03 January 2011

Error

 Look, if you will, at the foraminifera below. They look fairly similar, don't they? It's not really difficult to see the difference under the microscope, but in some foraminifera studies they're taken together anyway. They are of the same genus, and they have fairly similar environmental preferences.



I compare my own counts with that prominent work from the eighties and nineties. And these two species are lumped. Fair enough. But the authors also lump two species (I could only find a picture of the one; the other is damn small, and difficult to take a picture of without a SEM) that look nothing like each other, and have very different environmental preferences. Why consider them together? No idea! But I don't want to. And I had calculated the difference between the abundance of that category of two species in the old data and my own data. And calculated the difference as well when using only the not-so-small species in my own data. And wrote elaborate text about what that shows and why I prefer using the second option.

And then I calculated average difference in foraminifera assemblage. And accidentally counted both options. Which one should not do. And I noticed very late. I wanted to go to the office to do the very final touches on the manuscript that deals with these things. And then I found out...

That meant back to the drawing board. I recalculated the average faunal difference again. And based on these averages I had subdivided my study area in regions of similar change. And that had, of course, slightly changed. So my regions needed to be redefined. And then all the calculations on these regions needed to be redone. And all graphs had to be remade. And with the help of the data guys in Tromsø I had done the initial calculations in a quick and elegant way, but I didn't manage that in Plymouth, so I had to do these calculations on massive data files by hand. How to fill your christmas holiday! This way.

One would be tempted to just let it be. The conclusions of the manuscript won't be altered by this slight change! Why spend days on correcting it. But that would be wrong. And on top of that, finding this error immediately gave me Climategate visions. One should not allow any imperfection! Any destructive nutter can claim you were manipulating the data willfully to serve an alarmist purpose. By the way; yes, my data shows that in the Barents Sea the warm-water forams are winning at the expense of the cold-water species, and no, that doesn't depend on the calculation error. But no climate sceptic would check that. It's wrong! It's a conspiracy! So here I am, in my (borrowed) office in this further deserted building. Being inspired to next time check, check, and check again, and again check, my data before I turn them into a manuscript.One is never too old to learn of one's own mistakes...

No comments:

Post a Comment