Thursday, 30 August 2007

More on Rao et al. homeopathy tomfoolery

If anyone out there has read the stuff on this blog (here and here) about the recent special issue of the pseudojournal Homeopathy dealing with the memory of water, and the Rao et al. paper in it, they'll know there are some serious problems with the paper. If you're really interested in finding out just how rubbish this supposedly peer-reviewed paper is, I recommend the discussions of it on the JREF forum. Now Rolfe, one of the contributors to the forum, has drafted a letter to Peter Fisher, the editor of Homeopathy, pointing out the errors and problems with the presented UV-vis data. This includes my observations of graph tomfoolery, as well as many other points picked up by JREF contributors Rolfe, JJM and Pipirr. We haven't made any comment on the Raman spectroscopy data, which seems to be equally problematic (see here, look for comment No. 11). The letter is available here.

Assuming there's any response at all, watch this space...

Invitation to a debagging

I've been having a good laugh at the recent rash of media stories referring to Manchester's murder triangle (e.g. this one in the Grauniad). If you look at a map of said triangle, we live pretty much on the southern edge of it. There are bad areas around us in Longsight, Moss Side and Hulme, but Rusholme itself seems to be pretty safe. I am not among the one third of adults who are supposedly afraid to go out at night, though I was a bit worried when Jolan took us out for a beer at the Claremont in Moss Side. I've also tended to be somewhat sceptical of poor useless David Cameron, who has recently been flapping his soft, pudgy face on the subject of crime.

Well, I got shocked out of my complacency (at least a little bit) last night, when I got mugged in broad daylight, outside a pub and within yards of a busy main road. As I walked home from work, I took my usual short cut along Rusholme Grove past the Welcome pub. A young Asian lad asked me if I wanted to buy a mobile phone. I sure as hell didn't want to buy anything from some guy in the street I'd never met in my life, so I said no thanks. The kid kept following me and badgering me about his damn phone, which was when I started to realise that something here was not quite right. A bit too late, unfortunately, as the kid grabbed my shoulder bag, and tried to pull me towards him, saying 'come here, you dickhead'. 'What's the matter?' I asked, 'What's it all about?' Funny what you say in such situations, eh? No explanation was forthcoming, with the kid continuing to call me a dickhead and try to pull me towards him. At this point I noticed he had a friend with him. Reasoning that there was nothing in my bag except a copy of the Guardian, I got the strap over my head, and legged it. I ran straight home (all of 500 yards), and called the cops.

I must admit, I didn't expect much from the police, who I imagined had more pressing concerns on the notorious murder triangle beat, etc. I was quite wrong. About half an hour later, two plainclothes cops came round to the flat, and took me on a drive around the hood in an unmarked car, looking for the 'offenders'. Naturally we didn't find them, and I went to Greenheys police station to file a statement. The police were very professional, very thorough, and I couldn't help be impressed by the way they went about their job, which can't be easy (the murder triangle, and so forth). Anyway, I got a lift home in a police van. I got let out outside the flat and thanked the cops in the van, which then peeled off up the street, blue lights flashing. Another busy night for the guardians of law and order in Manchester's notorious etc etc.

I decided it was time for a beer, so Jolan and I went to Hardy's Well and played backgammon. In this neighbourhood, the excitement never ends. When we got home, there was a message on the voicemail from the police. They had recovered my bag, and it had been sent for forensic examination. I could have it back when they'd done with it. Again, I have to say I was filled with admiration for the police, who surely had more to worry about than a bag I got for free from a conference with a copy of the Guardian in it. I'm happy to get it back though. After all, Long Beach CA is a long way to go for a bag.

I was certainly a bit shaken up, but I'm sure I can take the same route home every day for the next twenty years and not have the same thing happen. Unfortunately, there are idiots everywhere you go in the world. I am going to be a little more careful, though, and it does make me wonder if the area is as safe as I thought.

