[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

Re: Torosaurus NOT Triceratops

I'm quite taken aback that anybody would suppose that other people would 
casually suggest that yet other people are _evil_.

> they made choices that others might/would interpret differently in
> coding these characters, perhaps because they (the characters that is)
> need much more data or study and/or may be really tough to
> interpret/code.

This happens, of course -- but there are many, many other potential sources for 
errors in a data matrix, at least the ones for phylogenetic analysis that I've 
had a closer look at. And yes, some of those result in _objective mistakes_ 
without the slightest evil or unscientific intent on anyone's part.

Typographic errors are quite common. I've made plenty myself.

While the computer programs don't care, there are two conventions on which 
numbers to assign to which states: first, the plesiomorphy is 0 and the 
apomorphy is 1; second, absent is 0 and present is 1. When these conventions 
clash, confusions result, and I've repeatedly momentarily forgotten which 
convention I've applied to which such character. I'm sure so do other people.

It also happens that people lose the line or column and end up entering the 
correct value in the wrong cell -- I'm actually going to publish on two cases 
that are by far most easily interpretable that way (as part of a larger 
manuscript of course); in those cases, people ended up coding body parts that 
aren't preserved and that nobody had ever claimed were preserved. (Coding as 
not-unknown, I mean.)

Then there are correlated characters. Yes, in many cases it is very hard work 
to find out whether characters are correlated, and there's no good way to deal 
with cases that aren't all-or-nothing -- but there are matrices in which 
_identical_ characters are included twice with different wording.

This list is not exhaustive.