Summary Checklists performing analyses for Research Skills: Analysis of Variance (880434-M-3, 880586-M-3)

Research Skills: Analysis of Variance (S) | SPSS guideline according to checklists

Checklist 1: Hierarchical multiple regression………………………………………….… 2
Reliability analysis……………………………………………………………………… 2
Computing a scale variable………………..………………………………………….. 2
Table with correlations and descriptives…………………………………………….. 3
Testing assumptions……………………………………………………………………. 4
Influential cases
Outliers
Multicollinearity
Normality
Independence of errors
Heteroscedasticity
Linearity
Perform hierarchical multiple regression (already done by testing assumptions).. 4

Checklist 2: Mediation analysis……………………………………………....…………….. 8
Testing assumptions………………………………..…………………………………. 8
Influential cases
Outliers
Multicollinearity
Normality
Independence of errors
Heteroscedasticity
Linearity
Perform mediation analysis by using PROCESS…………………………………... 11

Checklist 3: Moderation analysis (Hayes’ PROCESS macro, model 1)……………... 13
Computing an interaction variable…………………………………………………… 13
Testing assumptions………………………………..…………………………………. 14
Influential cases
Outliers
Multicollinearity
Normality
Independence of errors
Heteroscedasticity
Linearity
Perform moderation analysis by using PROCESS………………………………… 16
Create graph visualizing the interaction……………………………………………... 17

Checklist 4: Repeated measures ANOVA………………………………………………… 19
Testing assumptions………………………………..…………………………………. 19
Normality (within each group)
Sphericity (Mauchly’s test, Greenhouse-Geisser, Huynh-Feldt)
Perform repeated measures ANOVA for main effects (X1 on Y, X2 on Y).……… 21
Perform interaction effect for repeated measures ANOVA by using PROCESS
(X1*X2 on Y)…………………………………………………………………………….. 22

Checklist 5: Mixed ANOVA……………………………………………………………………. 24
Testing assumptions………………………………..…………………………………. 24
Normality (Kolmogorov-Smirnov test) per experimental condition
Homoscedasticity (homogeneity of variance)
Sphericity (Mauchly’s test, Greenhouse-Geisser, Huynh-Feldt)
Perform mixed ANOVA……………………………………………………………….. 27


1

,Checklist 1: Multiple (hierarchical) regression analysis
1. Introduction
a. Brief summary of the study aim
b. Hypotheses

2. Methods
a. Participants (N)
b. Demographics (e.g. age, gender)
c. Description of scales (i.e., mean, SD, reliability, sample item)
i. How many items + example + measurement
ii. Recoding
iii. Reliability (α > .70 is good)

Using drop-down menu
1. Analyze
2. Scale
3. Reliability analysis
4. Paste items that belong to the specific dimension in the box
of “Items:”
5. Click on “Statistics:”
6. Check the following boxes:
Item
Scale
Scale if item deleted
7. Continue
8. OK (or PASTE; paste places the command into the syntax)

Using syntax
RELIABILITY
/VARIABLES= VARIABLES
/SCALE('ALL VARIABLES') ALL
/MODEL=ALPHA
/STATISTICS=DESCRIPTIVE SCALE
/SUMMARY=TOTAL.

iv. Compute variable + mean + SD

Using drop-down menu to make a new scale variable
1. Transform
2. Compute variable
3. Type variable name in “target variable” box
4. Drag the variable in the “numeric expression” box and use
“+“ to add it up to the other variables you paste in the box
(e.g. (Variable X1 + Variable X2 + Variable X3).
! Do not forget to divide it by the number of variables in
order to get a scale → (X1 + X2 + X3)/3
5. OK

Using syntax to make a new scale variable
COMPUTE New variable name=(Variable + variable)/NR of
variables.
EXECUTE.




2

, 3. Results
a. General
i. Analysis chosen
ii. Dependent/independent variables included in analysis
iii. Table with correlations and descriptives (mean, SD)

Using drop-down menu
1. Analyze
2. Correlate
3. Bivariate…
4. Paste the variables you want to correlate/see mean and SD in
the box of “Variables:”
5. Click on “Options”
6. Check the box of “Means and Standard Deviations”
7. Continue
8. OK (or PASTE; paste places the command into the syntax)

Using syntax
CORRELATIONS
/VARIABLES= VARIABLES
/PRINT=TWOTAIL NOSIG FULL
/STATISTICS DESCRIPTIVES
/MISSING=PAIRWISE.

Table …: Descriptive Statistics of all variables included and correlations
M SD … … …
…
…
…
Total N = …. * p < .05 ** p < .01

iv. Interpretation of (direction of) correlations, means and SDs (e.g.,
floor/ceiling effects, multicollinearity) = descriptive statistics, correlations
o Floor/ceiling effects;
• Floor effect = very low scores,
• Ceiling effect = very high scores
• In those cases, there is not much room for a significant
effect to be found, because all scores are very low/high
already
o Multicollinearity
• The correlation coefficients serve as indicators of the
assumption of no multicollinearity for predictors.
.1 (small effect)
.3 (medium effect)
.5 (large effect → strongly correlated)
• If there are significant correlations between the
predictors, this is not necessarily very problematic if the
corresponding Pearson value is somewhere below 0.5.
0 = no relationship; (-)1 = perfect relationship
• The coefficient should be lower than .9 to assume there
is no violation of this assumption.



3