Hypothesis Testing Roadmap

Discrete Data 1 Proportion Test

1
Start Ho: Ppop = Pt
Tips to Remember Ha: Ppop ≠ Pt
t = target
1) Proper sample size selection is required 1 Levels 2 Stat>B.S>1 Prop.
for tests to be effective. of Interest (same as CI for P)
for X
2) Ha can be <,>, or ≠ 2 Proportion Test
3) If p>α, then fail to reject Ho Type Ho: Ppop1 = Ppop2
of Ha: Ppop1 ≠ Ppop2
If p<α, then reject Ho Attribute Number of 2
Y Stat>B.S>2 Prop.
Factors for >2
X ANOM
Variable
Ho: Ppop1=Ppop2 =........
Variable Data - Ha: At least one Ppop is different
Stuck Contingency Table Stat>ANOVA>ANOM
Not Normal Proceed with Ho: F1 independent of F2

DOE
caution >1 No. of Ha: F1 dependent on F2
X's Stat>Tables>Chi Sq Test
No
Kruskal-Wallis or Ho: Data is normal
Mood's Median Test 1 Ha: Data is not normal
Yes Are σ's
Ho: M1 = M2 = M3 ... Stat>B.S.>Normality Test
Ha: at least 2 ≠ equal
or
Stat> Nonparametrics> Is Y DOE, Logistic Regression
STAT>B.S.>D.D.S>G.S. >2
KW or MM Non-Normal Normal Ho: Ppop < or > f(x)
for each Ha: Ppop = f(x)
Mann-Whitney Test Levene's Test level of Stat>ANOVA>DOE
Ho: M1 = M2 Ho:σ1 = σ2 = σ3..... X? Stat>Regression>Logistic Regression
Ha: M1 ≠ M2 Ha: at least 2 ≠
Stat>Nonparametrics> Stat>ANOVA>
Mann-Whitney Test for Equal Variances
Normal
>2
Yes Levene's Test
Ho: σ1 = σ2 2 Variable Data -
Levels
Are σ's Ha: at least 2 ≠ Normal >2 Levels
of Interest 1 mu (µ)
equal Stat>ANOVA> of Interest µ or σ?
for X
Test for Equal for X
Variances 1 Bartlet's Test
No
Ho: σ1 = σ2 = σ3..... 2 1 Sample t test
Ha: at least 2 not equal sigma (σ) (used for Paired-t also)
F-Test
Stat>ANOVA> Ho: µ1 = µt
Test Ho: σ1 = σ2
Stuck Test for Equal Ha: µ1 ≠ µt
Median
Median Ha:σ1 ≠ σ2
Proceed with or σ? Sigma Variances t = target
Stat>ANOVA>Test for
caution to Stat>B.S>1 Sample t
Equal Variances (Requires
2 sample-t or (or use CI from
Stacked Data) or Stat >
Mann-Whitney One Sample Wilcox or graphical Summary)
No Basic Stat > 2 Variance
1 Sample sign Are σ's (stacked or unstacked)
Chi Sq Test
Ho: M1 = Mt equal Chi Sq Test
Ho: σ1 = σt
Ha: M1 ≠ Mt Ho: σ1 = σt
Ha: σ1 ≠ σt Stuck
t = target
t = target Ha:σ1 ≠ σt
Stat>Nonparametrics Proceed with Yes
Stat>B.S.>D.D.S>G.S t = target
& either a caution to One
(if S1 falls between CI, No Yes Stat>B.S>D.D.S>G.S
1 Sample Sign Way ANOVA Are σ's (if σt falls between CI,
then fail to reject Ho) equal
or a then fail to reject Ho)
1 Sample Wilcox

One Way ANOVA

Ho: µ1 = µ2 = µ3..... 2-Sample t-test 2-Sample t-test
Ha: at least 2 not equal Ho: µ1 = µ2 Ho: µ1 = µ2
Stat>ANOVA>One Way Ha: µ1 ≠ µ2 Ha: µ1 ≠ µ2
(then select stacked or Stat>B.S.>2-sample t Stat>B.S.>2-sample t
unstacked data) (Uncheck the Assume (Click the Assume Equal
Equal Variances Box) Variances Box)