University:
California State University, Northridge
Course:
PSY 524/L | Multivariate Analysis Computer and Lab
Academic year:

2005
Views:

378

Pages:

24
Author:

Oliver Donovan

MANOVA Lecture 12 1 2 45 Multivariate Analysis of Covariance 0011 0010 1010 1101 0001 0100 1011 • The linear combination of DVs is adjusted for one or more Covariates. • The adjusted linear combinations of the DVs is the combination that would have been had all of the subjects scored the same on the CVs. 1 S =S * (Y ) −S ( YZ ) (S 2 45 ( Z ) −1 ) S ( ZY ) Multivariate Analysis of Covariance 0011 0010 1010 1101 0001 0100 1011 • Each subjects score is made up of the DVs and the CVs 1 110   IQ      Y111 = 115  wrat − r  108  wrat − a  2 45 Multivariate Analysis of Covariance 0011 0010 1010 1101 0001 0100 1011 • So that each S is a combination of the original S plus the SSCP for the CVs and the covariances between the DVs and the CVs. Z Y1 1 2 Y2 45 Z  [ 2.00] [64.67 54.67 ]    Y1  64.67  2090.89 1767.89    Y2  54.67  1767.56 1494.22 Multivariate Analysis of Covariance 0011 0010 1010 1101 0001 0100 1011 (Z) −1 (ZY) S = S −S (S ) S * (Y) (YZ) 1 2 −1 * 2090.89 1767.56 64.67 S = − 2] [ 64.67 54.67] [   1767.56 1494.22 54.67 45 Multivariate Analysis of Covariance 0011 0010 1010 1101 0001 0100 1011 • s= Calculating Wilk’s Lambda is the same and for the most part the F-test is the same except calculating s and DF2: ( p + q )2 (dfeffect )2 − 4 ( p + q )2 + ( df effect )2 − 5 1 2 45 ( p + q ) − dferror + 1   ( p + q)( df error ) − 2   df2 = s ( dferror ) − −   2 2     Different Multivariate test criteria 0011 0010 1010 1101 0001 0100 1011 • • • • Hotelling’s Trace Wilk’s Lambda, Pillai’s Trace Roy’s Largest Root 1 2 45 Different Multivariate test criteria 0011 0010 1010 1101 0001 0100 1011 • • When there are only two levels for an effect s=1 and all of the tests should be identical 1 2 45 When there are more than two levels the tests should be nearly identical but this is not always the case Different Multivariate test criteria 0011 0010 1010 1101 0001 0100 1011 • When there are more than two levels there are multiple ways in which the data can be combined to separate the groups – 1 2 (e.g. one dimension separates group 1 from groups 2 and 3, a second dimension separates group 2 from group 3, etc.) 45 Different Multivariate test criteria 0011 0010 1010 1101 0001 0100 1011 • • Wilk’s Lambda, Hotelling’s Trace and Pillai’s trace all pool the variance from all the dimensions to create the test statistic. 1 2 45 Roy’s largest root only uses the variance from the dimension that separates the groups most (the largest “root” or difference). Different Multivariate test criteria 0011 0010 1010 1101 0001 0100 1011 • The various formulas are (E is error and H is hypothesized effect): – 1 2 Wilk’s Lambda - |E| |H + E| - It’s the ratio of error to effect plus error. Analogous to 1 – R2. Middle of the road in terms of how conservative a test it is. 45 Different Multivariate test criteria 0011 0010 1010 1101 0001 0100 1011 • The various formulas are (E is error and H is hypothesized effect): – 1 2 Hotelling’s trace – Trace(H/E)=C and you look up C in a table to get the F value. It is analogous to an F-test. Very liberal test. 45 Different Multivariate test criteria 0011 0010 1010 1101 0001 0100 1011 • The various formulas are (E is error and H is hypothesized effect): – 1 2 Pillai’s trace – Trace(H/(H + E)). Analogous to R2. Very conservative 45 Different Multivariate test criteria 0011 0010 1010 1101 0001 0100 1011 • The various formulas are (E is error and H is hypothesized effect): – 1 2 Roy’s Largest Root - (H/(H + E)) and it looks for the biggest difference. It is variable in terms of how conservative it is. 45 Which do you choose? • For the most part stick with Wilk’s lambda. It’s the most widely used Use Hotelling’s Trace if 0011 0010 1010 