One-way Independent ANOVA

Question 1

F ratio for an independent ANOVA?

Answer 1

still follows the effect divided by error format.
Effect - between groups variance
(consists of: treatment effect, individual differences, error due to sampling)
Error - within groups variance
(consists of: individual differences, sampling error)
so, we have individual diffs and sampling error in both sets of variance
BUT, we also have treatment effect in between groups, therefore we are effectively dividing the effect of the IV by the error we normally have in our data

Question 2

How do we report an F value?

Answer 2

F(df1, df2) = , p = .

Question 3

How do we know where differences lie?

Answer 3

• look at error bars

- post-hoc tests

Question 4

What does the ANOVA actually test?

Answer 4

tests the null hypothesis that all groups are the same.
It is an ‘omnibus test’ eg. it tests is there an overall effect
-significant value indicates that there is a low probability that differences would be observed if there i no effect in the population

Question 5

Assumptions?

Answer 5

Continuous (scale) dependent variable
normal distribution (histograms)
no outliers (extreme scores)
equal variance (Levene's)

-Levenes tells you if variances are roughly equal in each group

Question 6

Are ANOVAs robust to violations?

Answer 6

They are robust to violations of:
-normal distribution, outliers, types of data
BUT
if you violate equal variances in an independent ANOVA = must do non-parametric ANOVA