West Point Admissions and Separations are tightly coupled. The quality of an incoming cadet class influences its overall performance. A higher quality incoming class will have better outcomes at the Academy and in the Army. The Admissions group makes decisions that affect the quality of the class and therefore its outcomes at the Academy and in the Army.
This sounds common-sense. It follows that Admissions ought to make those decisions to build the highest quality classes that it can. But when we proposed, at this “How much can being white hurt you in admissions? post at Service Academy Forums, that Admissions frequently makes those decisions in a way that sub-optimizes class quality, that it does so for reasons of diversity and sports, and that this methodology ought to be scrutinized and challenged, we got comments like those below:
Underrepresented minorities have a greater chance of appointment in the Additional Appointee category – however, that does not necessarily reflect on their qualifications. Many AA’s are highly qualified and would win other districts, just as many district winners would not win districts other than their own.
Unqualified candidates do not get into West Point, and if admissions occasionally makes a mistake, they do not graduate.
jl123
Your statement that “20% don’t graduate” and insinuating they don’t graduate due to them possibly not being qualified or “mistakes” by the SA is a big reach in my opinion. Plenty of cadets, plebes, midshipmen etc say yes, take the oath, experience military life and decide it isn’t for them… That doesn’t directly correlate with them being not qualified or admissions mistakes. It implies the system worked.
heatherg123
The SAs have successfully admitted, trained and produced high quality officers, citizens and leaders for over 200 years. The process works, and the margins of the bell curve do not prove otherwise.
FiveByFive
Likewise, measuring “success” in the admissions system by who stays and who leaves is really not worth the time. Students leave for a variety of reasons, which of course can include not being cut out for the “academics” at some point, but that is only one issue (especially at an SA). Being anything other than “white” is no different in the “attrition analysis” even if some factors for why someone leaves a school may be more or less prevalent for an underrepresented group.
At an SA vs. a civilian school, making conclusions from the attrition rate as a measure of “unsuccessful” practices in admissions makes very little sense if any.
FiveByFive
All that merits looking deeper into the separations data. We address the following: Does preferred minority status confer an advantage in admissions, and is there an effect in separations? Does Admissions influence class performance and impact separations, and how much? Should the Academy be doing more to retain its cadets? If so, what should be looked at?
Minority Preference In Admissions
Always a controversial topic but one worth understanding before we move to effects on separations. Without understanding inputs, we cannot understand the outputs.
We previously analyzed admissions in our posts On Admissions Excellence , On Diversity As Strength, and Merit-Ranked Class Composition. In bullet summary order, these showed that the Academy is:
- deliberately not admitting the most qualified classes that it can, and
- it is doing so in a way that strongly indicates that it is (in whole or large part) for purposes of diversity recruiting and for recruited athletes
To recap, let us revisit this using admissions by SAT scores. First we show the absolute numbers of attendees by race in the years reported (2010-2018 for this post).
We lay self-reported race groups in a cumulative distribution chart against combined Math & Verbal SATs. The X-axis shows the combined scoring and the Y-axis shows the cumulative percentage of the group that attended West Point.
What immediately jumps out is that there are different curves. Keep in mind that this is cumulative percentages of attendance for each group at USMA at the scores.
This shows us that the race groups attending USMA have different characteristics that are not easily explained by other admissions criteria. If all else were equal–group characteristics, admissions standards, and cadet qualifications–we’d expect that at a given score level all groups would have similar percentages of attendance. For example, in this scenario all cadets who attend at the 1090 range would be equally below the academic mean and would have some compensating quality in their WCS to make up for it. So the lines would overlap.
But this is not what we see. We see different lines for the major groups.
What does this tell us?
To explain, let’s consider the counterfactual. In a perfect world with one set of admissions standards, equal representation criteria, and comparable aptitude distributions, there would be only one curve on the chart and all races would fall on it.
This leads us to the “iron triangle” of admissions. Admissions criteria, applicant qualifications, and representation at the Academy make up its sides. If one of those sides changes, one of the other sides must change accordingly.
In reality, group characteristics are real and different. Second, the Academy seeks racial (and gender) class composition goals. Therefore the admissions standards for different groups must be different. And we see that they are. For example, at the 1110 combined SAT score level, 52% of the Black cadets are attending compared to 12% of the Asian cadets and 14% of the White cadets.
Separation Types
Now to review separations types. The data contained separation reasons for each cadet (if the cadet didn’t graduate). We grouped separation types, including resignations, separations, suspensions, and a few other miscellaneous categories, into a several major cause buckets:
This was to more easily identify the cause of the separation from how it was effected.
