Most charity funders refuse to pay these - also the Medical Research Council as of this year stopped people putting open access fees explicitly into budgets (even though they mandate open access). The universities have to find the money themselves.

Medicine. I agree it's less open that many disciplines. Like I said, I think open access is generally a good thing. But in my vast experience, people actually doing research genuinely don't care, as they know that people at other universities will be able to read their work whether its open or not.

As an aside - a lot of universities are rejecting the 'Gold' open access standard (the author pays version) because it is horrendously expensive for authors (usually 1000-2000 per article). They are instead preferring the 'Green' open access model, where the journal keeps the copyright to the final copyedited version, but lets researchers distribute their own version on a personal or institutional website. This is probably the way of the future because we can't keep paying stupidly high open access publishing fees.

Well of course. As I said its about impact factor. If and when open access journals get decent impact factors, researchers will be more inclined to use them.

Well it is anecdotal, but I've been in literally hundreds of 'which journal should I send my paper to' discussions (I've been doing this a long time), and the factors that come up are (in this order) (1) impact factor (2) readership, ie which society is the journal affiliated with (3) likely success (4) cost of publication. Nobody has ever once said to me "I want to send to journal X because they are open access".

I think most would agree in principle that open access is a good thing, but when it comes to having your work seen, read and acknowledged by the right people it completely goes out the window. This is medical research btw, different fields may differ.

Researchers don't generally care about their papers being open access or not. They use open access journals because they are easy to get published in (they are mostly 'author pays' publications with very low standards) or because their funder mandates it.

Humans get old which is extremely unusual in nature.

It would have been enough for the subject's family to have posted the genealogy information - the subject may have known nothing about it. Still, you are right its not the fault of the researchers (as its impossible to fully anonymise a dataset while retaining its research usefulness).

Works out pretty well for the tobacco industry.

I'd also like to judge people on their ability to think, to listen to others, research existing knowledge, to appraise and weight up ideas, and this is a large part of what college teaches. This goes beyond 'knowing stuff' and 'people skills' (although these are undoubtedly important).

Somewhere there is an astronomer with his telescope the wrong way around trying to work out the orbital trajectories of the remnants of his lunch.

You could make them post an apology on the front page of their website.

Its a good point - I did wonder about this. I suspect there's a kind of bias going on whereby any result other than that with the face validity of this one would not have been published (the authors would have discarded it on the basis of their poor method). It's not possible to know how many combinations of searches and calculations were done before they arrived at this one. If that sounds cynical then it is, but I'm an (honest) statistician and I do have to work very hard to stoo myself falling into that trap sometimes. On the other hand, they could just have lucked out, or the sheer volume of positive tweets eminating from the blue states could have been resposible for the relative lack of racist tweets in those areas, which would have produced the same result regardless of possibly differing underlying levels of racism across states.

I can't believe how much of my weekend I've wasted thinking about this.

Can you show me any situation where less than 400 samples is enough to estimate fixed effects across a variable with 50 factors? This is in fact what was done, whether on not the guys doing the calculation realised it.

You are right but this means that the peer review filter is even more important so that what gets out to the media and beyond has at least some chance of being right. Also, having been through the process a few times I'd say academics are at least as guilty of overstating their findings as journalists. We want the headlines and the 'impact' as much as journalists to.