statistics review 5 comparison of means统计评估5比较的意思.pdf

Critical Care October 2002 Vol 6 No 5 Whitley and Ball Review Statistics review 5: Comparison of means Elise Whitley1 and Jonathan Ball2 1Lecturer in Medical Statistics, University of Bristol, Bristol, UK 2Lecturer in Intensive Care Medicine, St George’s Hospital Medical School, London, UK Correspondence: Editorial Office, Critical Care, editorial@ Published online: 12 July 2002 Critical Care 2002, 6:424-428 This article is online at /content/6/5/424 © 2002 BioMed Central Ltd (Print ISSN 1364-8535; Online ISSN 1466-609X) Abstract The present review introduces the commonly used t-test, used to compare a single mean with a hypothesized value, two means arising from paired data, or two means arising from unpaired data. The assumptions underlying these tests are also discussed. Keywords comparison of two means, paired and unpaired data, t test Previous reviews in this series have introduced the principals males admitted into an intensive care unit (ICU). The popula- behind the calculation of confidence intervals and hypothesis tion mean haemoglobin concentration in UK males is testing. The present review covers the specific case of com- 15.0 g/dl. Is there any evidence that critical illness is associ- paring means in rather more detail. Comparison of means ated with an acute anaemia? arises in many different formats, and there are various methods available for dealing with each of these. Some of the The mean haemoglobin concentration of these men is simpler cases are covered in this review, namely comparison 9.7 g/dl, which is lower than the population mean. However, of a singl