Lecture 11 – Comparing several means - summary page.docx
University of Waterloo
Lecture 11 – Comparing Several Means – Summary Page
- ANOVA – tell you whether the average score on a particular variable is
significantly different across three or more groups.
-t-tests: for comparing means of two groups
-ANOVA: for comparing means of 3+ groups
ie do 3 gen immigrants get better jobs than 1 or second gens?
Ie do cancer patients who take a new medication get better results
than with other drugs, no drugs, or alternative therapy.
-Asking if theres differences between groups.
-Why cant I run multiple T tests?
• Have a risk of 5% being wrong with our test, so if we do
another t test with another p value then we have an additional
5% chance of being wrong. Every time you do a new T-test
you’re increasing your likelihood for error.
1. To check for it, look at the histogram for our DV and look at
ii. Homogeneity of variance
1. Refers to wanting the variance to be equal for all groups.
2. Want the variances to be similar
3. Do a levine’s test and we want it to be non-significant
(<0.05) because that tells us we have a homogeneity of
4. Don’t want to see one leptokurtic kurtosis and one
iii.Independence of cases (except for the dependent/paired samples t-
1. Refers to that we don’t want any cluster sampling
iv. Interval level data 2
1. Want a DV that’s either interval or ratio
2. Unlike t-test,ANOVAuses an f-test. F-score is linked to a
certain p value which will tell you the likelihood of getting
your results strictly by chance.