Research on Permutation Test

In statistic, in order to compare two populations and identify the difference between the two treatment groups, statistical hypothesis testing is often used in practice. Z-test and T-test are common statistical testing and when we are conducting z-test and t-test, we are assuming that the data is normally distributed. However, in the case that normal assumption does not hold for the data, the result generated by z-test or t-test will not be reliable for drawing conclusion. In contrast to z-test and t-test, permutation test has the flexibility of the test statistic and minimal assumption of the data distribution. Thus, permutation test (Sayan Mukherjee, Polina Golland, & Dmitry Panchenko, 2003) was chosen. In this project, permutation test was used to analyze how different factors (i.e. gender, age, native language) affect the fluency score obtained. These findings are insightful in improving software system that assists in second language learning education of Mandarin Chinese.