1101 0001 0100 1011 • – – • 1 2 Manipulated (experimental) variables Very clean design with no internal validity problems 45 Pillai’s trace is the most conservative, but if your design has many problems (e.g. unbalanced, assumption violation, etc) pillai’s is supposed to be robust to these problems Assessing DVs 0011 0010 1010 1101 0001 0100 1011 • If multivariate test is significant • Run multiple univariate F-tests (one per DV) in order to see on which DVs there are group differences, this assumes uncorrelated DVs. 1 2 45 Assessing DVs 0011 0010 1010 1101 0001 0100 1011 • The overall alpha level should be controlled for considering the multiple tests 1 2 α overall = 1 − (1 − α1 )(1 − α 2 ) K (1 − α p ) • 45 The alpha levels can be divided equally or they can be set up to give more important tests a more liberal alpha level. Assessing DVs 0011 0010 1010 1101 0001 0100 1011 • If DVs are correlated than individual Ftests are problematic but usually this is ignored and univariate Fs interpreted anyway 1 2 45 Assessing DVs 0011 0010 1010 1101 0001 0100 1011 • Roy-Bargman step down procedure – Can be used as follow-up to MANOVA or MANCOVA with correlated DVs or as alternative to multivariate analysis all together. 1 2 45 Assessing DVs 0011 0010 1010 1101 0001 0100 1011 • Roy-Bargman step down procedure – – – The theoretically most important DV is analyzed as an individual univariate test (DV1). The next DV (DV2), in terms of theoretical importance, is then analyzed using DV1 as a covariate. This controls for the relationship between the two DVs. DV3 (in terms of importance) is assessed with DV1 and DV2 as covariates, etc. 1 2 45 Assessing DVs 0011 0010 1010 1101 0001 0100 1011 • Discriminant Function analysis – – We will discuss this more later but… – It uses group membership as the DV and the MANOVA DVs as predictors of group membership – Using this as a follow up to MANOVA will give you the relative importance of each DV predicting group membership (in a multiple regression sense) 1 2 45 Specific Comparisons and Trend Analysis 0011 0010 1010 1101 0001 0100 1011 • With a significant multivariate (and univariate) test and more than two groups, this needs to be followed with comparisons of the individual groups. 1 2 45 Specific Comparisons and Trend Analysis 0011 0010 1010 1101 0001 0100 1011 • Just like any test discussed previously, this can be done with planned or post hoc comparisons. 1 2 45 • Planned comparisons can be written into SPSS syntax and if post hoc you can adjust the test by the degrees of freedom to get a Scheffe adjustment. Unequal samples 0011 0010 1010 1101 0001 0100 1011 • If intended to be equal and no meaning to the imbalance, use type 3 sums of squares • If the imbalance is meaningful use type 1 sums of squares 1 2 45

Multivariate Analysis of Variance

Multivariate Analysis of Variance - Page 3

Multivariate Analysis of Variance - Page 4

Multivariate Analysis of Variance - Page 5

Multivariate Analysis of Variance - Page 6

Multivariate Analysis of Variance - Page 7

Multivariate Analysis of Variance - Page 8

Multivariate Analysis of Variance - Page 9

Multivariate Analysis of Variance - Page 10

Multivariate Analysis of Variance - Page 11

Multivariate Analysis of Variance - Page 12

Multivariate Analysis of Variance - Page 13

Multivariate Analysis of Variance - Page 14

Multivariate Analysis of Variance - Page 15

Multivariate Analysis of Variance - Page 16

Multivariate Analysis of Variance - Page 17

Multivariate Analysis of Variance - Page 18

Multivariate Analysis of Variance - Page 19

Multivariate Analysis of Variance - Page 20

Multivariate Analysis of Variance - Page 21

Multivariate Analysis of Variance - Page 22

Multivariate Analysis of Variance - Page 23

Multivariate Analysis of Variance - Page 24

of 24

Description

Free up your schedule!

Take 5 seconds to unlock