Mathematically it is quite a complicated exercise to calculate the size of a sample necessary to achieve a given statistical power in clinical trials. Altman's nomogram is a graphical method of assessing the power and statistical significance of a test at a variety of sample sizes. The right-hand vertical axis of the nomogram shows various power values, from 0.05 to 0.995. The left-hand vertical axis represents the 'standardized difference': a ratio which relates the difference of interest to the standard deviation of the observations. There are two axes within the nomogram, one for a significance level of 0.05, the other for 0.01, with total sample sizes indicated on each. The nomogram can be used to evaluate the optimal sample size once the power is specified, the significance level 5 per cent or 1 per cent is chosen and the standardized difference is calculated. This nomogram can be found at p. 456 of Douglas G. Altman (1991), Practical Statistics for Medical Research, London: Chapman and Hall.
Was this article helpful?