Independent Sample Tests - Comparing Two Proportions
Jump to navigation
Jump to search
The tests employed depend upon:
- Whether a weight is or is not applied (or, equivalently, a design effect has been specified).
- Settings in Statistical Assumptions (in addition to the settings below, tests are not conducted when a group has a sample size less than the number specified in Minimal sample size for testing).
If viewing Column Comparisons and an ANOVA-Based Column Comparison test has been selected in Multiple comparison correction then the following tests are performed where the proportion is represented as a vector of 1s and 0s:
- Multiple comparison correction= Fisher LSD: Multiple Comparisons t-Test (Fisher LSD)
- Multiple comparison correction= Tukey HSD: Tukey HSD
- Multiple comparison correction= Dunnett: Dunnett’s Pairwise Multiple Comparison
- Multiple comparison correction= Duncan: Duncan’s New Multiple Range Test
- Multiple comparison correction= Bonferroni (Pooled t-Test): Multiple Comparisons t-Test with Bonferroni Correction
- Multiple comparison correction= Newman-Keuls (S-N-K): Newman-Keuls (S-N-K)
- Multiple comparison correction= False Discovery Rate (pooled t-test): Multiple Comparisons t-Test with False Discovery Rate Correction
Else:
- Proportions = Non-parametric
- Data is not weighted: Pearson's Chi-Square Test of Independence
- Data is weighted: Second Order Rao-Scott Test of Independence of a Contingency Table
- Proportions = z-test:
- Data is not weighted: Independent Samples Z-Test - Comparing Two Proportions
- Data is weighted:Independent Complex Samples Z-Test - Comparing Two Proportions
- Proportions = t-test:
- Data is not weighted: Independent Samples T-Test - Comparing Two Proportions
- Data is weighted: Independent Complex Samples T-Test - Comparing Two Proportions
- Proportions = Quantum Proportions: Independent Samples - Quantum Column Proportions Test
- Proportions = Survey Reporter Proportions: Independent Samples - Survey Reporter Column Proportions Test