The statistical analysis of microfilarial skin snip counts: stabilization of the variance of microfilarial counts.

Abstract

Microfilarial skin snip counts from savannah and rain forest areas of West Africa and from Central America were analyzed. Before parametric statistical tests, such as the comparison of means, can be properly performed, the variance must be independent of the mean. The best transformation of the skin snip counts to make the variance independent of the mean was determined. Frequency distributions of skin snip counts are shown for several different transformations. In three out of four sets of data that were examined, the best transformation was found to be the fourth root of the counts, not the logarithm of the count as has been customarily used.