Large Experimental studies
A quote from a commenter on this post:
A major problem in alternative medicine studies is that small sample sizes produce positive results, but larger sample sizes produce negligible results. In science-based medicine, if larger studies show a medicine to be ineffective, but smaller studies show a medicine to be effective, we conclude that the smaller studies were flawed due to small sample size.
I'll let you guys comment on the rest of the taradiddles you see above.
Yes, alt treatments get by without having to do such large studies (which unfortunately can provide false positives--you can get a positive result for almost anything by making the study large enough). However, that helps to keep the cost down. Herbs and supplements are far, far cheaper than most prescriptions. If we're trying to keep healthcare costs down, do we really want everything to be uber-regulated and uber-expensive, especially innocuous things that have been used safely for a long time?I am confused as to how someone could come to the conclusion that the larger one's sample size, the more likely it is to provide positive results for "almost anything". The exact opposite is true, I.E. the smaller the sample size, the more likely one can provide positive results for "almost anything".
A major problem in alternative medicine studies is that small sample sizes produce positive results, but larger sample sizes produce negligible results. In science-based medicine, if larger studies show a medicine to be ineffective, but smaller studies show a medicine to be effective, we conclude that the smaller studies were flawed due to small sample size.
I'll let you guys comment on the rest of the taradiddles you see above.
Labels: alternative medicine, biases, blogging, fallacies
posted by Ziztur
at
3:00 PM
3 Comments:
I imagine the commenter means that if one has a null hypothesis of no difference, with a large enough study you can probably gather enough evidence to notice a difference, which is possibly true, however it is unlikely that you will gather enough evidence to note a climcally significant difference.
Imagine some sort of treatment which is, in absolute terms, effective on 1% of the individuals it is used on (for whatever definition of "effective" you want to use, the specifics don't matter for my example).
For a sample size of 1 individual, it's probable that it will return a result of 0% effectiveness - but there is a 1% chance that it will instead return a result of 100% effectiveness. For a sample size of 2 individuals, again, most likely it will return a result of 0%, but there's a chance it will turn up 50% effective. 3 individuals, probably turn up 0%, but can also turn up 33%. For any sample size less than 100, it will generally turn out to register as either 0% (less than actual effectiveness) or more than 1% (more than actual effectiveness).
For extremely large groups, there will almost always be at least several people affected, no matter how marginally effective the treatment. In a 100,000 person study, a 1% effective treatment will be effective on about 1,000 people. I think that's Andrea's point: for small groups, it's most likely to register as completely ineffective, while for large groups there will always be at least some individuals affected.
So I think can see Andrea's point. She needs to take a statistics course (she thinks the question is "Was the treatment effective on any individuals?", when in fact the question is "What proportion of individuals was the treatment effective on?" Even the most effective treatments known to man aren't 100% effective), but I can sort of see her point.
As for some of the rest of the taradiddles: herbs and supplements are cheaper than prescriptions because you get what you pay for (to an extent).
Real medicine is expensive because it needs to run real tests with enough subjects to yield a statistically significant result in order to prove it is effective. Alternative "medicine" is cheap because it doesn't need to prove anything, it can just stick any crap in a jar and sell it to unsuspecting dupes.
If we want to keep healthcare costs down, we could use the stuff that doesn't work. Or we could do absolutely nothing at all to treat health problems. That costs nothing*! Hurrah, we've solved health care! Let all people rejoice!
* Most alternative medicine, even homeopathy, does not count as "doing nothing", even though it has much the same effect as doing nothing.
Post a Comment
I will never delete a comment because I disagree with you, but if you're posting anonymously, at least give us a name so that if you make multiple comments we can tell you apart from the other anonymous people.
<< Home