Weekend reads: Death penalty for scientific fraud?; Why criticism is good; Cash for publishing

The week at Retraction Watch featured revelations about a case of misconduct at the University of Colorado Denver, and the case of a do-over that led to a retraction. Here’s what was happening elsewhere:

Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up on our homepage for an email every time there’s a new post, or subscribe to our daily digest. Click here to review our Comments Policy. For a sneak peek at what we’re working on, click here.

7 thoughts on “Weekend reads: Death penalty for scientific fraud?; Why criticism is good; Cash for publishing”

  1. Elsevier getting 15 million in damages is so important, it gets mentioned twice?
    (fourth point, and third-to-last point)

  2. With regard to the LIGO announcement of gravitational waves and the re-analysis by Danish researchers suggesting that the signal is really noise, does anyone know of an independent verification of the putative waves using the same raw data and the stated methodology?

    1. In terms of the LIGO detection, one doesn’t even need statistical analysis – the result is visible by eye (the first detection was at least 50 sigma, but in fact so strong they couldn’t really decide on how to estimate it, because there was no noise at that level to actually estimate it – they had to extrapolate the noise from far away in parameter space).

      But if you read the analysis of the Danish team, it’s obvious it’s garbage. For instance, they complain that once you subtract the best fit theoretical template, the remaining signal is correlated. Well, frankly, duh. The templates aren’t perfect matches – they’re theoretical templates constructed for possible signals, but they have finite resolution, so the actual signal is a little off the template. So, after subtracting the template, there’s a bit of signal left, which correlates with the bit of signal left in the other detector. Ditto when the noise is the sea of low power gravitational waves we can’t isolate from one another – that should have the same lag time and be correlated.

      1. The correlations persist for periods longer than 40 minutes, preceding transient arrivals by >10 minutes. The problem is that subtraction of template from whitened data seriously overestimates the majority of signal strength. The signals themselves are very generic broadband signals common to many non-gravitational sources at many time scales.

    1. This is entirely true. LIGO signals are order-subtracted, yet these processes are not also applied to coincident magnetometers or other multiscaled geomagnetic or space physical data. Transient vacuua and extended quasiperiodic coupling events may not have sophisticated exclusion protocols during LIGO data processing if detection thresholds are already inadequate or favor particular averaged windows that can obscure fine variation.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.