Thursday, 23 August 2007

Blue vs. pink

There's a study by Anya Hurlbert and Yazhu Ling in the latest issue of Current Biology (volume 17, No. 16, p. R623) that purports to show that women prefer the colour pink compared to men. The authors explained this as a result of hunter/gatherer era distribution of labour. Women did the gathering and thus needed to pick out red berries from a green background, while the men were out killing stuff. This was picked up widely in the media (for example, there's a Times article here). As you might imagine, it has also caused some controversy, with the Grauniad's Zoe Williams imploring us to 'Stop this idiocy now!'. Nonetheless, I don't get the impression that anyone has actually read the paper. There certainly appears to be no considered evaluation of the research methods employed in the study in Williams's diatribe. So, using my academic ninja skills, I looked up the paper and read it. I recommend this approach to any science journalists.

The research methodology of the paper seems to be sound. Essentially, the authors of the study got a group of men and a group of women, a total of 208 people. The subjects "used a mouse cursor to select, as rapidly as possible, their preferred color from each of a series of pairs of small colored rectangles presented sequentially in the center of an otherwise neutral CRT display." The results were quite striking, with women clearly having a preference for pink hues with respect to men. Interestingly, the differences for a Han Chinese sub-group were much less pronounced. Anyway, so far so good. Conclusion: Women like pink hues more than men do.

So why do women prefer pink? The authors write 'We speculate [my emphasis] that this sex difference arose from sex specific functional specializations in the evolutionary division of labour'. They go on to propose that the difference might relate to women gathering red berries, or the need for women to notice subtle changes in skin colour in 'their roles as care-givers and empathizers'. As evidence for this speculation, the authors employ something called the Bem Sex Role Inventory, showing that masculinity and femininity correlate with colour preference. What on Earth is the Bem Sex Role Inventory? It turns out that this works by asking the subject how they rate themselves against a list of 60 adjectives. 20 of the adjectives are 'desirable for men' (e.g. assertive, independent, analytical), 20 are 'desirable for women' (e.g. loyal, warm, shy) and 20 are 'gender neutral' (e.g. happy, tactful, jealous). People are asked to rate themselves for each adjective on a 7-point scale. Clearly, as scientific methodology this leaves a lot to be desired, and it seems to me that this is just testing cultural constructions of masculinity and femininity. I'm not convinced that it backs the authors speculation that innate biological factors account for sex differences in colour preference. In fact, the authors are quite careful about this, and state that 'while these differences may be innate, they may also be modulated by cultural context or individual experience. In China, red is the color of ‘good luck’, and our Chinese subpopulation gives stronger weighting for reddish colors than the British'. This doesn't seem to be reflected in any of the media stories about the research, which is what happens when you get reporting by press release. Anyway, in my opinion, the research can't distinguish between innate and cultural explanations of the observations, and the stronger preference for red hues among the Chinese subjects certainly seems to suggest that cultural factors are important.

So, what have we learned? Well, we've learned that women prefer pink hues in comparison to men. But I don't think we've learned why. So, when Williams says 'Humanity has nothing to gain from research into whether females prefer the colour pink', is she right? I don't think so. We've learned something, and for that reason alone I think the research is valuable. And if we can say something definite about the reasons for the differences found in the paper, we can say something about our society and why it works in the way it does.

Friday, 17 August 2007

Very bad science...

I already wrote an amateur rebuttal of the Homeopathy special issue on water memory here, but I was following the discussion about the Rao et al. paper at Ben Goldacre's site, where the articles are being journal clubbed. A contributor to Bad Science, RichH, noticed that some of the curves in figure 2, representing 'envelopes of difference' for different dilutions of two different homeopathic remedies, looked similar to the 'representative' examples of curves shown in figure 1, purporting to show that UV-vis spectroscopy can identify different homeopathic remedies.

