Account Share

 

Thread by @James_S_Murphy: "A long thread on the power of the @OfficialSAT to find "diamonds in the rough"; discredit the work of girls, African-Americans, Latinos, stu […]"

25 tweets
A long thread on the power of the @OfficialSAT to find "diamonds in the rough"; discredit the work of girls, African-Americans, Latinos, students in poverty, and first-gen students; and/or put a finger on the scale for the students who least need it.
Let's start with why colleges might use the SAT. If a student's SAT measures up w/ his GPA then it adds little value, other than confirmation. People who believe in the test argue, however, that the SAT can find diamonds in the rough and identify students with inflated GPAs.
In other words, the test matters when there's a discrepancy, of which there are 2 kinds:
1. HSAT--students whose grades don't reflect their ability, aka High SAT (HSAT) group.
2. HGPA--students whose SAT don't reflect their ability, aka High GPA (HGPA) group.
The other way to look at these same students is more negative:
1. HSAT--students whose SAT exaggerates their ability: aka High SAT (HSAT) group.
2. HGPA--students whose GPA exaggerates their ability: aka High GPA (HGPA) group.
College Board has recently been talking up HGPA as a case of grade inflation. A report published by the College Board suggests that we should be looking just as hard at SAT score inflation. eric.ed.gov/?id=ED563419
In 2010, the College Board conducted a study of 150,377 college students to see how well their test scores and HS GPA predicted their first year GPA (FYGPA) in college. FYGPA is the standard that the testing agencies use to measure the validity of their exams.
The study was especially interested in students whose SAT was discrepant with their HS GPA. They converted GPA and SAT into a standardized score and compared them. A discrepancy occurred when the SAT and HS GPA were at least 1 standard deviation apart.
Among the pool of students, almost 2/3 of the students had no discrepancy. And the share of students who were HGPA was almost exactly the same as those who were HSAT.
That makes it sound like test scores and GPA are about as good as each other since they're "off" as much as each other. If we look closely, however, stark differences appear along lines of gender, race, and class.
For under-represented minorities, women, low-income students, and first gen students with discrepant scores, the discrepancy is more likely to fall on the side of GPA.This chart shows that more female test takers are HGPA, male are HSAT.
The data for race also shows that African-American and Hispanic students have higher rates of discrepancy (41% vs 35% for the total group) and those with discrepancies tend heavily to HGPA.
Look at the pattern of SAT score discrepancies on income. We've long known that the wealthier you are the higher your SAT is likely to be. This report shows that the wealthiest students are more likely to have SAT scores that are inflated.
The pattern of SAT score discrepancies WRT parental education tells a similar story. The more educated your parents are, the more likely it is that your SAT scores are inflated.
Here's another way to look at the correlation between class and SAT scores. It is only among the children of parents with college degrees (and no grad degree) that we see a balance between HGPA and HSAT.
First gen students have higher rates of SAT score discrepancies and students whose parents didn't graduate from high school have, by far, the highest rate of discrepancy (HGPA). Notice that it ticks up for students of parents with grad degrees. That's a swing toward HSAT.
So, women, African-Americans, Latinos, poor students, and first generation students have a higher likelihood of applying to college with an SAT score that, relative to their GPA, hurts them.
White and Asian students are little more likely to reverse that dynamic, applying to college with an SAT score that, relative to GPA, helps them, and students whose parents made over $100k or had grad degrees are significantly more likely to have HSAT.
The College Board's sees these discrepancies as an argument for using both the SAT and GPA to evaluate an applicant's college readiness, and they provide this chart to illustrate their point.
It shows what happens when we base a prediction of a student's first year college GPA on just HSGPA, just SAT, and HSGPW+SAT. The third category is clearly the most accurate. Let's leave aside whether FYGPA is the correct measure of validity (I don't think it is.)
This chart is interesting because, in tandem with the other data we've looked at, it tells a story I've never heard before, about the kids with high SES and high SATs. It suggests that in some cases it is the SAT score that is inflated, not, as we have been told, HSGPA.
The chart shows that if you only use HSGPA to predict FYGPA, when you look at a student with a GPA significantly higher than her SAT score (or HGPA), you will overpredict what her FYGPA wil be by .21 points on a 4-point scale, e.g., you expected a 3.21, but she got a 3.0.
Using HSGPA alone underpredicts the HSAT student by .13 points, so you expected him to get a 3.0 but he got a 3.13.
If we switch to using SAT alone to predict FYGPA, things get worse. The HGPA student actually does .19 points better than predicted by her SAT. But the thing that really jumps out is that the HSAT student does .26 points worse than the SAT predicts.
The College Board has been shouting about HS grade inflation for the past year, but it looks like inflated SAT scores are even more misleading. Why aren't we hearing about this problem? Who is likely to have HSAT? Students who are wealthy and have highly educated parents.
What's news here is not that the children of highly educated & wealthier people do well on the SAT. The news is that many more of them do even better on it than they do at school & that phenomenon needs explaining. I suspect the answer is test prep.
This content can be removed from Twitter at anytime, get a PDF archive by mail!
This is a Premium feature, you will be asked to pay 30$/year
for a one year Premium membership with unlimited archiving.
Don't miss anything from @James_S_Murphy,
subscribe and get alerts when a new unroll is available!
This is a Premium feature, you will be asked to pay 30$/year
for a one year Premium membership with unlimited subscriptions/alert.
Did Thread Reader help you today?
Support me: I'm a solo developer! Read more about the story
Become a 💎 Premium member ($30/year) and get exclusive features!
Too expensive?
Make a small donation instead. Buy me a 🍺 beer ($5) or help for the 🛠 server cost ($10):
Donate with 😘 Paypal or  Become a Patron 😍 on Patreon.com
Using crypto? You can help too!
Trending hashtags:
Did Thread Reader help you today?
Support me: I'm a solo developer! Read more about the story
Become a 💎 Premium member ($30/year) and get exclusive features!
Too expensive?
Make a small donation instead. Buy me a 🍺 beer ($5) or help for the 🛠 server cost ($10):
Donate with 😘 Paypal or  Become a Patron 😍 on Patreon.com
Using crypto? You can help too!