Please enable JavaScript.
Coggle requires JavaScript to display documents.
Hypothesis Testing - Coggle Diagram
Hypothesis Testing
Hypothesis Test Basics
Three categories of hypothesis tests
Comparing statistic to hypothesis
Determining changes in data/process
Testing data fit to model
General guidelines for hypothesis tests
Compare statistic to reference criteria
Define null and alternative hypotheses
Begin with a statistic
Reasons for Running Hypothesis Tests
Translate statistical results to actionable insights
Ensure statistically significant differences
Sample T Test (Two-Sample)
Problem: Comparing billable hours between two attorneys
Steps
Run Test and Analyze P-Value
Select Appropriate Test (2-Sample T Test)
Set Confidence Level
State Null and Alternative Hypotheses
Hypothesis Tests for Discrete Data
2 Proportion Test
Proportion Test
Null Versus Alternative Hypotheses
Alternative hypothesis (Ha): Likely true if null is false
Null hypothesis (H0): No effect or difference
Type of errors
Type II: Accepting false null hypothesis
Type I: Rejecting true null hypothesis
Selecting the Right Hypothesis Test
Type of comparison (means, medians, variance, etc.)
Distribution of data
Number of levels of interest
Data type (continuous/discrete)
Proportion Test
Problem: Impact of environmental issue on student performance
Steps
Set Confidence Level
Select Appropriate Test (1-Proportion Test)
Run Test and Analyze P-Value
State Null and Alternative Hypotheses
Sample T Test
Problem: Effect of seeding change on corn yield.
Steps
Select Appropriate Test (1-Sample T Test)
Run Test and Analyze P-Value
Set Confidence Level
State Null and Alternative Hypotheses
Hypothesis Tests for Continuous Non-Normal Data
Mann-Whitney Test
One Sample Wilcox
Chi-Square Test
Steps to Run Hypothesis Tests
Decide sample size
Run test in statistical software
Set confidence level (alpha)
Choose appropriate hypothesis test
Interpret p-value against alpha
Translate results into business language
State null and alternative hypotheses
One Sample Wilcoxon Test
Problem: Identifying neighborhoods with homes having 4 or more bedrooms.
Steps
Select Appropriate Test (One Sample Wilcoxon)
Set Confidence Level
State Null and Alternative Hypotheses
Run Test and Analyze P-Value
Hypothesis Tests for Continuous Normal Data
2-Sample T Test
Chi Square Test (or 1-Variance Test)
Sample T Test (or Paired T Test)