Grant pressure amps up the scientific cheating?
May 14, 2013
Well this is provocative. One James Hicks has a new opinion bit in The Scientist that covers the usual ground about ethics, paper retractions and the like in the sciences. It laments several decades of “Responsible Conduct of Research” training and the apparent utter failure of this to do anything about scientific misconduct. Dr. Hicks has also come up with a very provocative and truthy graph. From the article it appears to plot annual data from 1997 to 2011 in which the retraction rate (from this Nature article) is plotted against the NIH Success Rate (from Science Insider).
Like I said, it appears truthy. Decreasing grant success is associated with increasing retraction rates. Makes a lot of sense. Desperate times drive the weak to desperate measures.
Of course, the huge caveat is the third factor…..time. There has been a lot more attention paid to scientific retractions lately. Nobody knows if increased retraction rates over time are being observed because fraud is up or because detection is up. It is nearly impossible to ever discover this. Since NIH grant success rates have likewise been plummeting as a function of Fiscal Year, the relationship is confounded.
DrugMonkey 
May 14, 2013 at 10:08 am
Well, we have to wait for NIH success rates to go up again.
LikeLike
May 14, 2013 at 10:21 am
Agreed that will be interesting
LikeLike
May 14, 2013 at 10:24 am
Maybe someone can write a grant application to raise success rates so that we can test the hypothesis.
(wonder, also, what inter-institute success rates might show. Are success rates variable enough and subfields funded specific enough that one could look across them v retraction rate in the subfield?)
LikeLike
May 14, 2013 at 10:40 am
I doubt it. There are certain technical associations in the cheatfucks record….that doesn’t break down by IC
LikeLike
May 14, 2013 at 10:55 am
Seems impossible to know. Since funding rates are at historical low and detection and ability to fabricate (because of technological ability) are both likely at historical highs, as DM says, confounding.
LikeLike
May 14, 2013 at 11:29 am
So is every graph that shows a correlation “truthy”?
LikeLike
May 14, 2013 at 11:34 am
No. Only the ones that are truthy.
LikeLike
May 14, 2013 at 12:43 pm
I’m not sure if the frequency of retractions due to fraud is benchmarked to retraction date (when it was discovered) or publication date (when it occurred).
LikeLike
May 14, 2013 at 1:13 pm
Wow – NIH success rates are so important that they appear to affect the rate of retractions in the entire world!
LikeLike
May 14, 2013 at 1:25 pm
Wow, retractions have doubled under Obama!
/teaparty
Of course, we all know the real reason for the rise is gas prices, or something like that. What kind of total fuckwit tries to play the correlation=causation trick with a bunch of scientists FFS? Talk about lambs to the slaughter.
LikeLike
May 14, 2013 at 2:42 pm
Ola – it’s truthy. Just leave it at that.
LikeLike
May 15, 2013 at 4:04 am
Cath – no, it’s the other way ’round. The rate of retractions in the entire world is so important that it appears to affect NIH success rates.
LikeLike
May 15, 2013 at 4:44 am
While the data might have other confounding factors, it’s pretty obvious that more competitive environment = higher incidence of fraud.
Seriously, is that such a hard conclusion to accept, or do you people think that scientist are somehow morally/ethically above everyone else? The question that we should be asking is how much of an impact does the environment has on encouraging fraud (relative to the “baseline” fraud rate)?
LikeLike
May 15, 2013 at 5:00 am
See this analysis folks
http://richvan.wordpress.com/2013/05/14/us-grant-pressure-and-retraction-rates-not-a-meaningful-comparison/
LikeLike
May 15, 2013 at 6:38 am
My anecdotal experience is that cheating increases when as the system becomes more arbitrary instead of correlated to merit.
This does not necessarily correspond to an absolute percentage of success. An award such as the Nobel prize has a very low success rate, but it is still relatively highly correlated to merit, and hence one would empirically expect to see less cheating.
Come up with an arbitrary system in which grants get rejected because of the wrong font or being a paragraph over an arbitrary page limit and the probabilities of cheating will go up (not that we have ever seen a grant system like that, have we?).
LikeLike
May 15, 2013 at 7:08 am
I think DM nails it on this. It’s like all the studies that try to link the “rise in autism” to something without taking into account the fact that many people with autism would until recently be characterized as just odd (if mild), or with some other mental disorder (if severe).
The idea of “truthiness” doesn’t mean that a conclusion is necessarily false, just that more than “it kinda makes sense” is needed to infer causation from correlation.
LikeLike
May 15, 2013 at 8:03 am
Was the Watson and Crick move highly correlated to merit, Busy?
LikeLike
May 15, 2013 at 11:23 am
There is a reason I wrote less cheating not no cheating, and it is not just Watson and Crick but also Banting and McLeod.
LikeLike
May 15, 2013 at 4:59 pm
@Busy
If the system’s arbitrary, why bother cheating?
You’d be much better off just using the right font or making sure you’re under the word limit.
I don’t see why cheating should arise or be beneficial in that sort of arbitrary system.
LikeLike