[Skip to Navigation]
May 1978

Statistical Approaches

Arch Ophthalmol. 1978;96(5):911. doi:10.1001/archopht.1978.03910050507028

To the Editor.  —We read with interest the observations of Drs Kupferman and Leibowitz on experimental staphylococcal keratitis in rabbits.1 However, we have some comments on the presentation and analysis of data in their paper.The authors assessed the results of various antibiotic treatments by determining the numbers of viable bacteria in the corneas. The results were expressed as arithmetic means. Biological data are often skewed and do not have a normal distribution; therefore, the usual statistical tests cannot be used. Such data should be transformed to logarithms to create a normal distribution. Then, a geometric mean rather than an arithmetic mean can be determined.The standard deviation is an indication of the variability of observations in a particular sample. The standard error, which the authors used, is an indication of the variability of the average among various samples. Although the standard deviation and standard error are related, the

Add or change institution