Abstract
It is observed here that bootstrap confidence intervals constructed under i.i.d. assumptions turn out to be 'conservative' under non-i.i.d. models, and thus they can be regarded as fall-back devices in the non-i.i.d. situations where exact inference is not available.
Original language | American English |
---|---|
Pages (from-to) | 67-75 |
Number of pages | 9 |
Journal | Journal of Statistical Planning and Inference |
Volume | 43 |
Issue number | 1-2 |
DOIs | |
State | Published - Jan 1995 |
ASJC Scopus subject areas
- Statistics and Probability
- Statistics, Probability and Uncertainty
- Applied Mathematics
Keywords
- 62G15
- Bootstrap confidence intervals
- Conservative inference
- Coverage probability
- Jackknife procedures
- Product-limit estimator