saptamana doispe Flashcards
What is unrelated ANOVA?
When you have more than 2 conditions
it is a one-way ANOVA because we have one IV with multiple levels
Why don’t we use t-tests instead?
Because carrying out multiple comparisons on same data increases the risk of type 1 error - more chance of finding a false positive
What are the assumptions of the unrelated ANOVA?
Interval or ratio data
Sample data drawn random from population
Variances are normally distributed
Variances of populations are equal
What can the total variation in set of scores be divided into?
Systematic variation: due to effect of IV on DV
Error variance: due to random or chance influence
Testin significance
Comparison made between:
Amount that means vary amongst each other
Amount each sample varies around its mean
What is the F statistic?
Compares variance between conditions and variance within conditions
F ratio = variance between group/variance within groups
What is between-conditions variance?
Measure of effect of IV
What is within-conditions variance?
Measure of error of variance (variability of scores due to random or chance_
If means differ a lot and/ or error variance is small
F value will be large
reject null hypothesis (scores from different conditions come from same population)
If means differ slightly, or there is large amount of variation within conditions
F value will be small
null hypothesis cannot be rejected
What are post hoc tests?
ANOVA looks at whether there is an overall difference between means - omnibus test
We need to find out where the difference is
We use Tukey HSD test which controls for type 1 error rate
How do we report?
Mean and SD and the n for each condition
Look at Levene’s statistic - we want it to be non-significant
Levene tests homogeneity of variance assumption
If test is significant - may need to consider non-parametric alternative
Report df, F value and significance
F (df1, df2) = f value, p = significance level
