That’s true, IFF that was the "sole purpose: of the UC officials. In fact, UC has given admission bonus points to SES apps for years.
But UC’s purpose as pointed out upthread is to find a way to admit more URM’s --into Berkeley and UCLA, primarily – in a state that prohibits affirmative action in its public colleges.
At the time of the Hout report listed above, the general weighting appeared to be something like below, listed from high to low. Students were given a boost in read score for being low SES (different from boost in test scores), but that boost was quite small. However, various personal/character factors that were correlated with low SES appeared to have greater weight than a direct boost for low SES, such as “lives and learns in an environment with limited academic opportunities, relative to the Berkeley applicant pool.” Both wealthy kids and low SES were given a boost for limited academic opportunities, but I expect the low SES kids were far more likely to be in this category than high SES, so the net result is increased low SES.
Estimated Most to Least Influential Factors in UC Berkeley Reader Score (2005)
1 . GPA
2. Eligible in Local Context (ELC)
3. SAT II Writing
4. GPA Context (Grades trend up/down, Senior year schedule, …)
5. AP Scores (5s appear to have much more influence than 4s)
6. Personal/Character (Limited academic/EC opportunities, Overcame obstacles, Had job, …)
7. SAT Scores (>750 score has more influence than low score)
8. Low SES + Demographics
The end result of this type of weighting has been that the UCs tend to matriculate a larger portion of not high-income kids than similarly selective peers. Some example numbers from the Chetty study are below.
Portion from Top/Bottom 20% at Highly Selective Publics (Chetty Study)
1 . UCSD – 40% from top 20% income, 11% from bottom 20% income
2. UCI – 45% from top 20% income, 9% from bottom 20% income
3. UCLA – 48% from top 20% income, 8% from bottom 20% income
3. UCSB – 48% from top 20% income, 8% from bottom 20% income
5. Florida-- 48% from top 20% income, 6% from bottom 20% income
6. UCB – 54% from top 20% income, 7% from bottom 20% income
7. UT Austin – 56% from top 20% income, 6% from bottom 20% income
8. GeorgiaTech – 59% from top 20% income, 5% from bottom 20% income
9. UNC: CH – 60% from top 20% income, 4% from bottom 20% income
10. Michigan – 66% from top 20% income, 4% from bottom 20% income
11. UVA – 67% from top 20% income, 3% from bottom 20% income
12. William and Mary – 73% from top 20% income, 2% from bottom 20% income
While demographic distribution of class are certainly important priorities for a public college like the UCs and were no doubt considerations in the recent policy changes, assuming the sole purpose of admission changes is either to increase low SES kids or to increase URM kids is a far too simplistic way of looking at college admission. There are many other relevant considerations and factors in the decision.
I don’t think anyone said it was the “sole purpose,” but it most certainly is a very important consideration. The increase in the number of low income admitted students in 2021 (10%) is about the same as the increase in admitted URMs (9%). My guess is there is substantial overlap between these groups.
And while you may view these tests as extremely helpful in creating your favored class profile, the UC system apparently views the tests as more hindrance than a help when it comes to fulfilling the UC’s educational mission.
As for bias correction, it could be (and is) used to address the shortcomings of factors which may “continue to disadvantage” low SES students, such as HSGPA or EC access. It may be that the UC system has determined that, in comparison, trying to correct the test score bias is unproductive and/or inconsistent with its mission.
Also, aren’t these tests supposed to provide a “standardized” comparison which transcends such factors as SES? If schools need to adjust the test scores up at the low end of SES and/or down at the high end, and the tests don’t tell us much that we don’t already know in the middle, then the value of "standardization” is illusory.
I would also add that CA has a much better financial aid program for low SES students than some of the other selective publics listed. Even with UT Austin admit of top x% across the state, the proportion of low income students is much smaller than CA. There are also geographical considerations within the state. William and Mary and UVA are located farther from the large metro areas, unlike the UC’s .
