The second variable you will be asked to fill in, is the β, or ‘1 – the power’ of the study to detect a true difference. In the jargon of hypothesis testing, α is set to address the type-1 error rate, type-1 error being the incorrect rejection of the null hypothesis. With the majority of disciplines, this is traditionally set at 0.05 3, although sometimes 0.01 is used.
Sigma vs alpha software#
In most statistical software packages, the first variable you will be asked to give is the α, that is the value we will set as our significance cut off, more familiarly known as the p value.
Let us first remind ourselves of the information required before sample size, or power calculations can be attempted. But in planning a clinical trial, there are several important factors we have to consider which no statistician or software can help us with. With the myriad of statistical software available nowadays, we no longer have to cast our minds back to the intricacies of the integration/differentiation equations, nor strain our eyes looking at the tiny numbers on the p, z and t tables in order to calculate sample size. Since it is neither practical, feasible nor ethical 1, 2 to study the whole population, a number of participants must be selected from the population, which is fewer in number, but nevertheless adequately represents the population so that inferences can be drawn from the study results 2.
Resident statisticians in a clinical department are often tasked with calculating the number of participants (sample size) required for randomised controlled trials.
0 kommentar(er)
