Objectives #3

Question 1

Treatment variables and classification variables

Answer 1

Treatment variables: variables that are manipulated through random assignment
Classification variables: naturally-occurring variables that are not manipulated
Some variables are necessarily in the classification category (sex or gender), but some could be either manipulated or naturally occurring.

Question 2

Random selection and random assignment

Answer 2

Random selection: every member of the population you wish to generalize to has an equal chance of being selected for your sample.
Random assignment: every member of the sample has an equal chance of being assigned to a particular condition/level.

Question 3

Importance of random assignment

Answer 3

It allows causal statements to be made.

Question 4

Experiment and observational study

Answer 4

Interpretation: Experiment⇒Treatment X caused a change in Y

O.S.⇒X is associated with a change in Y.

Question 5

Explain each design, illustrate the layout, and list the appropriate statistical analysis for it.
 Randomized group design
 Matched pair design
 Repeated measurement design
 Pretest-posttest design
 Observational study

Answer 5

See PDF

Question 6

Advantage of the matched-pair and repeated measurement designs over the randomized group design.

Answer 6

MPD: If the matching variable is highly correlated with the DV , we will have less variance as a result of the matching procedure ⇒ higher power than RGD.
RMD: each subject serves as his own control, causing higher power than RGD.

Question 7

α, β, power

Answer 7

Greater α, lower β ⇒ Increased power.

Lower α, greater β ⇒ decreased power.

Question 8

Four factors that affect power

Answer 8

Sample size ⇑, power ⇑
Underlying difference between means (μ1-μ2) ⇑, power ⇑
α ⇑, power ⇑
Variance/standard deviation ⇑, power ⇓

Question 9

Test, Null hypothesis, Purpose of test

Answer 9

See PDF

Question 10

Original example for each test

Answer 10

Let’s think now!

Question 11

Statistical significance

Answer 11

You have strong evidence that the population means are not equal.
Strong evidence that the mean difference is greater than zero.
Strong evidence of a treatment effect.

Question 12

Confidential interval

Answer 12

See PDF

Question 13

η squared, effect size g for independent samples two-group designs.

Answer 13

Eta-squared: proportion of variation on the DV explained by the IV. Value will fall between 0 and 1. (e.i. eta-squared=.45, 45% of the variation in Y is explained by X.)
Hedge’s g: provides the difference between means, in within-group standard deviation units. (e.i. g=.5, the mean of group 1 is .50 within-group standard deviations above the mean of group 2.)