Analogies are legitimate tests of reasoning ability. As I recall, however, the analogies of the "old" old test often relied on obscure words. This turned them into de facto vocabulary tests, defeating their purpose as pure assessments of reasoning. At the very least, vocabulary became a confounding variable in assessing the specific aptitude that analogies were meant to assess.

One could make a strong argument for plain-English analogies, absent the $10 words. If every test taker understood most of the words, then we could test for reasoning ability on a more normalized basis.

That's not to say testing vocabulary is totally invalid. Opinions vary. But if we assume vocabulary is worth testing, then we should test it independently.