I teach at a regional university serving a large low-SES population. Transportation costs are a big deal to my students, when income level is that low. Also, ability to work part-time etc. are other considerations. Students of mine who were top students in their high school have all these factors to consider. Even with good financial aid, it’s just not easy to make the $$ work to go to a more selective university that’s farther away. These are not just hypothetical observations - they come from conversations with many of my students.
If, according to some of you, standardized test scores, compared to other admission factors, disadvantage students of lower-SES the most, wouldn’t it make the most sense to correct for that disadvantage directly?
This also has the advantage of at least not disadvantaging (but actually helping) students of lower-SES who have high test scores (there’re lots of them around the country, BTW).
There’s no suggestion to adjust scores of students who aren’t in the low-SES category. Also, any score adjustment for students of lower-SES probably should be done based on their intended majors and the UC campuses they apply to. For example, much smaller boost may be offered to UCB EECS majors.
It’s generally not SES itself that gives someone an advantage/disadvantage on SAT/ACT. It’s more things that are correlated with SES. For example, suppose a low SES kid attends a top private middle/high school via scholarship that has rigorous classes that well prepare students for the test, has SAT specific prep classes in the HS, encourages students get extra time on SAT/ACT, encourages students to prep more and retake if score isn’t high enough, etc. That student would probably have various advantages over the average kid taking the test, in spite of being lower SES. The system described in the Hout report that gives an apparent greater boost for “lives and learns in an environment with limited academic opportunities, relative to the Berkeley applicant pool” than low SES would be less likely to flag the kid attending a top HS as getting a larger boost than a more simple low SES = +num on test score system. Other admission criteria like ELC (appeared to be 2nd most influential after grades) might also favor the low SES kid from a less resourced HS over the lower SES kid from the more resourced HS.
Lower SES also often impacts more components of the application than just scores, so it wouldn’t make sense to act as though it is a simple relationship that only impacts one admission factor. Recall the general public reaction when the CollegeBoard tried to make a direct adjustment to SAT/ACT score via their “adversity index” a couple years ago.
There is also the issue about how little weight test scores received in admission. As listed in my post above, SAT I scores appeared to be among the least influential factors in UC admission decisions. If test scores are among the least influential factors in admission, then giving a (small?) direct adjustment to test scores is unlikely to have a large impact on admission decisions. However, giving a boost to overall read score for lower SES has the potential for a larger impact on admission decisions.
“There are many other relevant considerations and factors in the decision.”
Well, of course, and UC already takes other relevant factors into consideration in it holistic application review. Clearly, UC could add more factors if they thought it would add value to the process. What they cannot take into consideration is race.
SAT/ACT tests are expensive for students and don’t add that much additional admissions value (per UC), so it makes sense for UC to drop them, particularly if such tests discourage persons of color from admissions.
If UCs don’t want to give that student an additional boost, it’s very easy for them to take into consideration of whether the student has already received a “boost” due to the student’s privileged primary and/or secondary education.
The score adjustment I’m talking about isn’t a public one. Each college (and/or department) can make an approriate adjustment based on the particular student in relationship to the rigor of the major and the college.
I’ve made similar comments in other posts about the influence of location on portion of low SES. Colleges located in LA, NYC, El Paso, and similar tend to do better in SES distribution than comparable colleges located in less urban areas.
That said, many of the UCs are not centered near low SES areas. For example, I live and am familiar with the San Diego area, so I will use UCSD as an example. UCSD is located in the La Jolla area of SD. Near the time of the Chetty college SES distribution analysis, Money magazine did an analysis that compared the cost for a 2 bedroom home in various real estate markets. The UCSD area of La Jolla was ranked as the most expensive real estate market in the United States at that time. It isn’t in area with a high rate of low SES kids. It’s true that UCSD is located relatively near a high population city, but many UCs aren’t in an especially urban area compared to other colleges on the list. For example, comparing Santa Barbara to Ann Arbor, Ann Arbor has both a higher population and a higher rate of low SES kids (poverty rate is ~double). We also don’t see that it’s just the UCs in particular locations that have a larger portion of low/middle income kids than typical peers – it’s all of them, regardless of location. While location is an influential factor, I don’t think is the primary driving factor in the larger portion of low/middle income kids at UCs than peer colleges.
