Star Ratings, Ranked Lists, One Number Per School. How to Read What Is Underneath

A single score attached to a school or a training program is doing a lot of work. It is compressing test results, attendance, graduation, sometimes college enrollment or wages, into one figure that can be sorted in a column. That compression is not dishonest, but it hides the trade-offs someone made on your behalf about what matters. If you are about to move for a school district, commit a teenager to four years, or spend money on a certificate program, it is worth knowing which decisions were baked in before you ever saw the number, and which parts of the underlying data you can pull for yourself in an afternoon.

How one number per school became the standard

The habit of scoring a school on a single scale is younger than most parents assume. For roughly the first decade of this century, federal accountability rested mainly on proficiency: the share of students at each school clearing a fixed bar on a state test, reported for the school and for subgroups within it. That design had a clear logic. It made low performance impossible to hide behind a school average, and it forced districts to report results for students who had previously been invisible in the aggregate. It also had a predictable side effect. A proficiency rate measures where students already are far more than what the school did with them, and schools serving more affluent, more stable populations score well on it almost automatically.

The correction was growth measurement. Instead of asking what share of students cleared the bar, growth models ask how much a student gained relative to students who started at a similar place. When federal law was rewritten in 2015, states were required to build accountability systems from several measures rather than one, and most of them added a growth measure, a graduation or persistence measure, an English learner progress measure, and at least one indicator of school quality that was not a test. The U.S. Department of Education oversees the approval of those state plans, which is why the ingredients differ from state to state while the general shape is similar everywhere. The commercial ranking sites and real estate listings that show you a score are almost always reweighting these same state-published ingredients into a formula of their own.

What changed after 2020, and why the older numbers stopped comparing cleanly

Testing was suspended in spring 2020 and heavily disrupted in 2021, with participation in some places falling far enough that results were flagged as not comparable to prior years. That created a hole in the middle of every trend line. Growth measures were hit hardest, because growth requires a prior-year score for the same student, and a missing year breaks the chain. Many states did not publish full ratings for two or three cycles, and when they resumed, the baseline had shifted underneath them.

Two quieter changes matter as much. First, chronic absenteeism moved from a footnote to a headline indicator in most state dashboards, usually defined as missing about ten percent or more of enrolled days for any reason. It rose sharply after 2020 and has come down unevenly, which means a school's rating can move several points on attendance alone while instruction has not changed at all. Second, enrollment itself became unstable. Cohorts shrank, families moved, and some schools now report results for a student body that is meaningfully different from the one measured three years earlier. A score built on a five-year trend at such a school is averaging two different populations.

The practical consequence is that a ranked list is a reasonable place to build a candidate list and a poor place to end one. If you are working from a roundup of the best schools in la or any comparable metro, treat it as a set of names worth investigating rather than an ordering you should trust to the decimal.

Three numbers to pull yourself

Every state publishes a school report card, and every one of them lets you see the components behind the composite. Pull these three, for the two most recent years available.

• Growth, separated from proficiency. Look for them side by side. A school with middling proficiency and strong growth is teaching effectively and enrolling students who started behind. A school with high proficiency and flat growth may be coasting on who walks in the door. If you are choosing for a specific child, growth is the closer proxy for what the school will do for them.
• Results for the subgroup your student belongs to. School-level averages routinely conceal a twenty-point spread between subgroups. Check students with disabilities, English learners, and any economic or racial subgroup relevant to your family. Where a subgroup is too small to report, note that as a gap in your information rather than as reassurance.
• Chronic absenteeism and teacher stability. Absenteeism tells you whether the building is functioning day to day. Staff turnover, often reported as a retention rate or a share of first-year teachers, tells you whether the program described in the brochure will still be staffed in two years.

Write the numbers down on one sheet, one row per school. The comparison is more useful in your own handwriting than in any dashboard, because you chose the columns.

Courses and training programs run on different numbers

For a certificate, associate degree, or trade program, the published figures that matter are completion rate, licensure or certification pass rate, and earnings or placement after exit. Each has a specific failure mode. Completion rates are often quoted only for first-time, full-time students, which can be a minority of the actual enrollment. Pass rates are sometimes reported for first attempts and sometimes cumulative, and the difference is large. Placement claims can count any employment rather than employment in the field, and can be measured at a point close enough to graduation that it captures the job the student already had.

Ask for the denominator in writing. "Of the students who started in the fall 2022 cohort, how many finished, how many sat the exam, and how many passed on the first attempt." A program that keeps clean records will answer in a day. Then check the credential independently: look the accreditor up in the recognized directory rather than accepting the seal on the website, and confirm with the state licensing board that the program's hours count toward the license you actually want. That last step is the one people skip and later pay for twice.

The part not to do alone

Data gets you to a shortlist of three or four. It cannot tell you whether the schedule works for your commute, whether the special education team returns calls, whether the shop equipment is the equipment employers in your area use, or how the eleventh graders talk to each other in a hallway. Visit during a normal week, not an open house. Ask to speak with a parent or a recent graduate whom the school did not select for you. If money is committed on the strength of a contract, an enrollment agreement, or a refund policy you do not fully follow, have someone who reads contracts for a living read it before you sign, and ask specifically what happens to your money if the program closes mid-term.

The numbers are better than they were, and the dashboards behind them are more honest about their own limits than the ranked lists built on top of them. Use the composite to narrow, use the components to compare, and spend your last week on the questions no dataset was ever going to answer.