I recently read something from the BMJ claiming to bust the myth that turkey makes you drowsy. And I thought that it would make a good example of the error of implied causation and how that is not what I’m talking about with homeopathy.
Then I realized that, at least for me, this wasn’t that kind of error: it was the error of believing what somebody told me was the science on the subject. I do not have the experience of turkey making me drowsy. I do like to relax after a big meal, but I don’t recall feeling especially sleepy. I usually don’t eat so much turkey anyhow, so maybe it’s just that. In any case, before I heard that turkey was high in tryptophan which makes you drowsy, I never would have said that. I might have said that eating a big meal makes you drowsy.
I’m glad that article put me straight. But wait a minute. Am I not just repeating the mistake? Before I heard that turkey makes you drowsy. Now I’m hearing that it doesn’t. When I read the article in detail it says that turkey doesn’t have higher levels of tryptophan than other meats and some other foods. It goes on to blame it on high consumption of carbohydrates possibly aided by wine. Nowhere in this article does it mention that somebody actually did the study on eating turkey. Maybe turkey does make many people drowsy, but it’s not just the tryptophan as the cause. Should I believe this article or not? Seems logical enough, but it also did when I was told the opposite.
Does science always progress towards greater accuracy, and therefore it’s safe to always take the latest data? And if so, then how much should I trust what is said now because tomorrow it may reverse itself?
I’m not saying that we should disregard the science, but I am saying that people should be skeptical of such things as well as “woo-woo” things. (And this is about something that doesn’t even have a lot of money at stake!) I don’t believe everything I hear from the SCIENTISTS up high. (They’re not really, but they are often treated that way in the media.) I’m skeptical of what I hear from them. It’s information that I will consider along with all the rest of what I know. Yes, the science matters, but so do other things. My key point: I am also certainly going to include my own experience in making judgments.
You might (or might not) call this a poor example, but in the medical field there are lots and lots of similar examples. Don’t get me wrong: I am not out to bash everything in medicine or science. I love science. There are some pharmaceuticals and techniques that I am personally very glad to have had available. What I am saying is that it’s not so black and white, with science as white vs. anecdote or intuition as black. I think that both have their place in the world.
Comments:
Of course you should not blindly trust scientists! If you ask any professional scientist they will tell you that there is an awful lot of nonsense in the literature. Further more you will find greater scepticism of a theory or result from those most closely associated with a piece of research than with most of the rest of the scientific community.
Some of the things you need to do to question scientific results are:
1. Put almost everything you know into a kind of “truth escrow account” neither believe or disbelieve it, that way you have no vested psychological interest in being right.
2. Always ask what question was asked what was compared and what was measured. Was the comparison made a sensible way to answer the question and was the thing measured a sensible point of comparison.
3. Ask which statistical technique, if any, was used to test the comparison and ask if the assumptions of that technique were met.
4. Think about what experiment you would do to test the result in a different way.
I’m a scientist, I’m often wrong. I do not want you to trust me I want you to be scientifically literate enough to question properly everything I do.
I agree wholeheartedly. This is my problem with most of the research that I’ve seen which claims that homeopathic treatment is no better than placebo.
It asks the wrong question: about a potentized substance not the homeopathic method.
It measures the wrong thing: does remedy X work on condition Y—that’s allopathy, not homeopathy.
It uses statistics in a manner that doesn’t fit the test. (A larger topic for another time.)
The researchers don’t seem to understand what they are testing, and so they don’t actually test it, but rather something that is sort of like it.
Therefore, their conclusions are wrong.
By the way, while I have gone beyond the question of whether or not homeopathic remedies “work,” i.e. are not just placebo, I do treat other claims I hear or read about in homeopathy with healthy skepticism.
The weird thing is that these are also the problems with apparently positive research published in the alt med journals. An example is a RPCT of arnica for post operative pain relief [1] the homeopathy in this paper may be bad but it can’t be as awful as the science. I may even start a blog just to write about this paper.
Anyway:
YOU DO NOT NEED TO US A SINGLE TREATMENT ON A SINGLE MALIDY 
to do a RPCT. You can just take a large number of random patients interview them and prescribe a treatment, they can then be randomly allocated into two groups, those that get what you prescribe and those who get placebo by the pharmacy and without your knowledge. You can use some sort of generic wellbeingness measuring questionnaire to track their progress. The important thing is that given large enough groups confounding factors ought to equal out so that only the effect of the homeopathic remedy comes through in the analysis.
