Work

A Statistics Primer

Year: 1998
Type: article
Abstract: The basis for scientific study is the search for truth.Hypothesis testing is a common way of providing evidence to support this search for truth.As discussed in a previous article, 12 study findings such as means or frequencies are associated with critical values from statistical tests along with associated probabilities or P values.Recall that tests for statistical significance (such as chi-square, Student's t, analysis of variance or ANOVA) are chosen during the design of a study and, regardless of the statistical test chosen, good study design requires that all tests of significance lead to a probability statement or P value."The P value provides the reader of the study with a guide for the likelihood that the statistical observation in a study is due to chance alone."8 By convention, P values less than 0.05 are accepted as statistically significant.This convention of statistical significance set at the 0.05 level, however, presents challenges to the reader in interpreting study findings.First, statistical significance (P Յ 0.05) may be present but may not be clinically relevant.Consider a previously presented example of the effects of two treatments on the outcome of length of hospital stay.8 Suppose the P value associated with the study findings is 0.006 -a highly statistically significant finding.However, the clinical result is less impressive because the statistical estimate associated with Treatment A results in a length of stay of 6.2 days and the estimate of Treatment B results in a length of stay of 6.8 days.These differences between results are meaningless.Even very small differences such as a few parts per million could be statistically significant if large numbers of subjects compose the sample.However, there may be no practical importance to such a finding.
Language: Spanish
Open Access status: green
OpenAlex ID: W1961043989
FWCI:
Cites: 13
Cited by: 18
Related to: 10