Regarding the suggestions that the UC schools should clandestinely adjust the scores of some students in order to preserve the use of these tests, I have to ask, to what end? How would preserving the tests in this manner better serve the mission of the UC system?
To answer it helps to set aside one’s own ideas on the ideal admissions process and consider the actual mission. According to the UC Senate Smarter Balanced Study Group (https://regents.universityofcalifornia.edu/regmeet/nov21/b3attach2.pdf):
UC has the responsibility to serve as many Californians as possible who meet our criteria for admission. However, present capacity levels of the institution, set by State budget allocations, mean that the university is unable to serve all the California high school students who merit admission. This reality makes it necessary to have an equitable admissions process that can identify students who are adequately prepared for and can benefit from the opportunities offered at the university. Equity is central to this effort, especially because educational attainment is one of the best predictors of lifetime earnings for individuals and college degree attainment remains the most effective means of ensuring social mobility. Also, as a state public institution, the UC is obliged to create a student body that is representative of the demographic profile of California.
Note that the mission is not to line up students up from most qualified to least qualified, so that the top 37k can be admitted. The mission is to choose between “adequately prepared” students based on factors beyond relative academic qualifications.
Which better serves this mission? Requiring test scores then clandestinely fudging them? Or dropping the requirement and focusing on other factors?
Which approach will more likely result in expanding the pool of low SES and underrepresented students “who are adequately prepared for and can benefit from the opportunities offered at the university?”
Which approach will more likely “create a student body that is representative of the demographic profile of California.”
Putting everything else aside, I don’t think officially adjusting SAT scores for certain applicant populations (as has been suggested) would pass the smell test. It seems like an approach that would be ripe for a lawsuit. It is easier to do away with standardized testing which leaves less room for folks to complain about being passed over by “less deserving” (i.e. lower scoring) applicants.
yes, it would be ripe for a lawsuit, but IFF it disadvantaged a legally protected class. For example, females, URM’s, disabled, and others.
No, not at all. Why would you think that?
They are meant to provide a standardized measure of skill, and skill is not equally distributed across SES ranges, for multiple reasons, but mainly because higher SES often means access to higher quality teaching and a peer group that values education.
The adjustment affects the mean scores, but the useful information is from scores that deviate from the mean. In other words, if a kid scores a 1350 from a poor performing school system where the mean is 900, that is very useful information. Therefore we cannot directly compare that to a student who attends a school system where the mean score is 1250.
Correct - but I never claimed it was. In my original post, I mentioned CA has robust FA for low SES students which may also be responsible for the higher percentage. This is all of course in addition to admissions factors.
For UT Austin, it’s auto-admit for the top x%, but we still do not see low-SES students at a level that such a policy would supposedly yield. I don’t know what’s going on there. Not enough FA? Not enough push from school GC’s? Something else? My own state has generous $$ towards tuition for college for low-SEs students, but there’s still a gap unless the student commutes to the university.
And my original post never claimed that UC admissions was the only factor in the relatively larger portion of low/middle income kids compared to peers. It sounds like we both agree UC admission system is one of several contributing factors.
Regarding robust FA, the average net cost among federal FA kids is below for the listed highly selective publics by income. UC seems to have a slightly higher average cost for low/middle income kids than the listed peers – not the highest and certainly not the lowest. So again while cost no doubt contributes, a low cost compared to peers doesn’t seem to be a primary reason for why UC has a larger portion of low/middle income kids than peers.
Average In State Cost for $30-48k, $48-75k, $75-$110k Income
Florida – $3k, $8k, $13k
Michigan – $6k, $10k, $18k
UNC: CH – $6k, $11k ,$18k
William and Mary – $5k, $12k, $20k
UC Average – $10k, $14k, $21k
GeorgiaTech – $11k, $15k, $18k
UVA – $12k, $!5k, $22k
UT Austin – $14k, $16k, $20k
As listed above, UT Austin was the most expensive of the listed colleges for all listed income ranges below the US median income, which no doubt contributes to why UT Austin does not appear to have as many low/middle income kids as the UC system, as you mentioned. Having near guaranteed seats at a flagship does not mean every low income kid who qualifies for top x% is going to apply to an unaffordable college that is a long drive from home, where he/she doesn’t know anyone who applies to or attends. Along the same lines, Florida’s especially low cost compared to peers no doubt contributes to why Florida has a larger portion of low income kids than all listed colleges except UCs.