I did some Corel Drawing, and resized and stretched the graphs in figure 2 so that the scales between figures 1 and 2 were the same. It seems that RichH was correct, and the curves in figure 1 are also shown in figure 2. This suggests that curves the authors compare in figure 1 are the most extreme spectra obtained for the different remedies, rather than representative ones. Most surprisingly, though, the part of figure 2a that shows the 'envelope of difference' for 30C Nat Mur is identical to the graph in figure 1 that supposedly shows the difference between 30C Nat Mur and 30C Nux Vom. In other words, the same graph is presented twice, and purports to show different things each time is presented. Apart from all the errors discussed here, it seems they got their graphs mixed up too. This seems like an extraordinary error to make in a formal paper, even in a pseudojournal.

I reproduce the part of figure 2a (at left), resized and stretched so scales match, next to the part of figure 1 (at right) below. It's easier to see that they're the same if you put them on top of each other, in different layers, in Corel Draw and turn one layer on and off. But hopefully you can see that the curves are the same when presented side by side. The only difference is that the symbols (open vs. closed circles) are swapped between the two graphs.

Of course, you shouldn't have to go through this rigmarole to compare graphs in the first place...

Edit: Philip Ball has picked this up on his blog. If you go there, you can also find links to his original Nature piece and the various comments on it at the Nature weblog. Ben Goldacre has also picked it up on his mini-blog. If I'm not careful, all this fame will go to my head...

Monday, 13 August 2007

Biking over mountains for fun

This blog is becoming far too serious, again, so it's time for some Pennine punishment. On Saturday I got up at 8:30 just so I could do some proper cycling. I was on the road by 9:15, and headed out through Longsight, Belle Vue and Gorton (all the sights), into Hyde and over to Mottram via Godley and Hattersley. At Mottram I took a left, over the hill and then a fast and long descent into Stalybridge. Then out towards Greenfield. A section of the main road was closed here, so I had to follow a diversion that takes you all the way to the valley bottom, before you climb steeply through a housing estate back to the main road. I'm unreasonably annoyed by this, as I know the ride already has plenty of hills. At Greenfield the ride really begins, as you begin the climb past Dove Stones Reservoir onto Saddleworth Moor. This is a longish climb, but not that steep, and I was able to get into a low gear and plod my way up easily enough, though drenched in sweat as the sun came out. You get good views of Dove Stones behind you and to your right before you get onto the featureless moor top.

Then a very fast and long descent into Homfirth, where I stop to have an apple. I'm in West Yorkshire now, and on roads I used to train on when I raced relatively seriously. In Holmfirth, you're really out of good options. You're in the valley bottom, and whichever way you go to get back to Manchester involves some serious climbing. I'm out to prove to myself that I'm fit, so I'm going over Holme Moss. Holme Moss is the highest main road in England, and it climbs about 400 m in 5 km or so. I once rode a 25 mile mountain time trial that started in Homfirth, and I went from Holmebridge to the summit in 22:15. This is not going to happen today.

I get up the lower half of the climb into Holme village OK, and you get a brief respite through Holme and Lane, and even a short downhill before the serious part of the climb begins. There's a series of hairpins, followed by the last gruelling section to the road summit. I run out of gears far too early, and I'm barely making walking pace by the last section. Every quarter of a mile the distance to summit is painted on the road, and I'm already suffering as I pass the 1 mile to go marker. Goddamnit, I grit my teeth and I get to the summit somehow. Beautiful views over West and South Yorkshire at the top, and then the steep descent to the Woodhead road.

This bit is narrow and lacking in fun as the trucks fly past, and I'm out of water. I head through Tintwistle and into Hollingworth, and start the climb into Mottram. The traffic here is solid, waiting for the lights at the top of the hill, and I ride down the middle of the road. Past Mottram there's a filling station, where I get some cold water down my neck and rest for a couple of minutes. Downhill into Hyde, then back through Gorton and Longsight, and I'm home at 1:30. The ride is only 80 km or so, but I reckon it makes up for it in climbing.

Friday, 10 August 2007

A personal perspective on industry funding

Prof. David Colquhoun makes some comments about the increasing proportion of research funding coming from industry here. First, a declaration of interest: my research funding at present comes from a consortium of oil companies, namely Norsk Hydro, Statoil and ConocoPhillips.

