Standard fixed-n confidence intervals produced at different sample sizes on accumulating data can be contradictory with high probability. We show a simple way to compute, instead, mixture confidence sequences for regression coefficients in generalized linear models. Simulations attest the need of using these always-valid inferences if more observations become available over time.
Mixture confidence sequences for regression coefficients in generalized linear models
C. Di Caterina
;
2023-01-01
Abstract
Standard fixed-n confidence intervals produced at different sample sizes on accumulating data can be contradictory with high probability. We show a simple way to compute, instead, mixture confidence sequences for regression coefficients in generalized linear models. Simulations attest the need of using these always-valid inferences if more observations become available over time.
File in questo prodotto:
Non ci sono file associati a questo prodotto.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
