UC admissions & high-school academic indicators

Each dot is a California high school. Compare any UC campus, any admission year or pooled period, and any indicator — academic strength (CAASPP, A–G), the unduplicated–pupil share (UPP), or the admissions funnel itself (apply → admit → enroll, against several denominators) — and what becomes of each school's entrants at UC: first-year retention and 4-/5-/6-year graduation. Search a school or click a dot for its full profile.

Companions: neighborhood context & admissions — census-tract measures of each school's surroundings (income, education, poverty), and how much of the achievement–admissions association they carry; and outcomes over time — graduation, retention, time to degree, majors, earnings and debt across entering cohorts, split by group, campus, source school, or GPA band.

30
scroll zoom · drag pan · dbl-click reset

Schools shown
Overall
Student-weighted
LCFF+ vs other (gap)
Correlation over time
Click a dot (or search a school) to see its full profile.
How to read this — and important caveats
  • Dots are California high schools, drawn one uniform size by default. Tick Dot size → Scale by sample size to size each dot by the denominator of the metric you're viewing — applicants for admit rate, admits for yield, A–G eligible for per–eligible rates, grade 9–12 enrollment for per–head rates, cohort for A–G completion — so a bigger dot means a larger base for that rate. Only schools at or above Min applicants are shown (a fixed sample floor). Scale ranges are padded by the largest dot radius so whole dots stay in frame.
  • Reading the rates. All outcome rates are shown as proportions (an admit rate reads 0.241, a yield 0.45) so the eight funnel ratios share one scale; the academic/context measures (CAASPP, A–G completion, UPP, ELWR) stay as the percentages they natively are. The admit rate is to the selected campus only (admits÷applicants there), not a UC system–wide rate.
  • “Mean across schools” (the dashed line and headline figure) is the arithmetic mean of the per–school rates — the central tendency of the dots, which is what this tool measures. The separate Student–weighted figure is the pooled Σnumerator÷Σdenominator rate, dominated by large schools and shown only for reference; the LCFF+ gap is likewise a difference of school–level means.
  • UPP & LCFF+. The axis labelled UPP is the Unduplicated Pupil Percentage — the CALPADS/LCFF share of a school's students who are low–income (free– or reduced–price–meal eligible), English learners, or foster youth, counted once if in more than one category, and so broader than a household–poverty rate. LCFF+ marks schools where that share exceeds 75% of enrollment — the Local Control Funding Formula's high–need threshold for additional support.
  • Year / period. Pooled periods sum the funnel across their years; single years show one cycle. Every rate is computed as a ratio of sums over years where both numerator and denominator are observed, so it is comparable across periods of different length.
  • Explore the cloud. Scroll to zoom, drag to pan, double–click to reset — handy for spreading piled–up dots enough to click one. Click the trend line to bring it in front of the dots (click again to send it back). Focus mode (top–right) hides the control bar and side panels for a clean, full–width view. You can divide any funnel stage by A–G eligible or by grade 9–12 enrollment via the metric menu.
  • Drop outliers. Ticking the box removes, for the current view, any school flagged by the standard Tukey 1.5×IQR rule on either plotted axis (the single metric in strip view). Dropped points leave the chart, the correlation r, the trend line, the mean and the LCFF+ gap; the number removed is reported. Because the headline mean is outlier–sensitive, this is the intended way to set extremes aside.
  • Headline r vs. the over–time panel. The headline correlation pools the selected period into one rate per school, then correlates across schools. The lower–right panel instead computes a separate r for each year and spans the full record — so the two measure different things: pooling several years per school damps year–to–year noise and can leave the pooled r stronger (more negative here) than any single year. The dashed line on the panel marks the pooled value for the selected period, with the axis widened to include it, so the two can be read together.
  • Base counts. A selected school's profile lists the raw denominators behind its rates — A–G eligible, grade 9–12 enrollment and graduating cohort — and every rate row also shows its own numerator ÷ denominator.
  • Cross-universe ratios. Applicant/admit/enrollee counts come from UC (CEEB); “eligible” (A–G completers) and enrollment come from CDE (CDS). Ratios mixing them are school-level indicators, not exact per-student rates.
  • Self-selection. A negative academic–admit correlation need not, on its own, reflect a school-level effect; the over-time view and within-LCFF+ comparisons are the checks against it.
