3 ms·
"but at close to 100% small differences in the percent points means a huge difference in outcome." Close, but the actual effect is if you subtract one large nu
by VLM 13y ago
"but at close to 100% small differences in the percent points means a huge difference in outcome."
Close, but the actual effect is if you subtract one large number with modest error bars from another almost equally sized number with modest error bars you get a small number that is smaller than either error bar aka meaningless.
This is based on the assumption that class sizes of 70 students and 7 grads from a small southern school is going to result in pretty big error bars if I somehow got data for the whole country, for more than one year. I'm not motivated enough to do it, but I did gather enough information to conclude that regardless of major, virtually all freshmen change their major before graduation (if they graduate at all...)
Upon reconsideration, it doesn't even matter if this one uni for one year is a perfect sampling of the whole country for many years, the verbal conclusion stands, almost everybody switches majors and the dropout rate for CS isn't stunningly higher than other fields. When the dropout rate for all fields is about 80% to 90% its hard to increase that by a relevant percentage to "prove" CS is harder. To be stunning or interesting or noteworthy you'd need the CS dropout rate to be, perhaps, triple the rate of .edu majors or history majors. Or at least some small integer multiple. However at 80% dropout rate a mere doubling would already be mathematically impossible at 160%.
Now if I found stats that 75% of incoming declared freshman geology majors graduated with a geology degree, and this was about the same percentage across all majors like .edu and nursing and art, except CS where it drops to a mere 10% or so because CS is so hard compared to the other majors, LOL, then the original poster would have made an accurate point with possibly actionable outcomes (maybe CS needs more tutors or better english speaking profs or, or pre-reqs to declare it as a major, or the graduation reqs are ridiculous, who knows). But it turns out there is no relative numerical problem with CS dropout rates, at least not noticeably worse than any other major, and certainly nothing actionable in the real world.
That would tend to indicate the original conclusion that CS is so much harder than other majors is simply false, or if it is true, its certainly not true because of the dropout rate claimed as the cause.