A Roost Rating is a single 0 to 100 number with a letter beside it. This page tells you what went into the one on your child's school, and how much of it you should believe.
What decides the grade
Two sources, and you can see which ones your school actually has on its own profile.
| Source | What it is | How much of the grade |
|---|---|---|
| The survey (AEAM) | Alberta's annual survey asking parents, students and teachers what the school is like day to day | 62.5% |
| The exams (PAT) | How the school's Grade 6 students did on Alberta's provincial exams, averaged over three years | 37.5% |
Those are the numbers for a school with both sources, which is 261 of the 406 graded schools on FamilyRoost today. The other 145 have exam results and no survey, and there the exams decide the whole grade. A missing source is never counted as a zero against a school.
The reverse case does not get a letter at all. A school with survey results but no exam results is shown as "pending", not graded, and there are currently 52 of them, most of them K-4 and K-5 schools whose students leave before Grade 6. We could compute a survey-only number, and we have decided not to publish one: a letter built on a survey alone would look identical to a letter built on two sources, and it would not be.
So check which sources your school has before you compare it to anything. Two schools showing the same letter can be resting on different evidence, and each profile says which.
On top of that sits Roost field research, which is not a percentage. It is a human-checked adjustment for things neither a survey nor an exam can see, logged with its reason. Most schools carry none.
The one worth looking for is a grade-configuration change. If a school is being shortened from K-6 to K-4, a family that buys a house on the strength of its grade may find their child displaced in two years. The academics have not changed; the decision you are making has. It is one of only two flags that can move a tier downward, the other being a documented school-internal incident.
Why the survey counts for more than the exams
Because it is the wider measurement. An exam result is one cohort, one week in May, four subjects. The survey asks about safety, engagement and how the school is run, from three groups who are there every day.
That does not make it clean, and the survey page is blunt about how it is not: the people who answer are not a random sample of the people who attend, and the three groups routinely disagree with each other by double digits. A grade resting mostly on a survey inherits both of those.
Inside the exam share we use two statistics of the same exams, 70% the equated share meeting the Acceptable Standard and 30% the average mark, for stability rather than accuracy (why that split). They are two readings of one test, not two independent sources, and blurring that distinction is exactly what we fault single-number rankings for.
The number we show but do not score
The Fraser Institute's /10 appears on school profiles. It does not move the Roost Rating. It is 0.975 correlated with the average exam mark we already calculate, so counting it would count one set of exams twice. We show it because it is the number most parents have already seen, next to our own reading of the same exams, with both dates, and the full argument is here.
From sources to a tier
The blended score is capped at 95 (no school is treated as "perfect"), then mapped to a tier:
| Tier | Name | Score |
|---|---|---|
| A+ | Very strong | 85.6 and up |
| A | Strong | 79.1–85.5 |
| B | Solid | 72.0–79.0 |
| C | Fair | 62.1–71.9 |
| D | Developing | below 62.1 |
Across the 406 graded schools today that works out to 34 A+, 109 A, 116 B, 98 C and 49 D. A missing score is shown as PND ("pending"), never as a D. If we don't have the data, we say so.
The same tie rule we apply to Fraser applies to us. Our number inherits the wobble in its ingredients: a single year of exam results can move about 11 points on a small cohort, and a survey year at a small school moves too. So treat two schools within about a point of each other as level, and treat a letter as a band rather than a ranking. A school at 79.2 and a school at 78.9 are the same school as far as this number can tell, even though one shows an A and the other a B.
Where we disagree with single-number rankings
Test-score rankings correlate heavily with neighbourhood income: they often measure the families a school enrolls more than the teaching a school does. That is the argument for keeping tests to a minority of the score, for surfacing survey and operational context alongside them, and for showing the Fraser number without letting it move ours. It is also why we cut a source that was mostly re-reporting a test we already read. The goal is a starting point for a conversation with a school, not a verdict on it.
This page explains methodology only. Specific school scores come from the live dataset and show their own data-freshness date on each school profile.