Regarding using just HYPSM vs top ~12 vs top 20 vs top 25 vs top 50, as stated in my first post in this thread, I believe yield more closely follows a combination of selectivity and early admit policy than top ## ranking in your preferred publication. Selectivity more closely correlates to which colleges a student chooses to apply to as backups, and which colleges are applied to as backups is associated with yield.
For example, in the area of upstate NY where I grew up, it was common for students (and parents) to prefer RPI over MIT. This group almost never applied to the more selective MIT as a backup in case they were rejected by their preferred RPI during RD since hardly anyone who is rejected by RPI would be accepted by MIT. It makes more sense to choose less selective colleges as backups. As such, this group doesn’t decrease MIT’s yield even though they preferred RPI over MIT. However, it was common for students who preferred MIT to apply to the less selective RPI as a backup, so they do decrease the less selective RPI’s yield. In this group, yield follows selectivity rather than degree of preference.
The stats below compare yield by selectivity, using admit rate as a proxy for selectivity. To compare more similar colleges, I only included private, non-religious, non-LAC, non-HBCU, not very small 4+ year colleges that reported ACT score stats . There does appear to be a bit more sharp difference between HYPSM and everyone else than expected based on the trend line. However, there does not appear to be dramatic distinction between being above a particular top ## threshold and below that threshold.
HYPSM – Average 75% yield
Top 6-10 lowest admit rate - Average 60% yield (64% without low outlier Caltech, Chicago is high outlier)
Top 11-15 lowest admit rate - Average 54% yield
Top 16-20 lowest admit rate - Average 45% yield
Top 21-30 lowest admit rate - Average 32% yield
Top 31-50 lowest admit rate - Average 24% yield