Student Growth Percentiles
Also known as: SGP, Conditional Status Percentiles, Betebenner Growth Percentiles, Quantile-Regression Growth Model
Student growth percentiles (SGPs) describe how much a student grew academically relative to peers with similar score histories. Introduced by Damian Betebenner in 2009, the method fits a series of conditional quantile regressions of a current test score on prior scores, then reports each student's growth as the percentile rank they occupy within the distribution of students who had the same starting point. A student at the 70th growth percentile grew faster than 70 percent of academic peers, regardless of their absolute achievement level.
Key highlights
- Separates growth from status, so low-achieving students who grow rapidly are recognized rather than penalized for absolute level.
- Highly interpretable: a single percentile communicates relative growth to parents, teachers, and policymakers without statistical training.
- Distribution-free in the response and robust to outliers because quantile regression does not assume normal residuals or model the mean.
- Supports forward projection of the growth needed to reach proficiency targets, linking normative growth to criterion-referenced goals.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use SGPs when a state or district has at least two occasions of vertically comparable assessment scores per student across a large population and wants a normative, easily communicated description of growth that is decoupled from starting achievement. SGPs are the backbone of many U.S. state accountability and growth-reporting systems. They are descriptive by design: they summarize where a student fell in the conditional distribution, not why, and they do not by themselves isolate the causal contribution of a teacher or school. When the goal is a defensible causal effect estimate, a value-added model with explicit controls is the more appropriate tool, and the two are often reported side by side.
Strengths & limitations
- Separates growth from status, so low-achieving students who grow rapidly are recognized rather than penalized for absolute level.
- Highly interpretable: a single percentile communicates relative growth to parents, teachers, and policymakers without statistical training.
- Distribution-free in the response and robust to outliers because quantile regression does not assume normal residuals or model the mean.
- Supports forward projection of the growth needed to reach proficiency targets, linking normative growth to criterion-referenced goals.
- Purely normative and descriptive: a growth percentile is defined relative to the peer cohort and is not a causal effect of any teacher, program, or school.
- Requires vertically comparable scales across grades and large samples to estimate stable conditional quantiles, especially in the tails.
- Median SGPs aggregated to teachers or schools inherit measurement error and nonrandom student assignment, complicating accountability use.
- Conditioning only on prior test scores omits other determinants of growth, so apparent differences across groups may reflect omitted factors.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
How do student growth percentiles differ from value-added models?
SGPs are normative and descriptive: they place each student in the conditional distribution of peers with the same score history and report a percentile, without claiming to isolate a causal effect. Value-added models aim to estimate the causal contribution of a teacher or school to growth, typically using mean-based regression with additional covariates. SGPs answer 'how fast did this student grow relative to similar peers,' while VAMs answer 'how much did this teacher add.' Many systems report both. See the related Value-Added Modeling entry.
Why use quantile regression instead of an ordinary regression on prior scores?
Quantile regression estimates the entire conditional distribution of current scores rather than just the conditional mean, which is exactly what a percentile interpretation requires. It is also robust to non-normal residuals and heteroscedasticity, common in achievement data, and it lets the relationship between prior and current scores differ across the achievement range rather than assuming one slope.
Can SGPs be aggregated to evaluate teachers or schools?
Median or mean SGPs are routinely aggregated, but this use is contested. Aggregates inherit individual measurement error, depend on the number of students, and can be confounded by nonrandom assignment of students to teachers. Treating a teacher's median SGP as a clean effect estimate ignores these issues; growth percentiles were designed primarily as a normative description of student progress.
Sources
- 1.Betebenner, D. W. (2009). Norm- and criterion-referenced student growth. Educational Measurement: Issues and Practice, 28(4), 42–51.
- 2.Koenker, R. (2005). Quantile Regression. Cambridge University Press.ISBN 9780521845731
You have read it. What now?
Cite this page
ScholarGate. (2026, June 22). Student Growth Percentiles. ScholarGate. https://scholargate.app/education/student-growth-percentiles