Contrast Coding Flashcards
1
Q
when do you use contrast coding?
A
when testing a specific hypothesis?
2
Q
what would you use if there wasn’t a specific hypothesis?
A
post hoc tests
3
Q
how many contrasts should you do?
A
K-1 = 1 less contrast than the # of groups
4
Q
what are the rules of contrast coding?
A
positive vs negative comparisons / eliminate a case which isn't involved by assigning a 0 / the sum of contrast should be 0 / equal groups in each contrast / choose sensible contrasts
5
Q
what do post hoc tests do?
A
compare all means to see where differences lie / correct the error rate
6
Q
how can you solve a type one error?
A
make groups independent
7
Q
post hoc safe option is:
A
Bonferroni
8
Q
post hoc for unequal variances is:
A
Games-Howell
9
Q
how can you make specific contrasts?
A
with a post hoc and cohen’s d
10
Q
what does the f ratio omnibus tell you?
A
about the whole model
11
Q
what doesn’t the f ratio omnibus tell you?
A
where the differences lie
