Output (Y)

Hypothesis Testing 1-Proportion Test

1 Sample HO: p1 = Ptarget
Continuous Categorical of Categorical Inputs (Xs)* Ha: p1 , <, or > Ptarget
Input (X)

Continuous Stat>Basic Stat>

Continuous Y Data Categorical Y Data 1-Proportion

Categorical 2-Proportion Test

Normality
Normality Test
Test 2 Samples Ho: p1 = p2
Data are Not Normal HHOO:: Data
Data is
are
Normal
Normal Data are Normal Ha: p1 , <, or > p2
HHaa:: Data
Data is
are
notnot
Normal
Normal Stat>Basic Stat>
2 or More 2 or More Stat>Basic
Stat>Basic Stat>Normality
Stat>NormalityTest
Test or
or
1 Sample 1 Sample 2 or more 2 Samples 2-Proportion
Samples Samples Stat>Basic
Stat>Basic Stat>Graphical
Stat>Graphical Summary
Summary Samples
For 2 or more
Samples,
Compare Variances Compare Variance Compare Mean Compare 2 Means Test for Chi Square Test
Independence HO: FA Independent FB
to Target differences
Levenes Test 2-Sample t-Test Ha: FA Dependent FB
HO: 1= 2 = 3 Chi2 Test Paired t-Test HO:1 = 2 Stat>Tables>Chi2 Test
Ha: i j for i to j HO: 1 = target HO: d = O Ha: 1 , <, or > 2
(or at least one is different) Ha: 1 target Ha: d , <, or > O
Stat>ANOVA> Stat>Basic Stat>2-Sample t Compare Variance
Test for Equal Variances Stat>Basic Stat>Graphical Stat>Basic Stat>Paired t uncheck: Assume = Variances to Target
Summary
(if target std dev. falls within
Chi2 Test
95% CI, then fail to reject HO; HO: 1 = target
Compare 2 otherwise, reject) Compare Variances 1 Sample Ha: 1 target
Medians Stat>Basic Stat>Graphical
Bartletts Test (> 2) or F-Test (2) Summary
Mann-Whitney Test HO: 21 = 2 2 = 23
(if several outliers, use Compare Median Ha: 2i 2j for i to j (if target std dev. falls within
Moods Median test) to Target (or at least one is different) 95% CI for std dev, then fail to
HO: 1 = 2 Stat>ANOVA>Test for Equal Variance reject HO: reject otherwise)
Ha: 1 , <, or > 2 1-Sample Wilcoxon
(if several outliers, use 1-
Stat>Nonparametrics> Sample Sign test)
Mann-Whitney (Moods Compare Mean
Median Test) HO: 1 = target Equal Variances to Target
Ha: 1 , <, or > target
Stat>Nonparametrics> Compare More Than Compare 2 Means 1-Sample t-Test
Compare More 1-Sample Wilcoxon (1- 2 Means HO: 1 = target
Than 2 Medians 2-Sample t-Test Ha: 1 , <, or > target
Sample Sign)
1-Way ANOVA HO: 1 = 2 Stat>Basic Stat>1-Sample t
Kruskal-Wallis Test (assumes equality of variances) Ha: 1 , <, or > 2
(if several outliers, use HO: 1 = 2 = 3 Stat>Basic Stat>2-Sample t
Moods Median test) Ha: i j for i to j Check: Assume = Variances
HO: 1 = 2 = 3 (or at least one is different) (uses pooled std dev) P-Value alpha,
If P-value alpha, then
then reject
reject H Hoo
Ha: i j for i to j Stat>ANOVA>1-Way If P-value
P-Value >> alpha,
alpha, then
then fail
fail to
to reject
reject H
Hoo
(or at least one is different) (select stacked or unstacked data) Alpha is usually pre-set at 0.05. Use
other values as appropriate.
Stat>Nonparametrics> If P-value
P-Value >> alpha,
alpha, ensure
ensure sufficient
sufficient
Kruskal-Wallis (Moods power and correct sample size if
Median Test) *NOTE: If X is Continuous and Y is Continuous,
necessary.
proceed with Regression and Correlation Analysis.
Hypothesis Testing Protocol
Practical Problem or Theory: State problem or state theory to be proven.
Statistical Problem: State Ho and Ha (with the correct test in mind)
Statistical Analysis: Test to be used, alpha and power. Run test, show printout, results and p-value
Statistical Conclusion: Based on p-value, Reject Ho or Fail to Reject Ho. (If the p is low, the null must go)
Practical Conclusion: Answer to your practical problem. Theory is proven true or not.

Hypothesis Tests by Data Type

Y
Continuous Categorical
How does change in X affect change
in Y? Logistic Regression (out of scope)
Statistical: Regression
Continuous
Graphical: Scatter Diagrams

X Means or Medians Different?

Statistical:
Are the outputs different?
Statistical: Chi Square, Proportions
tests
If Normal: t-tests; ANOVA
Graphical: Pareto Diagrams;
If Non-Normal: Wilcoxon, Mann-
Stratification
Categorical Whitney, Kruskal-Wallis
Graphical: Histograms; Box Plots
Variance Different?
Statistical: Test for Equal Variances
(Bartletts, Levenes)
Graphical: Histograms; Box Plots