The article discusses how to test the effectiveness of a treatment when dealing with regression toward the mean. The researchers found that using likelihood ratio tests on truncated distributions gives accurate results, even with moderate sample sizes. Comparing different methods, they discovered that the likelihood ratio test is the most powerful for testing treatment effects. The study also showed that Wald's test and score tests may not give reliable results unless the sample size is very large. Overall, the likelihood ratio test was found to be the best method for evaluating treatment effects in the presence of regression toward the mean.