A Quiet Change to the Lowell Admissions Process
A change in how SBAC scores are treated will reset the balance between private and public school applicants
There’s a story by Jill Tucker in the SF Chronicle claiming that private school students could have an edge in Lowell admissions this year. Here are the key paragraphs.
When the public school students took the state test, they were still reeling from the pandemic and didn’t know the exam would count for Lowell admission because the school board didn’t restore the merit process until after the academic year ended in June, said Rebecca Johnson, Lowell social studies department chair, in a letter signed by all Lowell department heads to district officials.
The educators have requested all students — public and private — be allowed to take the Lowell admission test in January, a request so far denied by district officials given costs, staffing and concerns about access.
I haven’t seen the letter. If that is indeed its substance, the argument is weak. The state test referred to in the article is the SBAC taken by all 7th graders in California public schools. The SBAC is a consequence-free test for the vast majority of students. Their performance depends on their ability and their internal motivation. Whether they do well or poorly has no effect on their future educational journey. The letter appears to be arguing that San Francisco students would have done better than they did if they knew the result counted for Lowell admission. But is it really fair to compare students for whom the result is important (e.g. those applying to Lowell) with those for whom the result is unimportant i.e. the vast majority of other public school students in California? Don’t we actually get a more accurate view of a test taker’s abilities if all the test takers have to rely on their internal motivation?
It’s possible that the article misses the substance of the letter. There has been a significant change to how standardized tests are treated in the admissions process that does change the balance between public and private school applicants and that was not, so far as I know, debated at any SFUSD board meeting or even publicized.
For those who aren’t familiar with it, here is a summary of Lowell’s three-band admissions system. Both the article and this post are going to focus just on Band 1, which is the only band where admission is based entirely on academics, and in particular on the 25 points that are allocated based on the results of standardized tests.
Background
Back in 2015-16, every applicant took the Lowell Admission Test which is a standardized test called the Terra Nova. The cut-off score for Band 1 varied a little from year to year. In 2015-16, it was 86.5 points out of a possible 89. A straight-A student would need to score in the 85th percentile in Math and the 89th percentile in Reading to get in (64 points for all those ‘A’ grades; 11 points for the 85th percentile score in Math; 11.5 points for the 89th percentile score in Reading). A student who got three As and one B in 8th grade would have to score in the 93rd percentile in Math and the 97th in Reading (or vice versa) to get in.
Advocates believed that having to take a separate admissions test reduced the number of eligible Black and Latino applicants to Lowell. This is the “access” point mentioned by the district in the Chronicle article above. The advocates argued that the applicants’ prowess could be measured just as well by the SBAC tests that students already took in 7th grade1. It also saves money for the district if they don’t have to pay for more Terra Nova tests to be scored. Private school students don’t take the SBAC so the Terra Nova test was retained just for them.
This posed a challenging problem. How could the district compare students when some of them had taken the SBAC and others the Terra Nova test? Percentiles on standardized tests cannot be directly compared. For example, many San Francisco middle schoolers take, or used to take, the SSAT (for admission to independent high schools) and the HSPT (for admission to Catholic high schools) in addition to the Lowell test and annual SBAC tests. It is well known that, since the population who apply to independent schools is very different than the population who take SBAC tests, a student who scores in the 95th percentile on the SBAC will typically score much lower on the SSAT - the SSAT is a “harder” test than the SBAC. The straightforward solution would be to compare the scores of people who took both the SBAC and the Terra Nova test and establish a rough equivalence. There would have been thousands of such students in the years after the SBAC was introduced and before the Lowell admissions system changed. By comparing their results on the two tests, the district could figure out what SBAC score was equivalent to Terra Nova percentiles.
Here is a portion of the admissions scoring table for 2018-19 admissions:
It appears at first glance as if they’ve established the equivalence I mentioned. As we can see, the scoring system was unchanged for private school students. To get 12.0 points in Math, a private school student would continue to need to score in the 93rd percentile. A public school student would now need to score at least 2646 on the Math portion of the SBAC. A natural assumption, and one that I made initially, was that 2646 was equivalent to being in the 93rd percentile. But it isn’t. The state doesn’t publish precise conversion tables for mapping between scale scores and percentiles but it does publish decile cutoffs. In 2018-19, for example, among 7th graders, the 80th percentile in Math was at 2632 and the 90th percentile was at 2683 so a score of 2646 would have been around the 83rd or 84th percentile.
The same is true for English. To get 12.0 points in English, a public school student needed to score a minimum of 2643 which is in the 80th or 81st percentile (the 80th percentile for 7th graders in 2018-19 was 2642 and the 90th percentile was 2683).
