The role played by the composite analogue of the log likelihood ratio in hypothesis testing and in setting confidence regions is not as prominent as it is in the canonical likelihood setting, since its asymptotic distribution depends on the unknown parameter. Approximate pivots based on the composite log likelihood ratio can be derived by using asymptotic arguments. However, the actual distribution of such pivots may differ considerably from the asymptotic reference, leading to tests, and confidence regions whose levels are distant from the nominal ones. The use of bootstrap rather than asymptotic distributions in the composite likelihood framework is explored. Prepivoted tests and confidence sets based on a suitable statistic turn out to be accurate and computationally appealing inferential tools. The Canadian Journal of Statistics 43: 18-41; 2015 (c) 2015 Statistical Society of CanadaResumeContrairement a sa version canonique, la version composite du log du rapport de vraisemblance joue un role moins central dans le calcul de tests d'hypothese et d'intervalles de confiance, puisque sa distribution asymptotique depend alors du parametre inconnu. Des arguments asymptotiques permettent de trouver des pivots approximatifs a partir de la vraisemblance composite. La veritable distribution de ces pivots peut toutefois differer passablement de leur reference asymptotique, conduisant a des tests et des intervalles de confiance dont les niveaux sont eloignes de leur valeur nominale. L'auteur utilise le reechantillonnage plutot que les distributions asymptotiques dans le cadre de la vraisemblance composite. Il constate que les tests d'hypothese et les intervalles de confiance prepivotes bases sur une statistique appropriee donnent des outils d'inference precis dont les proprietes calculatoires sont attrayantes. La revue canadienne de statistique 43: 18-41; 2015 (c) 2015 Societe statistique du Canada

### Prepivoting composite score statistics by weighted bootstrap iteration

#### Abstract

The role played by the composite analogue of the log likelihood ratio in hypothesis testing and in setting confidence regions is not as prominent as it is in the canonical likelihood setting, since its asymptotic distribution depends on the unknown parameter. Approximate pivots based on the composite log likelihood ratio can be derived by using asymptotic arguments. However, the actual distribution of such pivots may differ considerably from the asymptotic reference, leading to tests, and confidence regions whose levels are distant from the nominal ones. The use of bootstrap rather than asymptotic distributions in the composite likelihood framework is explored. Prepivoted tests and confidence sets based on a suitable statistic turn out to be accurate and computationally appealing inferential tools. The Canadian Journal of Statistics 43: 18-41; 2015 (c) 2015 Statistical Society of CanadaResumeContrairement a sa version canonique, la version composite du log du rapport de vraisemblance joue un role moins central dans le calcul de tests d'hypothese et d'intervalles de confiance, puisque sa distribution asymptotique depend alors du parametre inconnu. Des arguments asymptotiques permettent de trouver des pivots approximatifs a partir de la vraisemblance composite. La veritable distribution de ces pivots peut toutefois differer passablement de leur reference asymptotique, conduisant a des tests et des intervalles de confiance dont les niveaux sont eloignes de leur valeur nominale. L'auteur utilise le reechantillonnage plutot que les distributions asymptotiques dans le cadre de la vraisemblance composite. Il constate que les tests d'hypothese et les intervalles de confiance prepivotes bases sur une statistique appropriee donnent des outils d'inference precis dont les proprietes calculatoires sont attrayantes. La revue canadienne de statistique 43: 18-41; 2015 (c) 2015 Societe statistique du Canada
##### Scheda breve Scheda completa Scheda completa (DC)
2015
43
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in ARCA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: `https://hdl.handle.net/10278/5041280`
• ND
• 1
• 1