Sol Invictus wrote:No, but conclusions they drew from the data were incorrect.
If we put aside dating – which conclusions about common ancestry are incorrect?
it might as well result, for example, from languages in the same geographical region borrowing common words from the same foreign languages and each other.
The research data is freely accessible: http://ielex.mpi.nl/ If you believe their data were misleading or cognate judgement was massively incorrect, you can prove it by pinpointing the exact mistakes.
The research was supposed to exclude loanwors, according to current knowledge about all the languages studied. What constitutes a cognate was desided by trained linguists. As far as I can tell, the Lithuanian cognate data seems to be more or less correct.
Bear in mind that any research of this kind is based on assumptions and on our best current knowledge. If it is discovered that LV roka / LT ranka / PR rānka is a loanword from some language, then future research will have to reflect the new finding. But current research must be based on our current assumption that this is not a loanword.
that common ancestor for two groups that is significantly different from an earlier common ancestor of all three groups cannot be reconstructed
The trouble is that the reconstruction of the last common ancestor of one single node of just Latvian and Lituanian (that we assume existed till some 5th century AD) would not be significantly different from Balto-Slavic which could have existed millenia before. This is because we have a fixed set of data (no newly discovered sources). If you take a look at the list in Wikipedia, almost all the forms can be derived by comparison of just Latvian and Lithuanian (with a tiny bit of Prussian). This was basically the standard practice 20 years and before among Baltic linguists.