Zum Seitenende Übersicht Artikel Home & Impressum
First the link to this week’s complete list as HTML and as PDF.
If there is one point I have been keeping banging on and on and on about here all over the last months and years, it is the responsibility of co-authors. Just now an especially blatant example has come to light.
Last December science published a prominent article by LaCour & Green. I didn’t pick it up at the time and didn’t comment on it, but apparently it raised waves among the political science community. The slow and tortuous route to its debunking is described in detail by Singal. The editor and referees are not to blame, the data were an outright and complete fabrication and hints leading to that fact subtle and far from obvious. The role of the author’s supervisor is unclear. Perhaps she might have kept a closer look on what he was doing, but she was not involved with the paper as such. The only remarkable point about this whole affair is the role of Donald Green. He had been the main author’s undergraduate adviser at an entirely different university thousands of miles away and had nothing whatsoever to do with the supposed research. The only thing he’ll probably have known about it was what was obvious to everybody else being shown drafts at conferences:
“The results LaCour showed Broockman were, in fact, very cool, and like everyone else who had come across them, Broockman instantly knew they would be a hit.” Given the chance to add another jewel to his crown without expending any effort, Green jumped at it. But when the thing blew up, all he had to say was:
“I am deeply embarrassed and apologize.” In his interview he even had the gall to complain:
“Naturally, I resent being put in this awkward position for no reason.”
This is totally insufficient. Everyone is free to decide, either you are an author or you decline. If you are, you take full responsibility, moral, scientific, and legal, for everything in it. Here’s why. On seeing the results everybody in the field instantly knew they contradicted just about everything published before. True, they conformed to what many wished for and would have liked to believe being the case, but still an unknown graduate student would normally have been asked for more proof than that. But the well known and highly trusted Prof. Green’s name on the paper clinched it for many, him they trusted implicitly. To a large degree this race to add ill checked publications to one’s list is what makes the current level of fraud possible in the first place and the real perpetrators must take the blame without passing it on to scapegoats. That’s the whole point about graduate students not being fully fledged scientists, they have yet to prove themselves, morally as well as professionally. You can only be burdened with the blame after having given the full responsibility before. LaCour might have published on his own, but he was fully aware of what Green’s name would do for the paper – and so was Prof. Green.
Two more points. As Singal tells it:
“Moreover, the very few plum jobs and big grants don’t go to people who investigate other researchers’ work – they go to those who stake out their own research areas.” To paraphrase: careers and success in the political sciences go only to those, who eschew and refuse to work by scientific principles. That explains a lot.
Secondly I notice that LaCour does not place his response on his own webspace, where it may be collected and archived by robots, but on the highly volatile Dropbox server. This Orwellian mode of publishing can also be seen in political parties, who in spite of running large websites themselves place controversial stuff in disreputable and unstable places like Youtube. Those who make private copies are persecuted by copyright law like Winston Smith and his salvaged scrap of paper. We should all be aware of these shenanigans and draw our own conclusions from them.
I don’t understand the premise in Halevy & Halali at all. In a conflict situation the conflict happens, when one side decides to start it, irrespective of the other side’s preference for peace or not. After it started, one side usually wins and the other loses, but beforehand you don’t know which. In both the chicken and the prisoner models the side starting the conflict is always the one that wins, except when they both start, where none does. To me this looks like a totally different situation altogether and thus irrelevant.
In practical terms the promise of help to the passive loser and punishment to the active winner may entice the weaker side into provoking the stronger one to attack. There are good arguments that this is just what Poland did in the summer of ’39. Contemporary Polish propaganda showed a future border near the outskirts of Berlin.
And again in Kirk we get some data for the independent variable in figure 1 and none whatsoever for the dependent one. The weak trend may be real, but does it have any practical predictive value at all? It seems weak enough to be amenable to explanation by confounders, like parolees bent on taking up their criminal careers again actively seeking concentrated neighbourhoods even against circumstances making that difficult. The parolees dispersing, now that became possible, may be the very ones who’d not have reoffended irrespective of neighbourhood.
Marcia McNutt talks about less than helpful references for women and accounts for them as thoughtlessness in both male and female referees. May be so, but that’s certainly not the only possibility. Given that many women too were accorded the good qualities of creative, hard-working, and showing leadership, etc., is it not just possible, that some of them were lacking those? There is an open, active, and affirmed policy towards attracting women into STEM departments, so isn’t it just possible, that more women than men lacking those traits reach positions from which to apply for grants and fellowships? If so, what is the referee to do? Is he to lie, to highlight the candidate’s deficiencies, or to stress those positive attributes the candidate does have?
In their figures 1 e–g Mina et al. do not just show us the trends and regressions with their error margins but also all the raw data they are based on. Why do I never ever see the same in the psychological and social sciences?
Tack et al. is another article only showing us highly derived modelling outcomes but (including supplementary materials) no primary data at all to rate the value of those models. What does stand out from the data given is that the year 2013, exceptionally cold in spring and fall, shows exceptionally low yields while 2012’s yields, which was quite hot in spring and winter and average in fall, come out as average. In none of the yearly data shown can a falling trend over temperature be made out. Not that this was conclusive, because, as said, the relevant data and diagrams are withheld.
I don’t quite see the value of Higgins et al. Neither the age nor the duration, climate or climate variability can be constrained to meaningful limits. All we can say is, that values from each subsample fit together in a way conforming to established trends and contain no surprises.
I’ve just found another old but new to me article by Tversky & Kahneman. To my mind it suffers from the same shortcoming I have pointed out before. Choices are offered by humans and humans are fallible and not always honest. The choices offered here only make sense, when offered by omniscient gods on high and taken absolutely at face value. We are not trained to do that – and a good thing too – and we don’t easily shed that training. If I’m offered a certain $ 240 or a 25 % chance of $ 1000 I can believe the first offer. It was made in front of witnesses and is hard to get out of. But who is to notice if that promised 25 % in fact turns out to be only 20 %? In practice I routinely discount such projections and probabilities by at least a factor of two and comparing to those who don’t I have fared quite well that way.
A different but related argument can be made about loss. According to economists I should be indifferent to swapping my car for another of equal value and jump at the chance of swapping for one slightly better. Most of us don’t. We know what we have in the one we’ve got already and the other one may well turn out a patched-up dud. There may be a reason for offering us the choice and coming from a human there usually is. Even Mr Tversky and Mr Kahneman are not infallible and perfect gods and while they may imagine themselves as being seen as such, in actual fact they aren’t.
Third the theatre ticket. All tickets are identical and the same, but mass production was not around at the time our brains formed. Say you have closely examined and test driven a car and slept on it. When you come back to buy, you’re told it’s gone but another has just come in, that you can have for the same price. You have to decide now, because the play is due to begin in a few minutes. Would you do it? Our brains is a beautifully calibrated machine or we would not still be around. It may seem to malfunction in some highly artificial and hypothetical circumstances, but I for one do not live my life in a psychologist’s lab (at least I don’t believe I do).
Zum Anfang Übersicht Artikel Home & Impressum