I was referring to Weims 2006. In this paper, he offers more
hypothetical results rather than real evidence of having tested his
statements. No data matrices are presented, nor resultant analyses (he
does refer to a paper "in press"). He writes (p40): "I tested the
effects of adding 50 incomplete characters to a set of 50 characters in
the 16-taxon case using parsimony...adding the set of incomplete
characters generally INCREASES phylogenetic accuracy relative to
excluding them." Hell, in that case, let's just cut 50% from all the
characters of Wilson 2002 and call the results more accurate than Wilson
2002.