I’m not a believer, but I can offer some defense of the practice (if not the theory) of astrology.
I forget where I read it(from here, see section 5), but my understanding is that astrology-like practices were very useful in early wars. Suppose an invading state had to choose one of two cities to attack, and a defending state had to choose one city to defend, and if the defender guessed right they would repel the attack, and if they guessed wrong their city would fall. If this scenario is repeated over the course of a campaign, the best strategy for the attacker is to pick the city at random, because any other strategy can be learned and countered. But since humans are bad at doing things randomly themselves, the best strategy ends up being to resort to divination as a kind of random number generator.
Similarly, astrology might be beneficial where it injects beneficial randomness into people’s lives. In addition to making them more daunting foes in prisoners dilemmas, it may force them to take slightly more risk or to explore slightly farther from what they would otherwise explore, and those divergences will pay off in many contexts.
Somewhat paradoxically, another benefit could be that it crystallizes intuitions that might not otherwise rise to the level of action. A vaguely worded horoscope that someone interprets as talking about the very thing they’ve been struggling to make up their mind about can push them to the conclusion that they would have made anyway sooner, which is almost always better (if it’s right, you get compounding returns; if it’s wrong, you fail faster and move on sooner).
Worth noting his studies have been challenged on multiple grounds. My immediate thought on reading this was that it could easily be a statistical artifact, and the wiki points out that I am not the first to think so. He was testing a number of astrological beliefs on the same dataset, that makes spurrious correlations more likely and increases the threshold for significance. This is something that is obvious in retrospect, but was not widely accounted for until recently (cf. replication crisis).
That doesn’t seem to matter in Karpel Tunnel’s experimental design. Even if we hypothesized that libras are more likely to select descriptions that contain the word ‘romantic’, if that’s a statistically significant result it would be noteworthy.
I do think there are other confounds, though. For example, in the US we have hard birth date cutoffs for school enrollment, so that kids born in August are effectively a year younger than some of their classmates born in July (and no similar gaps for kids born in March). That could affect personality later in life, and create a birth-month-based difference in adult personality without having anything to do with the alignment of planets or stars.
[EDITs: typos, typos, typos]