Interesting. Does this include CA state grants? Or just federal?
It’s a federal reporting stat defined as follows:
Average net price is generated by subtracting the average amount of federal, state/local government, or institutional grant or scholarship aid from the total cost of attendance. Total cost of attendance is the sum of published tuition and required fees (lower of in-district or in-state), books and supplies, and the weighted average for room and board and other expenses.
Beginning students are those who are entering postsecondary education for the first time.
Title IV aid to students includes grant aid, work study aid, and loan aid. These include: Federal Pell Grant, Federal Supplemental Educational Opportunity Grant (FSEOG), Academic Competitiveness Grant (ACG), National Science and Mathematics Access to Retain Talent Grant (National SMART Grant), Teacher Education Assistance for College and Higher Education (TEACH) Grant, Federal Work-Study, Federal Perkins Loan, Subsidized Direct or FFEL Stafford Loan, and Unsubsidized Direct or FFEL Stafford Loan. For those Title IV recipients, net price is reported by income category and includes students who received federal aid even if none of that aid was provided in the form of grants. While Title IV status defines the cohort of student for which the data are reported, the definition of net price remains the same – total cost of attendance minus grant aid.
I don’t think it, thus the “supposed to.” But a number of posters have claimed that the advantage of using standardized tests is that scores are useful when directly comparing students from disparate high schools, such as a student “from a poor performing school system where the mean is 900” vs. a student in “a school system where the mean score is 1250.”
The reality is that many parents and test advocates claim that schools can and should make the exact type of comparison that you say “we cannot compare.” And if low SES kids are getting admitted with 1350 scores, then parents of rejected 1500+ scoring kids aren’t going to be keen on accepting your deviation from the mean explanation. Instead they are going angrily claim that the system is unfair and that their kid deserves admission because the “standardized” score is higher. If most people don’t understand or accept the distinction you propose, then it risks further exacerbating already growing tensions between various demographics.
And again, the question is, to what end? Why are test scores necessary to meet the UC mission described above? There are other mechanisms by which schools can identify standout students in generally low performing schools without the baggage of test scores.
Test scores are a component in a holistic admissions assessment at the schools that use them. Really, the question should be, how can the University of California rely only on GPA, which requires the UC AO’s to compare applications from the 1,300 high schools in California and the 26,700 high schools in the United States based solely on subjective assessments of employees of those high schools? It is worth noting that those making the subjective assessments of students at prep schools have a MASSIVE incentive to overstate the academic performance of their students.
Say whatever you want about the College Board, but they don’t give two spits who gets a 1500 and who gets a 1000 on the SAT.
Agreed. But that doesn’t really address whether using test scores advances the UC’s educational mission, does it? For example, how would requiring test scores expand the pool of low SES and underrepresented students who can benefit from the opportunities offered at the university? And how would requiring test scores help the UC’s create a student body that is representative of the demographic profile of California?
While the UC system doesn’t rely “only on GPA,” GPA has long been a important factor, although not quite in the manner you suggest. The UC system has no need to reinvent the wheel here. It was already identifying “adequately prepared” students based on test blind criteria even before it officially went test blind.
For example, the Eligibility in the Local Context (ELC) program guarantees UC admission to the top 9% at each California high school, based on UC GPA. In this program, grade inflation at Prep High doesn’t impact the kids at Rural High, because the students are only being compared to students at their own high school.
To answer your question of how can they do this, it may sound like an arduous undertaking from a logistical perspective, but the way it works is that each school periodically submits transcripts from the top 15% of their classes after junior year, and the UCs calculate a baseline GPA for each school. Again, this isn’t new. It is already in place.