Idk man

Seite 1

TIV, DV, confounding variables

A confounding variable = something that might affect your findings in unwanted ways - must be kept constant so it doesn't

Data

Can be quantitative or qualitative

Can be nominal, ordinal interval or ratio

Types of tests: difference and correlational

Tests for simple comparisons (1 factor wth 2 conditions): t-tests, Mann Whitney, Wilcoxon

more comparisons:

1 factor w/ 3 conditions: ANOVA, Kruskal-Wallis, Friedman's

2 factors w/ 2 conditions each (4 total, 2x2):

3 factors w/ 2 conditions each (8 total, 2x2x2)

ANOVA: Analysis Of Variance - parametric test

Variance: (standard deviation)^2 = measure of spread/dispersion

It compares within group error variance to between group error variance.

Non-parametric alternatives: Distribution-free tests, "simple" treatment of data, Kruskal-Wallis (non-par, between, similar to Mann Whitney), Friedman's (non-par, within, similar to Wilcoxon)

No matter how many factors/IVs an ANOVA has, there is always only one DV.

ANOVAS can be: between participants, within participants and mixed (at least one factor between, one within etc)

ANOVA 2x2: gives you main effect factor 1 (variance for both conditions), main effect number 2 and interaction effect

Interaction can be represented graphically: parallel lines => no interaction, non parallel lines => no interaction

Seite 2

ANOVAs don't tell you where the differences are, for that you need to use post-hoc tests

Example: simple effects (compares both factors of one study with one condition of the other factor)

to do this in SPSS, you need to split the file to get all the comparisons. For repeated measures, no prep, for mixed: split the file by the between part factor, for between, split by one factor, unsplit, split by other factor

For more than one DV, we use MANOVA!

MANOVA: Multivariate Analysis Of Variance

It asks: is there any effect of IV on the linear combination of DVs?

ex of combination: add them together, but if you want some to weigh more you can multiply them then add them

Assumptions of MANOVA: multivariate normality, homogeneity of variance, DVs are not strongly correlated (if they are, it's harder to pick out individual differences)

BUT MANOVA doesn't tell you WHICH DV is affected, you need post-hoc tests

Could use individua anovas for each dv, corrected with Bonferroni.

Seite 3

THEMATIC ANALYSIS