When they originally announced the college credit thing, I asked them if they had considered partnering with HBCUs on this.
So, I hope they do go somewhere with it.
I was looking at Math SATs for some HBCUs. Howard is first with a 75th percentile of 635, Spelman is second with 600 and Morgan State has a 75th percentile of 530. I’m at a school where the 75th percentile is 600 and I know how few people at this university can pass even the first exam which is still going to be a requirement. I wish them luck but I don’t envy who will be in charge of providing the education and rigorous examinations,
That’s depressingly low, but probably 90th or 95th percentile is a more relevant metric.
Or better yet (but likely substantially harder to find) the average math SAT score among math & stats majors. Those are the people who might choose actuarial work. The students with the 75th percentile math SATs probably mostly aren’t even STEM majors, let alone math & stats specifically.
I’m not sure it’s relevant how much better Harvard’s communications majors are than Spelman’s at math. Ain’t none of those students going to become actuaries anyway.
Regardless, by almost any academic metric, I would expect Howard, Spelman, and Morehouse to be the top 3 or at least in the top 5. I know less about Morgan State.
When given quartiles, any score above Q3 + 1.5IQR is considered an outlier (analogous to a Z score greater than 3). At Morgan State a Math SAT above 665 would be an outlier. Based upon my experience 680 is about as low as you want to go (although there are exceptions). While SATs can be criticized on many grounds I have found them to be a very good predictor of ultimate success on actuarial exams.
1 Like
Would that assume a normal distribution? If so, are math SAT scores among HBCU students normally distributed?
I gotta think that if you look only at the math/stats majors you’re going to see a different picture emerge. And for the purposes of recruiting more African-American actuaries they’re (the math/stats majors) the only students who matter.
Regardless, I stand by my assertion that 75th percentile is not the right metric (though the right metric is likely much harder to find).
I don’t think it matters if only 1% of HBCU students can cut it on actuarial exams compared to 3% or even 10% at some other school.
If you have a goal of recruiting more black actuaries (and one could potentially argue whether this is a worthwhile goal, but I’m assuming that it IS an actual goal of the societies), then finding and targeting the ones who have the aptitude to pass the exams coupled with the interest in the subject matter is going to be important.
HBCUs produce 25% of all African-American STEM majors. So you could try to find the 3% of math/stats majors who are African-American and attending one of the 3,900 non-HBCUs or you could look at the 73% of math/stats majors who are African-American at one of the 107 HBCUs.
Even if we assume that the African-American math/stats majors at non-HBCUs are on average better qualified than the ones at HBCUs (a statistic not actually demonstrated in this discussion) I think you’ll still get more “bang for your buck” by focusing on the ones at the HBCUs. Especially the top academic ones like Howard, Spelman, and Morehouse. And given the proximity of Spelman and Morehouse and the connection between those schools they’re almost a two-fer.
So you guys are looking at the wrong end, in that you are looking at selectivity of the schools, SAT scores, and all that.
I’m more concerned about the output – heck, how many people will actually leave with the degree? And then how many people will become actuaries?
I had specifically recommended Spelman in writing to the SOA, because they have some math-specific programs that I thought the SOA could partner with, and Spelman has an excellent reputation. (also, Spelman is a women’s college)
Anyway, here’s something re: graduation rates (albeit 4-year graduation rates, and obviously pre-pandemic):
My employer has recruited actuarial talent at HCBUs, and we’ve successfully hired some excellent entry level actuaries. Fwiw.
I didn’t see anyone looking at the selectivity, FWIW, and that would certainly be the wrong metric.
I was the one who pointed out that HBCUs produce 25% of all African-American college graduates with STEM degrees (although I shortened it to “majors” in my post because I thought it was implied that this meant they graduated, but maybe it wasn’t.)
I will say that graduating with a math / stats degree is probably also not the ideal metric in that lots of people of every color have the degree and still don’t fare well on exams. And we’re not trying to increase the number of black exam-takers we’re looking to increase the number of black exam-passers (I assume). So I think the combination of a math/stats major and math SAT score is going to be more predictive than merely targeting all math/stats majors.
To answer your question about the definition of an outlier, the upper/lower fence definition is due to Tukey and is designed to be distribution free. Of course the probability of such an outlier would depend upon the distribution. With a normal you would expect an outlier on the right with probability .005, while with an extreme value distribution such as the Gumbel it would be about .025.
1 Like
I hadn’t heard of gumbel. I had a quick look at wikipedia, interesting rabbit whole to go down there. Says it’s related to to Gompertz. I.wouldn’t say this in public, but, cool.
It also calls it the double exponential. The pdf looks like it’s one of those that’s ‘testable’ because it’s full of cool properties.
Cool. Til.