Hypothesis testing framework

-State the null and alternative hypotheses.

-Calculate the value of the test statistic.

-Identify the reference distribution.

-Find the**P-value** for the observed test statistic

-Check the assumptions.

-State a conclusion.

-Calculate the value of the test statistic.

-Identify the reference distribution.

-Find the

-Check the assumptions.

-State a conclusion.

compares 1 mean ⇌ fixed number (μ1=μ0).

One-sample Z-test

one mean, know σ

One-sample T-test

one mean, unknown σ

One-sample Proportion

one proportion. H0: p = p0

compares mean1 ⇌ mean2 (μ1=μ2).

Two-sample T-test

Two means.

Confidence interval

We are 95% confident that, the population parameter will occur within the range of values.

Two-sample Proportion

Two proportions.

compares subject with paired mean1 ⇌ mean2 (μD=0). EX. [before-after] [left-right] [experimental-control]

Paired-sample T-test

Two measurements on each subject.

compares category1 ⇌ category2 (C1-C2 = no association)

Chi-square test

Two categorical variables measured on each subject.

expected value

R-squared: amount that DV is explained by the IV.

Linear regression test

Two quantitative random variables measured on each subject. SPSS: coefficients (t)-(sig).

compares 3+ independent (group) means. (H0: all μ's are equal)

One-way ANOVA

3+ means. H0: group means are all equal.

-What is the assumption?

-Where do you look to check the assumption?

-What do you expect to see if the assumption is valid?

-What do you see?

-What is your conclusion?

-Where do you look to check the assumption?

-What do you expect to see if the assumption is valid?

-What do you see?

-What is your conclusion?

Normality-checking

Q-Q plot (close to line)

Linearity-checking

Residual-Predicted

linearity- (random scatter)

constant spread- even (left-right)

linearity- (random scatter)

constant spread- even (left-right)

Independence

Random allocation etc.

All of Stats-table

P(not A) = 1 -P(A)

P(A and-or B) = P(A)+P(B)-P(A&B)

P(A&B) indep = P(A)xP(B)

P(A&B) disjoint = 0

P(A and-or B) = P(A)+P(B)-P(A&B)

P(A&B) indep = P(A)xP(B)

P(A&B) disjoint = 0