Sunday, 20 April 2008

Homeopathy and wikipedia

Homeopathy-related articles on Wikipedia have been subject to probation for some time, because of edit wars between pro-homeopathy and anti-homeopathy contributors. The situation has become such a shambles that Wikipedia has opened an arbitration case on all its homeopathy-related pages. As a result of his editing behaviour, Dana Ullman has managed to get himself banned from editing any homeopathy-related pages for a period of three months.

The arbitration page is long and parts of it are somewhat tedious, but it is interesting in that it demonstrates some of the problems science has in presenting evidence to laypeople in a clear way. The literature on homeopathy is complex, consisting of a mixture of poorly conducted studies that show some (at best equivocal) evidence that homeopathy might work, and well-designed studies that show homeopathy has no effect. (A meta-analysis by Shang et al. that was published in the Lancet elegantly showed that the largest and least biased studies showed the least effect for homeopathy). It is then easy for advocates of homeopathy to emphasise the apparently positive studies, and claim that sceptics of homeopathy do not consider all the evidence. In other words, advocates of homeopathy take the studies at face value; they don't look at the studies and analyse them to see how they were conducted and whether the results make sense.

Because there are some people who think homeopathy works, and there are some peer-reviewed journals of homeopathy and other 'alternative' therapies that superficially appear to be scientific, it is possible to claim that there is some scientific controversy about the effectiveness of homeopathy. As Wikipedia is based around the idea of neutral point of view (NPOV), it seems superficially sensible to present a 'balanced' view whereby evidence for and against homeopathy is described. But, crucially, this is not a balanced view; it gives undue weight to poorly conducted studies. Looking at the evidence as a whole shows that homeopathy is nothing more than a placebo. Quite apart from that, it contains ideas (e.g. that the 'potency' of a remedy increases the more it is diluted) that are in conflict with well established science. A neutral point of view is surely that homeopathy is extremely implausible and that there is no evidence that it works.

This sort of thing is not just a problem for Wikipedia. There is a general problem of spurious 'balance' in scientific reporting. Another clear example is the BBC linking to anti-vaccine websites in stories about vaccines. Science is a notoriously complicated business, but there are times when the scientific evidence is clear, when it is evaluated properly. The challenge is to show that this is the case, without coming across as some kind of monolithic establishment that wishes to crush all dissent.


Joe Dunckley (aka Steinsky) said...

I've not been following this, and don't edit Wikipedia much these days, but Wikipedia's policies actually have something going for them.

Admittedly, the best route would be saying: this the good science, the end -- it would be best not to perpetuate the manufactroversy through an encyclopaedia. But having to take the he-said-this/she-said-that approach of reporting everything and attributing every claim to a reference or quote should make the pro-homoeopathy side look more ridiculous than just reporting the science: you get to quote somebody important pointing out just how fundamental their misunderstanding of the methods and results of science is. After all, if it's relevant and supported by the references, it fits Wikipedia's inclusion criteria.

Paul Wilson said...

I think you have a point. The only thing is that with the rise of CAM journals you can find a supporting reference for pretty much any daft statement you wish to make.

I do very much like the idea of Wikipedia, but I think it struggles to handle situations like this, where marginal papers can be cited and confusion can be sown by people who have an agenda.

Of course, it isn't just Wikipedia that has this problem: it is no better or worse than the mainstream media.