Scientific Misconduct in Geoscience
Well, its been fun noting that the bulk of the issues in places like RetractionWatch are in fields far from geoscience, but we now have our very own cesspool of misbehavior (one that, indeed, RetractionWatch covered but GG missed). And, it would seem, it is pretty unusual. Unlike the high profile cases of fraud where scientists are publishing fake results for fame or p-hacking to make a silk purse from a sow’s ear, this involved technicians at one lab at the U.S. Geological Survey not following protocol. What sets this apart in some other ways is the duration of the transgression: this represents 18 years of misbehavior.
The basic story is outlined in an AP report (via the Denver Post) and much of the rest of what was going on is detailed a bit more in the Interior Department’s Inspector General’s report. What it amounts to is that a USGS lab, which made measurements for investigators both inside and outside the USGS, failed to operate their mass spectrometer sensibly. It sounds as though the usual procedures for calibrating the instrument were not followed and that records of measurements and calibrations were not properly kept (which, it seems, has left everybody unsure just how far off the measurements really are). In lieu of doing the proper calibrations, the staff adjusted numbers in a direction that would kind of look right. About the most complete public explanation is in a USGS announcement of the problem from back in May:
During this time frame, some data were manipulated both to correct for calibration failures and to improve results of standard reference materials and unknowns. Some of the original files (“raw data”) from the ICP-MS instrument are unavailable, thus the measured concentrations cannot be re-checked for accuracy. As a result, the trace element and rare earth element concentrations reported from these ICP-MS analyses are considered suspect and are to be used with caution.
What is scandalous about all this is that this same lab previously had just about the same problem running from 1996 to 2008. That notice is in some ways clearer about the transgressions of this lab:
The most serious shortcoming was the adjustment of raw data to a standard when the instrument reading for the standard was beyond acceptable limits, or when the frequency of repeat analyses of standards was insufficient. In general, adjustment of raw data to account for instrument drift is acceptable practice within strictly defined limits. During the period in question, the maximum adjustment of instrument readings, guided by calibration standards, was not allowed to exceed 10%. However, in some cases the adjustment exceeded 10% and/or was not constrained by an adequate number of control standards. Original instrument readings no longer exist for about 80% of the analyses in question and we are unable to determine the acceptability of drift corrections for most of the samples analyzed during this period. For these reasons, 1996-2008 data from the USGS ERP Inorganic Geochemistry Laboratory should be described as “semi-quantitative” and should be used with care.
In other words, for 18 years the USGS ran a lab that produced questionable data. They couldn’t get the right numbers from their calibration standards and so fudged everything so it would look about right. Awhile back GG asked, how ethical is geoscience? The answer might be, far less ethical than the absence of retractions would suggest, if a lab can misbehave over such a long time and continue to do so even after being caught at it and disciplined. And remember, this is misbehavior that really wasn’t for personal gain nearly so much as emerging from some blend of incompetence and laziness.
At least this time the USGS closed down the lab. But the duration of the transgression reveals startlingly poor oversight. It also seems to show that few if any of the lab’s users were verifying the lab’s fidelity by either including unmarked standards or running splits of the same samples with other labs.