Generalized Estimating Equations (GEE)
Generalized estimating equations estimate population-averaged regression effects for correlated data, such as repeated measures or clustered observations, using a working correlation structure and robust sandwich standard errors.
일반화추정방정식은 작업상관구조와 강건한 샌드위치 표준오차를 사용하여 반복측정이나 군집 자료처럼 상관된 자료에서 모집단 평균 회귀효과를 추정하는 방법입니다.
How do I run a regression on repeated measures or clustered data with a binary or count outcome?
결과변수가 이진형이나 빈도형인 반복측정 또는 군집 자료로 회귀분석을 하려면 어떻게 하나요?
Generalized estimating equations extend generalized linear models to correlated observations by specifying a working correlation within clusters and using robust standard errors that remain valid even if that correlation is misspecified. They fit when the interest is in population-averaged effects for continuous, binary, or count outcomes measured repeatedly or within clusters.
When to use Generalized Estimating Equations (GEE)
- Binary or count outcomes measured repeatedly on the same subjects
- Patients clustered within hospitals or students within classes
- Population-averaged effects are the target of inference
When to use something else
- Subject-specific effects or variance components are needed: use a generalized linear mixed model
- Very few clusters: use a mixed model or small-sample corrected standard errors
Assumptions and requirements
- Correctly specified mean model and link function
- Clusters independent of one another
- Enough clusters for robust standard errors to be reliable
- Data missing completely at random, or weighted GEE otherwise
What the analysis reports
- Coefficients or odds ratios with robust standard errors
- Wald tests and confidence intervals
- Estimated working correlation
- QIC for comparing correlation structures
How to report it
In a GEE model with an exchangeable working correlation, [predictor] was associated with higher odds of [outcome], OR = [value], 95% CI [[value], [value]], p = [value].
Related methods
- Generalized Linear Mixed Model
- Linear Mixed-Effects Model
- Logistic Regression
Run Generalized Estimating Equations (GEE) on Quorant
- Data you need
- a table (CSV or Excel) with an outcome variable, a cluster ID and predictor (feature) columns
- Price
- 400 credits for one analysis up to 10,000 rows (complexity level 3); larger or wider data costs more. Optional methodologist review: 1800 credits.
- Checking
- Python results are cross-checked against R reference outputs in every deployment
- Before you pay
- A free pre-analysis diagnosis checks your data and gives an exact quote; nothing is deducted until you approve it.