have catchy titles and positively-phrased abstracts, or their authors will have to find jobs in the real world -- trash-collecting would probably be appropriate.

Reasons why [ahem] certain blogs are not on my list --> include some writers' support for NOTIONS THEY WANT TO BELIEVE IN using "evidence" as nebulous as the claims they decry in their competitors.  If X isn't allowed to use mouse studies as support, than Z isn't, either.  If questionnaires aren't adequate to support hypothesis-F, then they're not allowed as proof of hypothesis-G.  What's sauce for the goose is ... oh, you know.

You can't have it both ways, doc -- and you only damage your own credibility by CLAIMING to be more discriminating.

9 comments:

  1. That G wasn't me, amirite? LOL

    ReplyDelete
    Replies
    1. :-) you're right! I had x and z in the previous sentence, so I needed two other letters. admittedly, F stands for food-reward, and G just follows it.

      Delete
  2. Is this self inflicted aggravation? You did read that blog post.

    ReplyDelete
    Replies
    1. just skimmed it.... :-) when I read things that annoy me, I often quit early and run back "home" to rant about it!

      Delete
  3. One thing about association and causation, though (as you know): if you don't find a relationship, there's probably no causation.

    But I thought about this after my last blog post on fat: Roseto, PA had almost no heart disease when they were eating lard. But the French have lower rates of heart disease than many other countries even though they smoke a lot. It doesn't mean that smoking is protective. I'm not sure how to address this.

    ReplyDelete
    Replies
    1. Dr. Feinman once wrote a post about how associations sometimes DO imply causation, but my gripe is with those who whine about correlations then turn around and quote them, themselves. :-)

      the "French paradox" has a lot more confounders than fat intake and smoking -- the traditional culture is significantly different from the Anglo- and German-American ones, but I fear that is dissolving -- PITY!

      Delete
  4. Recently Razzwell posted an interesting comment on Dr.Eades blog

    "Most scientific ideas are wrong. Most experiments are wrong,too, the first time they are done. This is not taught often enough.
    ALL science can EVER do is say what is plausible. What is more or less likely. At best, highly likely or highly unlikely. It can show us what is wrong. We can know we are wrong, but never that we are completely right.

    Newton's laws for planetary motion were found incorrect about the orbit of Mercury. This was spotted a few hundred years later.
    Correlations, in science, do not mean anything UNLESS there is an underlying principle that deeply explains the correlation.
    Evidence alone does NOT determine theory choice as Albert Einstein new very well. This is a VERY important point.
    I talk to some of the best science educators in the world . I would love to correct most of the Blogsphere's misinformation about "laws." They remain always provisional and always perfectible. Physicists fully expect them to be modified with further advancements. "Immutability" is not even remotely a characteristic of a "law."
    All of these Internet gurus are so easy to expose."

    Many people act like they know the ultimate truth,or bring authorities to support their opinions, who look like the people who must know answers to all questions, but it is always a good possibility that the final truth is way to complex , and it may satisfy no one.

    ReplyDelete
    Replies
    1. yep, he hit the nail on the head on that one. I think I remember he was talking about that infamous "first law of thermodynamics" which so many people take out of context....

      I have mixed feelings about Razwell -- I tend to agree with him most of the time, but sometimes wonder about his mental health, as he sometimes claims some rather outrageous things. It surprised me when Wooo ended up banning him.

      Delete
    2. He is definitely not a normally citated authority. His trolling antics didn't do any good to his credibility , and he sounded irrational many times. So I was amazed how reasonable the comment I re-posted sounded. It is spot-on.

      Delete