  • Suppression & reliability. UC masks small admit/enroll cells; at Berkeley ~a third of applicant schools have suppressed admits. When the numerator is masked in the years carrying most of a school's applicants or admits, a pooled rate would otherwise collapse onto a small surviving year (e.g. an apparent 1.00 yield, or a 0.50 admit rate from 3 of 6 in one year while 31 applied over the period). Such rates are marked low-confidence: drawn as hollow dashed dots, excluded from the headline r, the mean and the LCFF+ gap, and shown in a school's profile as their actual numerator ÷ denominator with the number of suppressed years.
  • Strip position. In the strip view each dot sits exactly at its value on the axis; only vertical spacing is used to separate overlapping schools, so a dot's horizontal position always matches the number in its profile.
  • Newcomer schools. A few schools for recently arrived immigrant students (e.g. Oakland International) show ~0% CAASPP proficiency while still posting A–G completers — students meet graduation requirements while acquiring English. That 0% is real data (it may round from a small masked percentage), not a missing value.
  • Impossible denominators. A handful of schools have a corrupted grade 9–12 enrollment far smaller than their applicant count (e.g. Wheatland). Rows where a funnel count exceeds total enrollment are dropped from the “÷ enrollment” metrics rather than plotted as >1.
  • Capacity dynamic. High-admit inland campuses (Merced, Riverside, Santa Cruz) show a positive academic–admit correlation — they admit most qualified applicants — so read them differently.
  • ELWR met at entry. The writing–requirement figure is the share of a school's UC enrollees who satisfied the Entry Level Writing Requirement at entry — via the AWPE exam or other qualifying methods (AP/IB/SAT/ACT/transferable course) — i.e. who did not need first–year remedial writing. It is not “ever met”: enrollees only (self–selected, small N), latest year per school, indicative only.
  • SAT / ACT. Where available a profile shows the school's UC–enrollee average SAT reading (EBRW, 200–800) and ACT (writing–section takers, 1–36) for the latest year on record. These exist only for ~2006–2020 (none after 2020), are enrollee–only, small, self–selected samples — context, never an academic axis. SAT “writing” appears only on its old 200–800 section scale, as the post–2016 essay is not comparable.
  • Applicant & admit GPA. The two GPA axes are UC's own recalculated weighted, honors–capped high–school GPA (sophomore–junior A–G coursework — the figure UC uses in admissions), averaged over the selected period. Applicant GPA gauges the strength of a school's UC applicant pool: an applicant–specific measure, distinct from CAASPP (which covers all grade–11 students). Admit GPA is the average among admitted students — the GPA bar UC actually accepted — and is partly downstream of the admit decision (a more selective campus shows a higher bar), so read it as a level UC set, not a pure measure of school strength. A profile also shows enrollee GPA and the admit−applicant gap (how far above the pool the admitted students sat). Admit GPA is blank wherever UC suppresses a school's admits.
  • UC graduation rates by high school. From UC's freshman-outcomes dashboard: for each school and fall entry cohort, the share of its UC freshman entrants (all nine campuses pooled) who persisted one year and who graduated within 4, 5 or 6 years. A cohort appears only when the school sent ≥10 entrants; rates are integer percents; 6-yr rates are mature through entry 2019, 4-yr through 2021, retention through 2024. Two ways to use them, matching two hypotheses: (1) as an academic axis — UC 6-yr grad (prior cohorts) pools, N-weighted, the five cohorts entering 6–10 years before the selected period: the most recent rates fully observable — and knowable to admissions readers — by decision time. If graduation reveals something fixed about a school's students, it is an academic proxy; if readers calibrate on prior graduates, it is the very signal they act on. The over-time panel shows the correlation re-anchored year by year. (2) as an outcome — the retention/graduation rates of the period's own entrants (e.g. vs CAASPP: does school-level academic strength predict success after enrollment?). The same-period rates are also offered on the X-axis: though measured after admission, a cohort's eventual completion can be read as an indicator of academic strength already present when it applied. Scope. The two roles use different series. As X-axis indicators, grad metrics pool the school's entrants to any UC campus (richest coverage; a school-level trait), so a school can carry one even when its admits or enrollees to the selected campus are zero or suppressed. As Y-axis outcomes, retention/graduation are campus-specific — entrants to the selected campus only — since an outcome of this campus's admissions should use this campus's students. The per-campus dashboard series reports a school-year only when ≥10 students entered that campus, so campus-specific outcomes show far fewer (and larger) schools. Profiles show both blocks with their entrant counts, and funnel counts suppressed in every year display as “—” rather than a false 0. Dot size for grad metrics is the entrant-cohort N. Campus filter does not change the All-UC grad series; a profile additionally shows the campus-specific 6-yr rate where reported.