10. t Test and Paired t Test

Topics and Notes

1. t-test for normal population means
   • a. t-distribution revisited.
   • b. Assumptions of the t-test.
   •   ■ Population must be normal and variance is unknown
   •   ■ If sample size is small, a test must be used
   •   ■ If sample size is large, the t-test can be approximated by normal test.
   • c. Steps for t-test
   •   ■ The same 6-step procedure of normal tests applies to t-test
   •   ■ Only critical value method is required in this class
2. Paired t-test
   • a. Understanding paired samples: taking measurement from the same subject twice (e.g, measurement before and after a person received treatment)
   • b. Take different of the paired measures (either before measurement - after measurement or vice versa) to convert to a single sample
   • c. Assumptions on the differences of the corresponding the paired measurement (in this class): normally distributed with unknown variance.
   • d. t-test for the difference of dependent population means based on the paired sample
3. Lecture Note
• t-tests and paired t-tests HTML PDF

Practice and Interactive Apps

1. Practice exercises #10 WEB LINK
2. [Optional]Read sections 9.1-9.2, 11.3 of Navidi's textbook
3. t Tests for Normal Population Means INTERACTIVE APPS