Week 5 MANOVA Flashcards
(13 cards)
how to check if the H0 of equal covariances matrices can be rejected?
see table BOX’s test of equality of covariance matrices. note: F (df1, df2) p
what is the influence of large sample size?
MANOVA
makes H0 of equal covariance matrices very sensitive, so (too) easy rejected.
how to check if the null hypothesis of equal group variances can be rejected? if H0 is rejected is this a problem?
MANOVA
see table leven’s test of equality of error variances.
only if not robust.
how to check if the multivariate tests are significant? What are you able to do if a test of a component is sign?
MANOVA
zie table multivariate tests: note. F(df1, df2) Fvalue p for every test.
check wether a variable has a sign univariate effect.
how to check if a variable has a sign univariate effect? if a variable is sign, what could we do now?
MANOVA
see table tests of between subjects effects–>see corrected model–>note: F (df1, dferror) F value, p
bonferoni
how would you apply bonferoni on univariate tests. if the test is still sign what could we use now?
MANOVA
calculate new p (see formula sheet)–>check if univariate effect is still sign.
use tukey post-hoc tests
how would you interpret the univariate effects using tukey post-hoc.
MANOVA
welke groepen verschillen van elkaar.
Multiple comparison tabel > kijk naar de Mean Difference en Sig. van de significante univariate effecten die gevraagd worden > Noteer: Y1 en Y2: Mean Difference = …, p = … en licht toe welke dan significant hoger is dan de ander.
Let op! In de tabel van Multiple Comparisons van de Tukey post-hoc test is de Bonferroni correctie al toegepast, vergelijk deze Sig. dus gewoon met een alpha van .05.
How to check how many discriminant functions are significant?
DA
tabel Wilks’ Lambda. –> noteer van elke sign function. X2(df)= , p
How much variance explained of the data is accounted for by the significant discriminant functions together
DA
see table eigenvalues>see cumulative % of highst sign function
For each significant discriminant function, indicate which two variables have the largest absolute correlation with the discriminant function.
DA
check structure matrix. for each sign function (vertical) note the highst one
how do you determine if the correlations in a function have a general or a contrast function?
Wat zijn de General function/algemene functies? tabel Structure Matrix > binnen welke significante functie correleren alle variabelen positief of negatief? > generieke functies.
Wat zijn de Contrast functies?: tabel Structure Matrix > welke significante functies hebben zowel hoge absolute positieve als negatieve correlaties?
For which two groups does the classification work best?
DA
tabel Classification Results > Zie percentage rijen > Kijk naar de percentages waarbij dezelfde groepen met zichzelf vergeleken worden. > hoogste waarde = beste classificatie.
where do you finde the overall classification
DA
left underneath the classification results table
