general linear model and ANOVA (w8) Flashcards
what does the general linear model assume
that the ‘residuals will be normally distributed
that different statistical tests combine linearly (add together in a simple way)
purpose of GLM statistical tests
assess strengths + direction of relationships + differences
strengths of interventions + manipulations
what does ANOVA stand for
ANalysis Of VAriance
what are ANOVA’s for
categorical differences between:
- different conditions in a study (eg drug v placebo)
- different groups in a study (eg old v young, intervention v control)
in t test: how many and what type of predictor and outcome, what is the group and measure
1 categorical predictor (group = IV)
1 continuous outcome ( measure = DV)
in ANOVA how many and what type of predictor and outcome, what is the group and measure
1 categorical predictor (group = IV)
1 continuous outcome ( measure = DV)
what are the different types of ANOVA
one way, factorial (or multi-way), repeated measures (or within-subjects), mixed
in one way ANOVA: how many groups, how many and what type of predictor and outcome, what is the group and measure
3+ groups
1 categorical predictors (group=IV)
1 continuous outcome (measure=DV)
in factorial (multi way) ANOVA: how many groups, how many and what type of predictor and outcome, what is the group and measure
2+ groups
2+ categorical predictors (group, sex)
1 continuous outcome (measure)
in repeated measures (within subjects) ANOVA: how many groups, how many and what type of predictor and outcome, what is the group and measure
1 group
2+ categorical predictors (intervention, state)
1 continuous outcome (measure)
in mixed design ANOVA: how many groups, how many and what type of predictor and outcome, what is the group and measure
2+ groups
1+ categorical predictors (intervention=within, drug=between, sex=between)
1 continuous outcome (measure)
what is an interaction between 2 variables
when differences in one variable are affected by differences in another variable
what are assumptions in ANOVA
observations are independent (ie from different people, different times)
variances of different groups or conditions should be roughly equal
residuals (unexplained variance or error) will be roughly normally distributed
what assumption is made in RM-ANOVA, what do you need to do if you have 3+ groups)
the differences between levels of a variable should roughly equal variance (3+ groups or conditions in a variable, check sphericity)
what can post-hoc tests provide
p-values for differences
what type of ANOVA is it when one group trained over 4 weeks, each week they took carbs, guarana, caffeine or placebo, measure perception of fatigue before and after training session
repeated measures ANOVA
what type of ANOVA is it when one group trained over 4 weeks, each week they took carbs, guarana, caffeine or placebo, 40 men, 40 women, measure perception of fatigue in each group after training session
mixed (within and between) ANOVA
what type of ANOVA is it when
each of 4 training groups had carbs, guarana, caffeine or placebo, groups formed separately by sex, 10 people each group, 80 people total
measure perception of fatigue in each group after training session
factorial (multi way) ANOVA
what’s a MANOVA for
multiple outcome variables (eg height and weight)
what’s ANCOVA for
continuous predictor variable (eg cholesterol)
what’s logistic regression for
discrete outcome variables (eg diabetes/not)
what’s multivariate regression for
multiple predictor and outcome variables (eg cholesterol, height; diabetes/not, weight)