Home banner
A-Z Index

Quick way to the find the information that you need...

More button
Register with FRAME

Although you do not need to register, any information you provide will be confidential and used only by FRAME to improve the website

Register button
Account Login
Forgot password?

The Journal


Alternatives to Laboratory Animals - ATLA

Download latest issue button Download back issues button Subscribe to ATLA
Contact Us

Tel icon

Tel: +44 (0)115 9584740

Tel icon

Fax: +44 (0)115 9503570

Make an Enquiry

Bootstrap inference for local populations.

Lunneborg, C.E.

Drug Information Journal, 35(4), 1327-1342 (2001).

The randomized available case study, in which a nonrandom set of cases (patients, animals, laboratory runs) is randomized among two or more treatments, is a staple of biomedical research. Traditionally, such studies have been analyzed as though the cases were a random sample from an infinitely large population (1). The resulting statistical inferences address incorrect populations. More importantly, in the presence of response measurement error these inferences are inappropriate for the, correct populations, understating the differential impact of treatment (2). In this paper I develop and illustrate a nonparametric bootstrap approach to inference in such studies, an approach that is faithful to the local origins of the randomized cases and can account for the influence of measurement error.