Which Statistical Test Do I Use?
When working with statistics it is critical that you choose the correct test. Choosing the correct test is a relatively simple exercise once you understand a few key items about your data. Here we'll try to guide you through some of the basics of choosing between T-tests and X2 (Chi Squared) tests. Please see our glossary page for definitions of the terms that we are using:
Is your data nominal, ordinal or qualitative?
1. If it is nominal or ordinal that implies that you should use a X2 test. Is your data independent or paired?
2. If it is independent you should use a Pearson X2 test
2. If it is paired then you should use a McNemar's X2 test
1. If it is qualitative, you should (probably) use a T-test. Is your data independent or paired?
2. If it is paired the you should use a paired T-test
2. If it is independent do the two data sets have and equal variance?
3. If the two data sets have equal variance then you should use an unpaired T-test
3. If the two data set have unequal variance then you should use a Welch test (AKA modified T-Test)
Is your data nominal, ordinal or qualitative?
1. If it is nominal or ordinal that implies that you should use a X2 test. Is your data independent or paired?
2. If it is independent you should use a Pearson X2 test
2. If it is paired then you should use a McNemar's X2 test
1. If it is qualitative, you should (probably) use a T-test. Is your data independent or paired?
2. If it is paired the you should use a paired T-test
2. If it is independent do the two data sets have and equal variance?
3. If the two data sets have equal variance then you should use an unpaired T-test
3. If the two data set have unequal variance then you should use a Welch test (AKA modified T-Test)