*** Lab #3 Two-way analysis of variance ***
Purpose: To demonstrate how two independent variables (factors) interact to produce outcomes different from what would have occurred if either was tested alone. In this lab, as in Lab #2, one of the factors must be a true IV, but the other may be a truly categorical, non-manipulated Subject Variable, (e.g. gender, race, educational status).
Instructions: Using your proposal topic, and the four variables you created for the Lab #1 assignment,
(1) Create two more independent variables that will serve as a two factors in a factorial design
(FAC1 and FAC2). Each factor should have two levels (e.g. low and high, yes or no, conditions A and B).
The new variables should be called FAC1 (for Independent var1) and FAC2 (IV2) and given Variable Labels.
You must develop an explanation (or use an existing one) which would logically predict how the variables would operate together to produce the interaction. The example I have used suggests that FAC1 (hi and lo supervisor surveillance) will interact with FAC2 (lo and hi worker need for autonomy) to produce effective performance (C variable) for workers who need direction (lo autonomy) who work under a hi surveillance (much direction) supervisor. As well, workers who need much independence (hi autonomy) will perform better under lo surveillance. On the other hand workers in other conditions will perform less well since they are under the type of supervision less appropriate for them.
(2) You must "assign" subjects to groups in such a way to insure that a significant CROSSED INTERACTION occurs.
NOTE: You must create data that will support the hypothesis for your crossed interaction on your primary DV with a probability for rejection between p >.05 and < .001.
There should be no main effects for either variable (FAC1 or FAC2). Interaction must be significant for your primary DV (e.b. A) but not necessarily for B and C. What significant differences, if any, would you expect to find for B, C and D? Why?
Hint: think about the relationships you established among the four variables in Lab 1.
NOTE: If you are using a student version of SPSS, you will not have Multivariate under Analyze/General Linear Model. Therefore, you can only run one DV at a time using Analyze/General Linear Model/ Univariate. Using SPSS in the labs, you can run all DVs under Multivariate procedures.
Hint: it will help to set up a 2 X 2 table for the four cells and plug in "dummy" means to determine how the means in each cell will have to vary. It will also help to sort your dependent variable before you attempt to assign subjects to conditions.
You can do this by including the SORT CASES BY...command and LIST VARS= ....
(3) Now add the new variable(s) in the data list and "assign" subjects to groups by entering the new values (1, or 2, for each subject). If you have 30 subjects, you will have to assign 7 to two cells and 8 to the other two cells.
(4) Describe and explain how all four means fit (or follow from) your theory or reasoning.
Lab 3 data definition
VARIABLE LABELS:
FAC1 therapy
FAC2 gender
VALUE LABELS:
Fac1 1 psychodynamic 2 behavioral
Fac2 1 male 2 female
Data analysis for Lab 3 Two way anova (easy as 1,2,3)
Steps:
#1. Create data: After defining your data variables and labels, in data view, choose View [value labels]. This will make it easier to work with changing subject values to fit your hypotheses.
To calculate how your “assigned” Ss are distributed to conditions (cells):
Analyze/Descriptives/Crosstabs
a. Fac1 -> Row
b. Fac2 -> Column
#2. Test your hypotheses: To test your hypothesis (no significant main effects, but a 2-way interaction), use a Two-way ANOVA:
Analyze/General Linear Model/Multivariate (for SPSS student version, use /Univariate and run each DV separately)
a. DVs -> Dependent variable(s)
b. Fac 1 -> Fixed Factor
c. Fac 2-> Fixed Factor
Choose under: Options
d. Descriptive statistics
e. Estimates of effect size
(move each fac plus the interaction fac1*fac2 over to the "display means for")
f.. Plots: fac1 -> Horizontal; fac 2 -> separate [add]
#3. Analyze data:
Inspect Cross tab results to ensure that Ss are (almost) equally distributed into four cells, i.e. 7 or 8 in each.
Inspect Univariate ANOVAs to make sure:
1. There are no main effects for either fac1 or fac2
2. There is a significant interaction (probability of rejection between p < 05 but not less than .001).
3. The means are in the predicted direction.
Explain results in the narrative.
Note: (Important!) If results are counterintuitive, you must be careful to explain how the theory predicts the findings.
Import your Academic tables and figures with numbers and labels into your word processing document.