In other words, students coming from public school were held to a much lower standard than private school students and to a much lower standard than public school students from prior years. In theory, it could be that the SBAC is a much harder test and scoring in the 84th percentile on the SBAC is equivalent to scoring in the 93rd percentile on the Terra Nova. But the proof that the public school students were being held to a much lower standard is that the score required for admissions in Band 1 jumped suddenly. It was 86.5 in 2015-16, and 87.0 in 2016-17 then jumped to 88.0 in 2017-18 and stayed at 88 in 2018-19 and 2019-20 and 2020-21. Public school students didn’t suddenly get much smarter in 2017-18 but hundreds of extra students were suddenly obtaining 88 or more points because the scoring system made it easier for them to get those points.
A straight-A private school student could have obtained admission prior to 2017-18 by scoring in the 89th percentile in both Math and Reading, thereby earning 87 points. Since then, a straight-A private school student has needed to score in the 93rd percentile in both Math and Reading to get the required 88 points. Meanwhile, a public school student could now obtain admission by scoring as low as the 81st percentile in ELA and the 84th percentile in Math.
Here is the same table for 2019-20 admissions. The points mapping is identical to the 2018-19 table.
Here is the same table for 2020-21 admissions. The points mapping hasn’t changed.
This points table has other problems beyond the different standards applied to public and private applicants. Most notably, it is too unforgiving of a blemish on the student’s grade report. A single ‘B’ in 8th grade is enough to cause a student to miss admission. I would argue that a student who scores in the 99th percentile on Math but only gets a ‘B’ in English (perhaps due to being an English learner) is much much more suitable for Lowell than a straight-A student who scores in the 85th percentile on both.
Admissions for 2021-22 and 2022-23 were done by lottery. After the recall election, the board voted to revert to a merit-based admissions system. Let’s look at how the points are being allocated for the current admission cycle which is for the 23-24 academic year but uses the 2022 SBAC scores.
Notice that the scale scores required to obtain particular numbers of admissions points have changed. Previously, a student needed a score of 2646 in Math and 2643 in ELA to obtain 12 points in each of those subjects. Now those scores would be good enough for only 10.5 points and the standard to obtain 12 points has risen to 2697 in Math and 2693 in ELA.
What percentiles do these scores map on to? Here are the scale score percentiles for 2021-222. A 90th percentile score for ELA is 2682 which is good for 11.5 points, the same as a 90th percentile score on the Terra Nova test. A 90th percentile score for Math is 2668, which is good for 11.0 points compared to the 11.5 earned by a 90th percentile score on the Terra Nova test. It seems that, in the new points table, the percentiles required of public school students are now a little bit higher than those required of private school students. That's not necessarily wrong: the goal should be to apply the same standard to public school and private school applicants and if the Terra Nova test is in fact a bit harder than the SBAC, this points table might be the fairest possible table. It is perplexing, however, that there was no public notice or explanation or board debate about this change, or at least none that I have seen. Why did they decide to change the table this year? How did they come up with these particular point allocations?
[Update: 14-November. I figured out why they had to change the table this year. The highest and lowest achievable scale scores changed. In 2018-19 and preceding years, the highest achievable scores in grade 7 were 2745 in ELA and 2778 in Math. Now they are 2810 in ELA and 2820 in Math.]
A couple of predictions:
The new points table will lower the admissions cutoff back to something like 86.5 or 87. That is a positive change because it makes the process more forgiving: students can now afford a ‘B’ in one subject if their standardized test scores are high enough.
It will lead to an increase in the number of private school applicants gaining admittance to Lowell but a smaller increase in the number who actually choose to attend. Lowell is a backup option for many private school applicants - their first choices tend to be private high schools, most of which are at all time enrollment highs. Last year’s huge budget cut at Lowell has made the school less attractive to parents who are used to paying for their kids’ education.
It’s not generally possible to use 8th grade SBAC scores because the SBAC is administered in the Spring, well after high school offers have gone out.
The impact of the pandemic on scores is not as straightforward as you might think. Although average SBAC scores were lower in 2021-22 than in 2018-19, the scores of the best students actually rose. In ELA, the median dropped by 6 (2555 to 2549), the 80th percentile dropped by 3 (2642 to 2639), the 90th percentile dropped by 1 (2683 to 2682), and the 99th percentile actually rose by 35 (2745 to 2780). In Math, the median dropped by 20 (2527 to 2507), the 80th percentile dropped by 19 (2632 to 2613), the 90th percentile dropped by 15 (2683 to 2668) and the 99th percentile again bucked the trend and rose by 19 (2778 to 2797).
It seems reasonable to me that SFUSD would favor previous students when granting admission to Lowell. This unannounced change is troubling.