The article compares different tests to see if data sets are normally distributed. They used five tests on two sets of data and simulated data sets. The Kolmogorov-Smirnov test is best for large samples, while the Jarque-Bera test is best for small samples. The researchers recommend using the Jarque-Bera test for small samples and the Kolmogorov-Smirnov test for large samples to check for normality.