# PSYC 3000 Lecture Notes - Lecture 1: Multiple Comparisons Problem, Analysis Of Variance, Dependent And Independent Variables

19 views1 pages

2 Aug 2016

School

Department

Course

Professor

PSYC 3000

Jan 5

Example

-dependent variable is how many words are recalled

-independent variable is the unique instruction for each group

Analysis of the example

-k = the number of groups.

-in assignment table 1 is the raw material for all tests that will be done.

-you could do an independent samples t test between two groups to see if they are significantly

different.

Multiple comparisons

-In anova you test the hypothesis that there is no difference between all the groups. Once you find out

that there is a difference between some groups, you can do individual t test.

-first problem: the s pooled in each t test is different. So instead we should create an s pooled with the

data from all the groups and use this s pooled in every t test conducted. So basically the denominator of

each t test will always be the same.

-first problem is solved quite easily.

-Second problem: PC error rate is the same as alpha. FW error is however much higher. This is the

problem. IF you do too many t test your FW error rate will be so high a type 1 error is inevitable.

Lowering the PC (usually to 0.01 instead of 0.05) or lowering the number of comparisons (by choosing

wisely which comparisons you want to make instead of blindly comparing all groups)

Bonferroni formula

-the problem with lowering your PC error rate too small is that it will make it very hard for you to attach

significance to any of your test. The power is thus lowered.

-The ideal is to figure what your desired FW is and rearrange your PC and comparisons in order to

achieve the proper FW.

-need to explain in assignment 4 what the error rate is. How it was achieved and what the PC and FW

error rate are

-most common correction.

find more resources at oneclass.com

find more resources at oneclass.com