The article compares six different tests to see how well they can tell if a group of numbers comes from a normal distribution. The tests were tried on different sample sizes and levels of contamination. The results showed that for small samples, all tests had low power. But for n = 20, the Shapiro-Wilk and Anderson-Darling tests were the best. For n = 60, the Shapiro-Wilk and Liliefors tests were most powerful. And for large samples, the D'Agostino-Pearson test was the most powerful.