Cadets with lower grades improved academically if they socialized with cadets with high GPAs

Thursday, July 29th, 2021

Economists Scott Carrell and James West had noticed a pattern, a particular peer effect, at the Air Force Academy, Top Dog: The Science of Winning and Losing explains:

Cadets with lower grades improved academically if they socialized with, and spent more time around, cadet friends with high GPAs. The high-performers rubbed off on the low-performers, dragging them upward. Having friends whose SAT scores were 100 points higher than yours led to a half-grade improvement in GPA.

[...]

Carrell and West started by identifying which of the 1,314 incoming cadets had lower SAT scores and GPAs. These were the students most at risk of dropping out. They were assigned to special squadrons with a makeup of extra numbers of high-achievers. Compared with normal squadrons, these socially engineered squadrons had a few more low-performers, many more high-performers, and — to make room — fewer middle-performers.

[...]

More of the at-risk cadets were crumbling, not fewer.

[...]

Within a test squadron, the low-performers were self-segregating into cliques, to insulate themselves from the endless ranking and comparison.

[...]

Remember those squadrons comprised of leftover middle-performers? It turned out that their academic performance dramatically surpassed expectations.

[...]

When the field is too large, and the chance to be near the top is slim, people don’t try as hard.

Comments

  1. McChuck says:

    Wow, it’s almost like the researchers and social engineers had never met actual human people before.

  2. Gwern says:

    “Carrell and West started by identifying which of the 1,314 incoming cadets had lower SAT scores and GPAs. These were the students most at risk of dropping out. They were assigned to special squadrons with a makeup of extra numbers of high-achievers.”

    I haven’t read the study in question, but as described, this is nothing but regression to the mean. You filtered out and selected extremes on a noisy measure (SAT/GPA) who were, however, admitted based on their strengths elsewhere, and then you find that later on, additional noisy measures are higher rather than also extremely low. Regression to the mean.

  3. VXXC says:

    Duty vs. Tasks: We need courage not intelligence to be entrusted with Duties.

    Intelligence is for performing TASKS.

    Duty requires COURAGE, moral and physical.

    Meet our Elites and their striving Managerial Class: We selected for intelligence and ended up with sociopaths dominating smart cowards.

    Yes, it’s true across the board in our elites and certainly in the Army. I saw and see nothing in the Air Force to indicate the pattern does not hold.

Leave a Reply