Impacts on Separations
Now that we have examined some of the inputs, let’s look at outputs.
Separations rates by race group and SAT category shows that for the classes analyzed (2010-2017 and keeping in mind that 2018 had not yet graduated when this dataset was obtained) Blacks separate at 26% rate, Other at 22%, Hispanic at 21%, White at 16%, and Asian at 15%.
We see separation percentages roughly following the same type of curve as the test score/attendance curve. Lower SAT scores tend to separate more and earlier than higher SAT-scoring groups. But that is not all. If all separations were Academic, we’d probably see much more closely aligned lines. But there is a significant difference in magnitude and slopes.
What else could cause that?
Separations Data
Above we saw that the groups have different overall separation rates.
Now we present a table (“Table 1”) showing the breakdowns of separations by category and race. Percentages are calculated in columns.
Motivation is the leading category of overall separations, followed by Academic, Honor, Conduct, and Religious reasons. So indeed many cadets feel the Army isn’t right for them. But another quarter of separations are due to Academic reasons and Honor, which is often tied to Academic pressures (think of cheating on a test or paper).
Below (“Table 2”) is the same data with percentages calculated on rows.
This gives us the breakdown of separations by race category within each of the separation causes. For example, Asians make up 5% of Academic separations, and Whites 47%.
We provide in Table 3 a view of separation buckets and rates by sex. Percentages are again calculated on column.
For example, we see that 23% of male separation occurs in the Academic category compared to 18% for women. This shows that women tend to be more studious than men, less prone to behavioral problems, and more prone to injuries and motivational problems. All of this probably surprises no one.
So, given this information, what can, or should, Admissions do?
What can Admissions do?
We see that Motivation and Academics are the top 2 drivers of attrition followed by Honor and Conduct. We also see that different “slices” of the Corp have different retention opportunities that can be addressed in Admissions.
We see that women tend to separate at much higher rates for Motivation and Medical reasons (probably related) than men do. Perhaps such testing could address this and enable the Academy to screen for the correct cadet class instead of engaging in harmful sex-based recruiting. Women are being injured badly enough to leave the Academy at double the rate men do, but the gender class compositions are proudly trumpeted on class profile information.
The single biggest group source of attrition is White motivation, which at 45% of White separations is 37% higher than the rest of the groups’ average rates. This leads us to ask: Why are Whites more prone to motivational separation than other groups? Perhaps they don’t get the support other groups do?
Regardless, there is something there that the Academy can address through counseling, psychometric profiling prior to admission, or other measures. The data doesn’t tell us what the cause is, but it is a huge opportunity.
In Academics, Blacks wash out at 42% rate (table 1), or 72% higher than the rest of the groups’ academic separation average rates. This is 28% of all academic separations. We also see higher Honor separation rates in the Black group than others, again likely tied to cheating and academics-driven pressures. Both of these are disproportionate to Black representation at the Academy.
The buried lede is that this could be avoided by maintaining consistent admissions standards for test scores. Admitting 50% of Black cadets at a standard-deviation lower of SATs than other groups cadets (by comparison most of the rest of the Corps is around 10%-15% at the same score range) puts that group at high risk for academic problems.
Someone may challenge the link between test scores and graduation rates. We have explored that link in previous posts. There is clear correlation between them and substantial academic literature backing it up.
Conclusion
West Point Admissions has opportunity to build classes that are much less likely to wash out. They could lessen the number of Academic and Honor separations simply by having consistent admissions criteria for all groups similar to what they apply to Whites or Asians. They could add another profile test weighted in the to the applications process to evaluate personalities against those who graduate and those who quit.
But they do not. Instead they pursue class composition goals unrelated to winning wars and setting the Army up for future core leadership. The Academy prefers touting diversity achievements and sports recruiting to building the strongest classes possible.
This is doing no one any favors. It does not help the cadets who struggle because they were misplaced into a rigorous academic environment. It does not help the cadets who quit because they were misplaced into a cultural environment. It doesn’t help the women who are hurt bad enough to leave. It does not help the qualified, motivated applicants who were rejected in favor of cadets who would struggle or fail. It does not help the academic environment, the Academy’s reputation, perceptions of various groups, or the soldiers who get officers who struggle cognitively (and yes there are exceptions but we are talking about large numbers).
But it does help somebody, or else it wouldn’t be happening. We will leave who, and why, as an exercise to the reader.