Notes MRM 2
There are open book tests and multiple choice questions to help prepare for your exam.
The test will be open ended questions seeing how you can interpret.
Moderator is the interaction effect, for example communication skills has a positive effect on
student satisfaction. They are strengthening each other.
Mediation / mediator is the how, how does the independent variable have an effect on the
dependable variable.
Arrows always are beneath the arrow and go from independent to dependent variable.
ANOVA
The samples need to be mutually exclusive.
One-way ANOVA has one predicter variable/ Independent variable
The alternative hypothesis is there is at least one difference in the dependent variable mean score
between the PV categories. Keep that in mind.
F-test is the test statistic for this test. F-values look at explaining variability. F-distribution is also
different from a T-distribution.
Variability is split up in two groups: Difference between groups (explained) vs differences within
groups (unexplained)
ANOVA compares the variation between the groups.
There is the Total sum of squares = Model sum of squares + Residual sum of squares
R2 or R squared is the proportion of total variance. Equals to the variability explained by the model /
total variability. This is not a official statistic.
A proper statistic is the F-test. The F ratio = explained variability / Unexplained.
There are different DF between these sums, u divide the sum by the df. For getting into the Mean
squares. U calculate the degrees of freedom by:
SSm = K – 1
K is amount of groups
SSr = n – k
N = number of units / data points per group
K = groups
F value like T value can be looked up in a table.
Multiple comparisons cause artificial inflation of alpha. The chance increases because you take the
5% E.G. three times (family-wise error rate). To correct for this we adjust for test at the end.
, We choose Bonferroni for this course, the alpha gets divided over all comparions. E.G. if there are 5
comparisons each gets 1%.
There are other types of ANOVA when some assumptions are not met.
Notes week 2 Interaction & Factorial ANOVA
N-way independent Anova has 2 or more PV’s. N stands for the amount of PV.
The N-way adds more realism and control to the analysis and they give a clearer view of the cause
and effect.
The interaction effect (moderation)
- The effect of one PV on the OV is moderated by another PV
Interaction helps with various ‘problems’ in research, interaction helps resolve conflicts between
earlies theories/results for example.
When looking at ANOVA for factorial design the SSm or sum of squares of your model gets divided
over three sums of Square. Namely SSa, SSb and SSaxb.
Factorial ANOVA in SPSS: Analyze > GLM > Univariate. Check descriptive stats, homogeneity test and
estimates of effect size.
Partial N2 (Eta) effect size of the individual PV’s Interactions.
We use the corrected model and the corrected total.
Then look at pairwise comparisons for main effects.
Plotting interactions
analyze > GLM > Plots
Main effect on the horizontal, second predictor variable we put on separate Lines.
To check If the interaction effect is significant we do the Simple Effects Test (a checkable box when
analyzing in SPSS). This spots out a table with Sig. in there.
Week 3 Regression
Regression is like doing ANOVA but then with quantitative data. We are looking at how much of the
variance in our data can be explained by our model.
The alpha (intercept or startin point) and beta (slope). In y = ax + b. Alpha would be B and beta
would be A, X remains the same.
R2 is the amount of variance that gets explained by the model. This is calculated by dividing the SS M
by the SST. This is interpreted as percentage of total variation.
There are open book tests and multiple choice questions to help prepare for your exam.
The test will be open ended questions seeing how you can interpret.
Moderator is the interaction effect, for example communication skills has a positive effect on
student satisfaction. They are strengthening each other.
Mediation / mediator is the how, how does the independent variable have an effect on the
dependable variable.
Arrows always are beneath the arrow and go from independent to dependent variable.
ANOVA
The samples need to be mutually exclusive.
One-way ANOVA has one predicter variable/ Independent variable
The alternative hypothesis is there is at least one difference in the dependent variable mean score
between the PV categories. Keep that in mind.
F-test is the test statistic for this test. F-values look at explaining variability. F-distribution is also
different from a T-distribution.
Variability is split up in two groups: Difference between groups (explained) vs differences within
groups (unexplained)
ANOVA compares the variation between the groups.
There is the Total sum of squares = Model sum of squares + Residual sum of squares
R2 or R squared is the proportion of total variance. Equals to the variability explained by the model /
total variability. This is not a official statistic.
A proper statistic is the F-test. The F ratio = explained variability / Unexplained.
There are different DF between these sums, u divide the sum by the df. For getting into the Mean
squares. U calculate the degrees of freedom by:
SSm = K – 1
K is amount of groups
SSr = n – k
N = number of units / data points per group
K = groups
F value like T value can be looked up in a table.
Multiple comparisons cause artificial inflation of alpha. The chance increases because you take the
5% E.G. three times (family-wise error rate). To correct for this we adjust for test at the end.
, We choose Bonferroni for this course, the alpha gets divided over all comparions. E.G. if there are 5
comparisons each gets 1%.
There are other types of ANOVA when some assumptions are not met.
Notes week 2 Interaction & Factorial ANOVA
N-way independent Anova has 2 or more PV’s. N stands for the amount of PV.
The N-way adds more realism and control to the analysis and they give a clearer view of the cause
and effect.
The interaction effect (moderation)
- The effect of one PV on the OV is moderated by another PV
Interaction helps with various ‘problems’ in research, interaction helps resolve conflicts between
earlies theories/results for example.
When looking at ANOVA for factorial design the SSm or sum of squares of your model gets divided
over three sums of Square. Namely SSa, SSb and SSaxb.
Factorial ANOVA in SPSS: Analyze > GLM > Univariate. Check descriptive stats, homogeneity test and
estimates of effect size.
Partial N2 (Eta) effect size of the individual PV’s Interactions.
We use the corrected model and the corrected total.
Then look at pairwise comparisons for main effects.
Plotting interactions
analyze > GLM > Plots
Main effect on the horizontal, second predictor variable we put on separate Lines.
To check If the interaction effect is significant we do the Simple Effects Test (a checkable box when
analyzing in SPSS). This spots out a table with Sig. in there.
Week 3 Regression
Regression is like doing ANOVA but then with quantitative data. We are looking at how much of the
variance in our data can be explained by our model.
The alpha (intercept or startin point) and beta (slope). In y = ax + b. Alpha would be B and beta
would be A, X remains the same.
R2 is the amount of variance that gets explained by the model. This is calculated by dividing the SS M
by the SST. This is interpreted as percentage of total variation.