Clearly, industry funding can lead to conflicts of interest, especially where the research is into the efficacy of a product manufactured by the funder (e.g. in drug trials). But this isn't necessarily the case. In the work I'm doing, the companies are interested in finding out something about the geology of rift systems, so they can apply that knowledge to the assets they have with similar geology. In this case, the companies have no interest in influencing the results. They have an interest in us conducting good science, because that will help them recover more petroleum. In return, we get access to the best quality data, and publish (hopefully) good basic research on the evolution of rift systems. We also get our wages paid. Everyone's a winner. It's possible to envisage a situation of conflict of interest if we were involved in, say, estimating reserves, because that might have an effect on the share price of the funding company. But as a rule, this is not the case.

Of course, it isn't all good news. Companies are sometimes cagey about releasing data for publication. This can be a real problem. During my PhD, a company that had provided seismic data for me to work with (but did not fund the project) did not want the data published. In fact, they were unhappy about me even publishing my interpreted line drawings of the data. I eventually managed to get around this and get the work published, but the reviewers were not particularly happy that they could not independently evaluate my interpretations against the data. Unfortunately, this situation is so common in petroleum geoscience research that work often gets published without anyone getting to look at the original data. Obviously, this is not the way to get good science. I work at a university, a public institution, so I think that any work I do ought to be publicly available. Where projects are funded by industry, this is not always (or even usually) the case.

The other problem relates to what kind of work gets done. Clearly, industry is only going to fund projects that it has an interest in. Hence here at Manchester the petroleum geoscience group is awash with money, but other groups doing research that is no less interesting struggle for funding. Again, a university is a public institution, and it ought to work on behalf of the public. Working on behalf of industry is not always the same thing (although sometimes it can be).

Industry funding is not necessarily bad in itself. But we need to be clear about its drawbacks. I'm not convinced these drawbacks have been carefully considered in the rush to increase industry funding.

Thursday, 9 August 2007

Pseudoscience and impact factors

Just out of interest, I looked up the journal 'Homeopathy' (see below) in ISI's Journal Citation Reports. It isn't listed, which cheered me up a bit. However, the Journal of Alternative and Complementary Medicine is listed, and has a impact factor of 1.104. In comparison, the Canadian Journal of Earth Sciences, a solid and respected regional earth science journal, has an impact factor of 0.943. So perhaps it would be a better career move to try to publish some pseudoscience that purports to prove the efficacy of homeopathic treatments, than to try to produce a solid and useful piece of earth science research.

I'm sure this is not the intention, but including such pseudo-journals in these rankings has the effect of making them seem respectable. After all, the numbers don't lie, right?

Wednesday, 8 August 2007

A cracking example of a pseudojournal

It turns out there is a journal called Homeopathy. It looks very like a proper journal. It is peer reviewed, it is indexed in the Web of Science, and it is published by Elsevier. If you look at the page layout, it looks not unlike the Journal of Structural Geology; it has that journal format to it. But this is definitely not a scientific journal. It is a pseudo-journal that publishes pseudo-science, and its reviewers are pseudo-scientists.

The journal has recently published a special issue on the memory of water (as so often, Ben Goldacre's site is responsible for the heads up). This is an important topic for homeopaths, because it provides a way in which homeopathy might work. The water that contains the homeopathic 'remedy' is diluted such that none of the 'remedy' can possibly remain, so homeopaths fall back on the notion that the water somehow structurally 'remembers' the remedy. Some proper scientific work that got published in Nature suggested that the actual length of time water can structurally remember anything is on the order of 50 femtoseconds, so this is probably a non-starter. Those brave homeopaths have taken on the blinkered scientific establishment and had a go, nonetheless.

Oh, it's a shambles.

