Emil O W Kirkegaard

These aren't obscure articles. Grigorenko et al. (2001) has been cited 154 times (45 times by Sternberg). The Sternberg et al. (2001) piece has been cited 346 times (90 times by Sternberg). He cites both to support his triarchic theory of intelligence. 4/n

What's the point of showcasing these excerpts? Because unethical behaviors are probably not uncorrelated. If getting published and/or cited is rewarded, then accuracy, pursuing truth, and ethics are easily sacrificed. #Sternberg #SternbergGate #psychology 5/n

This example isn't proof that #Sternberg is a bad scientist. (He's not corresponding author on the Grigorenko et al., 2001, article.) But I do think that @OmnesResNetwork is right to suspect #Sternberg of QRPs. 6/nhttps://twitter.com/OmnesResNetwork/status/989321722611585024 …

But I don't envy the individual(s) who check #Sternberg's work. His work is pretty painful to read, as I've said before. #psychology #SternbergGate 7/7 ENDhttps://twitter.com/Russwarne/status/971250816228233216 …

OK, initial questions: 1) perfectly OK to use the same sample more than once to ask different questions. It's good practice to say you're doing so, of course, but it surely isn't a huge problem if the questions asked from the sample are sufficiently different. Are they here?

Yes, you can use the same sample in different articles...if you explicitly say they're the same sample. (I've done it.) The variables in the articles are overlapping. Both include English & Luo vocabulary scores and a Raven's Progressive Matrices score.

OK, so sufficient overlap in the results that you think it's problematic, and presumably without citation.

2) Can you link us something explaining why a correlation matrix needs to be positive definite? I tried understanding this once, but I woke up in a pool of my own sweat clutching a marmot in a town I'd never been to wearing something else's clothes.