ACVIM CARDIOLOGY CANDIDATE FUNCTION: UNDERSTANDING CORE STATISTICAL METHODS USED IN VETERINARY LITERATURE

摘要

Consumers of scientific literature must assess the validity of results based on careful assessment of the reported materials and methods. It is standard for investigators to report the statistical methods that were used in the study, however, this information is often passed over by the reader, who assumes that the proper methods were employed. This may or may not be the case. Some journals employ a statistical editor whose job it is to ensure that proper methodology and reporting is performed. Somejournals leave evaluation of these methods to the associate editor or to the individual reviewers. Thus, quality control of statistics is highly variable from study to study and from journal to journal. Mark Twain famously stated "There are three kinds of lies: lies, damned lies, and statistics." Consumers of scientific information that understand commonly used core statistical methods can better evaluate what they read and make determinations as to the validity of reported results. This review will cover some basic statistical premises including evaluation of normality, multiple comparison testing, linear regression and use of Bland-Altman plots, and survival curves.