USMA Acceptance Rates: Who Didn’t Make It

Let’s conduct a thought experiment on the West Point acceptance rate.

The question

West Point is considering admitting two groups of candidates. Both groups contain 178 candidates and have been deemed 100% “qualified”.

You get to decide which group is offered admission. Keep in mind, 178 candidates is about 15% of an entering class.

Keep in mind, you are the public, the taxpayer, and future recipient of the services provided by the Officer Corps of the United States Army. If your kids are going into the Army, then they may have to serve under these future officers.

Who do you choose?

Based on the group profiles below, including test scores, number of High School Varsity Letters, and class president/eagle/gold award stats, and what you know about West Point’s mission, do you admit group A:

Or do you admit group B:

Go ahead, write it down.

Group Descriptions

We assume most people will look at the test scores, the leadership evidence, and the athletics accomplishment of the two groups and pick group A. We would.

Group A’s minimum test scores are around the 60th percentile, plus there is more leadership and sports participation (higher average number of high school letters per candidate) than Group B. This means that A-group is not just a bunch of library-bound nerds. The max test scores are at the top of the charts, and the averages are around a healthy 90th-percentile range. This means the Whole Candidate Scores should be fairly high.

Group B contains cadets scoring in the 15th percentile or less of ACT tests. The 860 SAT is around the 20th percentile. The averages are around the 50th-60th percentile range. B’s leadership indicators are almost non-existent and the sports participation, in letters, is far below A’s. Would admitting this group really meet the standard that gets West Point to its vaunted sub-10% highly-selective admissions rate?

A Real Choice

The big reveal is that this is a real choice that West Point faced for the class of 2019.

The class was arbitrarily chosen from recently graduated classes. We analyzed the class applicant and cadet data, obtained by FOIA from the Academy, to develop a close estimation of the Whole Candidate Scores for whole set of candidates for the class. We determined who attended based on separation and assigned company codes. Then we rank ordered the cadets who were “qualified” and did not attend, and those who attended (therefore were admitted to) West Point. See end of post for more notes on data and analysis.

Group A represents the top 25% of candidates who were deemed qualified and did not attend West Point. 178 candidates is 25% of the qualified-no-attended group.

Group B represents the 178 “bottom” cadets (ranked by our WCS) who actually attended.

The Academy had an opportunity to admit a clearly superior group of cadets and did not.

Why would that be?

Decision Reasons

“Well UD” you might say. “We don’t know that group A wasn’t offered admission. Or even if they were offered admissions, they might have declined to attend. Maybe they weren’t nominated. Maybe Group B was all primary nominations.”

Ok, perhaps. It’s true that we don’t have records of offers extended or nomination sources or types. But Group A candidates are all qualified (and highly qualified, relatively speaking) candidates. We also know that qualified candidates for class of 2019, according to USMA, numbered around 2300 and were admitted at about a 50% rate (whether “admitted” means “attended” is another question).

Maybe they are the evidence of offer yield percentages. That is, maybe this is the group that all got a West Point admissions offer and opted to go somewhere else (not likely though after full application and getting qualified, but let’s consider it). After all, they’re a pretty solid group of candidates and have other options.

So let’s play along and assume that’s the case. Let’s assume Group A all got offers and declined them, and present Group C, the next 178 qualified-not-attended candidates:

Reminder, group B:

Who would you choose to attend? Group C is still objectively much stronger than Group B. But Group B was admitted and attended.

And this is for one class, class of 2019. We’ve evaluated 356 of the 818 qualified cadets who did not attend. Even with an 80% yield rate, USMA should have been able to place a much better class than it actually admitted.

We are highly confident that we’d see the same pattern, year over year, in other classes.

So you might say further, “So? And even if those things aren’t completely true, West Point probably had reasons to make the decisions it did.”

Sure it did. Let’s talk about another set of differences between Group A and B.

Racial Preferences

Let’s compare Group A and B on another plane, group demographics. We break out groups A and B by the coding in the data, assuming the racial and sex codes are as follows.

A-ha! We note that group B is much more “diverse” than group A, which is a huge selling point, if one is seeking to satisfy a politically-driven agenda rather than graduate the best qualified Army officers!

Preferred groups get a huge boost here. Blacks, Hispanics, Native Americans, and Other get the benefit, with Black candidates seeing tremendous relative overrepresentation. This comes to the detriment of White and Asian candidates, who gave up the slots.

So West Point gets diversity bonus points for passing over qualified candidates. And with West Point’s documented racial preferences goals in admissions it is not surprising that they’d lower standards to meet those goals (or “quotas” if you’re not a lawyer).