In it, Martin F Chaplin has a paper entitled 'The Memory of Water: an overview'. But this paper contains no evidence of water having a memory at all. Chaplin's examples of 'evidence for water memory' have nothing to do with water 'remembering' a solute that has been diluted out of existence. For example, 'human taste is quite capable of telling the difference between two glasses of water, processed in different ways (eg one fresh and one undrunk for several days.)' Yes, that’s because they will contain different amounts of dissolved gases from the atmosphere. What would be really interesting is if you could taste the difference, say, between two 30C dilutions of different solutes prepared under identical conditions. Chaplin also mentions a 'memory effect' related to clathrate formation. Again, this isn't relevant to whether a solute that is diluted out of existence has any effect on the water structure. And again, there is a mention of 'slow equilibration' of solutions. But this isn’t relevant to cases where no solute is present.

Then there are the 'experimental papers'. The paper by Elia et al. is so poorly organised and written that it isn't clear what they did. However, they do say 'It is important to emphasise that, from the studies so far conducted, we cannot derive reproducible information concerning the influence of the different degrees of homeopathic dilution or the nature of the active principle (solute) on the measured physicochemical parameters'. What this seems to say is that none of the results they present are reproducible, so they can't really be considered to represent results in any scientifically meaningful sense.

The paper by Rao et al. seems to have some serious problems with the reported spectroscopy, which is dealt with here. My personal favourite section is where the paper suggests that 'it can be argued' that the succusion process (shaking of the homeopathic remedy between dilution steps) could produce pressures of up to 10 kbar. Perhaps this point can be argued, but the authors don't make any attempt to argue it. They also state that grinding in a mortar and pestle can produce pressures up to 20 kbar, for which they give a reference to a paper by Dachille and Roy, which is number 22 in their reference list. In fact, their reference no. 22 is to a Bates et al. paper in Science, which looks at high-pressure forms of Germanium. The paper contains no mention of mortar and pestle at all. This sort of bad referencing would be picked up by any thorough review. But anyway, 10 to 20 kbar? Now, as a geologist, I know that pressure in the crust increases by approximately a third of a kbar per kilometre, depending on rock density. Thus 10 kbar is equivalent to a depth of 30 km, and 20 kbar is equivalent to a depth of 60 km. So if I shake up a homeopathic solution, I can get a pressure equivalent to 30 km of rock, and if I grind up some shale in a mortar and pestle, I end up with a blueschist facies metamorphic rock. 10 kbar is on the order of 10,000 atmospheres. This claim seems extremely unlikely, and again is the sort of thing that would be picked up by any half-way competent reviewer.

The paper by Louis Rey seems interesting, but there are issues with reproducibility. None of the graphs contain error bars. Although the author says that his results have been replicated by another lab, he doesn't seem to have tried to replicate them himself. It is just stated that some graphs look different, without any discussion of how statistically significant the differences might be. Again, it's difficult to know whether the paper shows anything at all.

Then there is a paper by Vybiral and Voracek, which looks at some interesting physics of water. However, the authors conclude that the 'autothixotropy' effect that they observe is absent when they use de-ionised water. Thus their effect is related to ions in the water, and not any 'water memory' effect which would be due to ions that were previously in the water, but are no longer there.

After that there are some 'theoretical' papers, which contain no new evidence, and some bizarre excursions into the world of quantum mechanics, which luckily are dealt with here.

In short, the entire issue seems to contain no good evidence in support of its title 'The Memory of Water is a Reality'. And yet this will now be cited by homeopaths for evermore as peer-reviewed scientific evidence for how homeopathy might work. Of course, all the evidence suggests that homeopathy doesn't work at all, so there's no need to explain how it works in the first place. What a waste of time.

To give Homeopathy its due though, it did publish a paper by Jose Teixeira, entitled 'Can water possibly have a memory? A sceptical view'. The paper concludes that 'any interpretation calling for ‘memory’ effects in pure water must be totally excluded', on the grounds that 'the longest life of any structure observed in liquid water is of the order of 1 ps'. So much time and effort could have been saved.

Edit: There's a far better overview than the above somewhat amateur attempt at Philip Ball's blog, which is why he writes for Nature and I don't. I can't link directly to the post itself, but look for the August 03 2007 entry entitled 'A bad memory'. There are also comments from Martin F Chaplin there.