A research on the importance of testing the normality assumption in microbiological data

The most used distribution in statistical analysis is the normal distribution. Parametric tests (e.g. one sample t-test) require that the data are normally distributed. In this study, milk somatic cell count data (SCC) used to test the normal distribution was obtained from a farm for the first and second month of lactation. According to the findings of the present study, SCC data of the first month showed normal distribution. However, the SCC data of the second month did not show normal distribution. Since the first month data showed a normal distribution, one-sample t-test, which is one of the parametric test methods, was applied for comparison with a specific reference value; since the second month data did not show a normal distribution, the Wilcoxon One-Sample Signed Rank Test, which is the non-parametric equivalent of the one-sample t-test, was applied. When the parametric test was applied to the second month data that did not show a normal distribution, results that did not comply with the standards in terms of SCC were found. When the same data was analyzed with the nonparametric test method, results that complied with the standards were obtained. It is noteworthy that different results are obtained in both analyses. As can be seen from the research results, since existing data sets in the field of microbiology may tend to show large variations, it should be tested whether the data show normal distribution before determining the statistical analysis method. According to the research results, the normality test must be applied in the statistical analysis of microbiological data showing large variations.