Athletics

One other relevant question is how many of Group B were recruited athletes vs. diversity appointees. Both of those categories comprise the bulk of candidates who are admitted as “Additional Appointees” from the National Waiting List, and where Admissions has complete discretion over who is selected to receive an offer. [see our previous article on WCS for more on this].

Our only means of inferring that a candidate was a recruited athlete was to examine who later, as a cadet, participated in intercollegiate (“Corps Squad”) athletics. Such data were available only for Group B, since Groups A and C were not admitted.

We found that 92 of Group B participated in intercollegiate athletics. Some (if not many) of those 92 were thus likely to have been recruited athletes. It would follow that the other 86 of Group B, and those of the 92 who were not recruited athletes, or most of them, despite their lower Whole Candidate Scores than those in Groups A and C, were admitted to help meet the “class composition goals,” discussed in our article on West Point’s class composition goals.

Effects

We noted in On Admission Excellence that aptitude matters for retention. For groups of 178 cadets with SATs medians of 1330 and 1050, we’d expect all-cause separation at 13% and 22% rates respectively. This means we’d expect Group A to separate 23 cadets and Group B to separate 39 cadets.

(Keep in mind that the all-cause separation rate contains all separations. So if it were drilled down to more “controllable” separations like academics, honor, and conduct, the separation effect would be even larger than indicated in those numbers.)

In other words, by not holding to standards and creating the best class it can, USMA is accepting that approximately 16 more cadets in the examined cohort will fail and separate. This comes at cost and trauma to the cadets and wasted effort on the part of the Academy and taxpayers.

Conclusion

West Point doesn’t get to “<10% admissions” by holding everyone to the same standards. There are clearly different standards for different groups.

West Point admits parts of its classes with different standards to meet demographic goals. It is consciously passing over more-qualified candidates for racial composition and other purposes. In doing so it is admitting candidates who will struggle academically and conduct-wise, drop out at higher rates, and not be as good officers, as it could have done.

If you still think the Academy had no choice but to admit group B and did the right thing, you should ask what evidence to the contrary would persuade you otherwise. How many more qualified candidates have to be passed over–to be discriminated against because they checked the wrong box on the application–before we find that West Point maybe has its priorities wrong?

It’s disappointing that West Point’s leadership tolerates building classes to look good to politicians and press releases. It should be focused rather on taking the best classes it can to build the best Army it can.

Factual corrections and thoughtful criticism are welcome.

Data Notes

To provide transparency, below is further data regarding the data sources and methodologies used for this post. Factual / methodological feedback is welcome. (pardon the block editor incorrect list numbering)

1. Group profile methodology included developing “pseudo WCS scores” as described in our post “A Class Admitted on Merit” [link]. Briefly, we used our dataset and the GainServiceAcademyAdmission WCS calculator to approximate Whole-Candidate Scores for entire classes of applicant files, including both those who attended and did not attend. We then ranked by “pseudo-WCS” the overall group and compared it to the class of cadets that actually attended. This became the basis for evaluating the lowest-ranking group that attended West Point and comparing it with the top of the qualified group that did not attend.
2. Data file referenced: Class Year Data post, class year 2019 data [direct file link].
3. “applicant_qualified” is a field directly accessible in the data file. See column AB below. Note also that “applicant admitted” flag reads “N” for all candidates, even for those who had grade point scores and cadet companies assigned.

1. To work around this data error, we added “Attended” as a binary dummy variable for 2019 dataset and in other analysis. If a record showed a company assignment then the field would read “1” indicating that the cadet attended West Point. There were some records with no C/Q/MPAs but with company assignment, those still showed as “attended”. This field lets us filter on those cadets attended and not attended. Update 3/30: Thanks to a reader asking this question: What about cadets who quit during Cadet Basic Training? On further investigation we found that those cadets who resigned during New Cadet Training had neither an academic company nor GPA scores recorded, but only the separation reason. We have removed those cadets (numbering 2 total) from group A, filled with the next 2 qualified-not-attended candidates on the list. None was found in group C.
2. Varsity sport letters: We referred to columns in the data as shown in an example below with each row representing a cadet record. No key was provided to the sport description. However we can reasonably assume that, for example in the second row, “MBB” is Men’s Baseball or Basketball, and that field “hs_sport_1_ltr” of 4 means the candidate had 4 letters (or 4 years lettering) in the sport. Of course if there is a better key to reading this we welcome it. We then found summed the total letters earned for each candidate record to get to the total letters per candidate, and from there calculated the max / min/ averages for the groups.

1. There are some discrepancies in the total numbers of cadets and applicants in the files we used compared with those published on West Point’s class profile page. We speculate that these are due to differences in terminology, timing, or turnback and separation status. Where we use the West Point website source we reference or link it, otherwise data points are from our file and analysis.