The article compares different tests used to check if data follows a normal distribution. They looked at four common tests: Anderson-Darling, Chi-square, Kolmogorov-Smirnov, and Shapiro-Wilk. The Shapiro-Wilk test had the best error rate, followed by Kolmogorov-Smirnov, Anderson-Darling, and Chi-square. For continuous data, Shapiro-Wilk was the most powerful test, while Chi-square was best for discrete data. However, all tests performed poorly for mixed data with equal means and variances.