You can even be cleaver and note that you would expect the placebo group not only to fair worse than the treated group but also to have their remedy altered more often as the homeopath struggles in vain to adjust the remedy in the face of little or no improvement resulting from the lack of real treatment. So you can set up two statistical tests:
1. Does the placebo group have lower wellbeing than the treated group?
2. Does the placebo group have a greater number of changes of prescription than the treated group.
This kind of thing is hardly a regulatory study, but you are light years away from that kind of thing. You need to build a base of evidence from a variety of experiments to convince us sceptics and you can worry about more advanced things later.
Also stop demanding that scientist do the work, we are busy doing our own things. It is the homeopaths that have an interest in homeopathy and it is up to them to get the data. In addition it is a disgrace that you are treating the general public without the proper evidence to know you are actually helping them.
1. Homeopathic Arnica montana for post-tonsillectomy analgesia: a randomised placebo control trial
Homeopathy, Volume 96, Issue 1, January 2007, Pages 17-21
A. Robertson, R. Suryanarayanan and A Banerjee
I find the fact that some of these studies show results to be very significant, because the deck was stacked against homeopathy due to the methodology. The fact is that many of the remedy X for condition Y studies do not show a result. Some do. I depends on the condition and how well the single remedy is selected.
There is a new one of these studies with a positive result in the journal Chest. They have moved on to a phase II trial.
I am not demanding that anybody do anything. I do prefer, however, that people stop criticising homeopathy using studies which were not designed to actually test homeopathy.
Some studies have been done that aren’t single remedy studies, with good result. More of both kinds of studies are being done, as I understand it. I am not in the research business. Nor am I interested in selling homeopathy to people who are skeptical. I only want to try to provide straight information and counter the misinformation and let people decide for themselves.
There are several problems with this.
The first is that you believe there isn’t proper evidence that homeopathy is helping people. I disagree. I say that you dismiss evidence because you don’t like the implications of it.
Another is that this is the state of medicine. The majority of treatments that doctors use are not directly backed by studies, but are in fact anecdotal. Most are off label uses of pharmaceuticals.
Most importantly has to do with freedom. If what I am doing is not intrinsicly harmful, I have the right to do it and people have the right to access it. I do not commit fraud by making promises of cure, only to give the best homeopathic treatment that I can. And I do in fact help a majority of them. If I did not, I would not be doing what I do.
Yes, you have mentioned this before. A very interesting point. I would be grateful if you could post a reference to the original source of that information.
Darn it! I can’t find my orgininal notes where I heard about that study. Still, if I just use Google…
A Standford study came up with a 21% number in 2001. Certain types of drugs have much higher numbers.
According to this article, the AMA estimated it at 40%.
And this group cites sources which give numbers from 25% to 60%, with an 80% number for children.
These are only for pharmaceuticals, not all treatments. The source I want to cite was considering all type of medical interventions, not just the percentage of prescriptions.
In any case, without my source I’ll reduce my “majority” to “commonly.” By the way, I’m not condemning medicine for this. It’s not illegal for doctors to do this, only for pharmaceutical companies to advertise off-label uses. All I’m saying is that doctors have the freedom to use their experience. And I believe this is an essential freedom. Like most things, sometimes this is bad, sometimes it’s good, but hopefully more often good than bad. I do think, however, that doctors probably ought to disclose when they do this.
John, The Chest paper is very interesting. I’ve only skim read it and I want to give it careful consideration befor I respond.
Also Woodchopper would you have a look at it. I have to say I can’t see anything obviously wrong with it at first glance. Looks like it was properly double blinded, looks like sensible things were measured, three of the five things subjected to statistical testing gave p-values < 0.0001! so its not like one of those flaky studies with p-values ~ 0.045 in 1 of 20 things tested.
Derik, John
I just saw the reference to the Chest article. Indeed, I think its the first I have seen in which the homeopathic group actually displays a significant difference to the control group. (As you say, its in a completely different league to all the ‘revolutionary paradigm shifting’ studies with a p-value of 0.5).
I’m not a doctor so I can’t evaluate it properly. But the only issues I could see were the small sample size (25 + 25) and that the placebo control group appeared to be marginally iller at the start of the trial.
But still, this is an interesting study. If its replicated with a much larger sample then people should take it seriously.
In the comment above, I obviously ment “As you say, its in a completely different league to all the ‘revolutionary paradigm shifting’ studies with a p-value of 0.05).”
Post Your Comment: