Scientific Online Resource System

Conference Proceedings

P-value - между статистическата и клиничната значимост

Н. Матева

Abstract

Statistical inference may be thought as a body of methods to choose the `most plausible` assumption, or `the most reasonable` way of behaving in situations where we only have partial, incomplete or indirect information. P-values are used to assess the degree of dissimilarity between two or more sets of measurements or between one set of measurements and a standard, and these concepts are closely related to the statistical hypothesis testing.

The P-value is actually a probability, usually the probability of obtaining a results as extreme as or more extreme than the one observed if the dissimilarity is entirely due to variation in measurements or in subject response - that is, if it is the result of chance alone. P-value depends on a number of conditions. To describe the results of statistical hypothesis testing, the statisticians have taken the word `significance`. Statistical significance is a technical term, while medical or clinical significance is a wider concept and usually means `importance`. Some results are both statistically and clinically significant, and some are neither statistically nor clinically significant. The more troublesome results are significant in only one of these senses. When samples are very large, small differences may be statistically significant even though they have no importance in clinical practice or possibly even in public health. At the other extreme, small samples may produce large differences so imprecisely determined that they are not statistically significant at the levels usually required. Studies with low statistical power can lead to an unjustified solution for a given treatment or for another medical result. P-value should be taken into account when making a decision, but should not be the sole basis for such decision.


Keywords

statistical significance; P-value; statistical hypothesis testing

Full Text


References

Emerson John D. and Colditz Graham A. Use of statistical analysis in the new England Journal of Medicine. In: John Bailor, Frederick Mosteller, editors. Medical Uses of statistics. NEJM Books;1986. p. 27-38.

Davis R, Ribner HS, Keung E, Sonnenblick EH, LeJemtel TH. Treatment of chronic congestive heart failure with captopril, an oral inhibitor of angiotensin-converting enzyme. New England Journal of Medicine 1979;301:117- 21

Freiman JA, Chalmers TC, Smith HJr, Kuebler RR. The importance of beta, the type II error, and sample size in the design and interpretation of the randomized controlled trail: survey of 71 “negative” trials. New England Journal of Medicine 1978; 299:604-9.

Ware James H., Mosteller F., Ingelfinger Joseph A. P Values. In: John Bailor, Frederick Mosteller, editors. Medical Uses of statistics. NEJM Books; 1986. p. 149-169.


Refbacks

Font Size