MultivariateTopics1
Do we want Wilks’ Lambda to be: Large or small? Why?
Small; measures variance not explained
Socioeconomic Status = w1Educ + w2Income + w3Wealth an example of?
Discriminant function (composite / latent variable)
Which FA technique allows factors to be correlated?
Oblique
Which data reduction technique analyzes all variance?
CFA / SEM
Primary difference between “path analysis” and Factor Analysis
Only MV in path analysis; Latent in FA
If multivariate effect (Wilks’ lambda) significant, what is next step?
Examine univariate results for each DV.
A mortgage company uses Age, Income, & Debt to predict who is or is not eligible for a loan. Which procedure?
Discriminant Analysis
Tells us proportion of variability in the MVs accounted for by a factor
Eigenvalue or SSL
Which data reduction technique allows MVs to load on more than one factor?
EFA
You review an article on SEM with a diagram showing ovals and rectangles. Types of Vs?
Rectangles: Measured/Observed; Ovals: Latent
This procedure minimizes chance a researcher will miss measuring “right” DV.
MANOVA
Ideally, in MANOVA the DVs should be correlated?
Moderately
What is the primary tool in FA used to make factors more interpretable?
Rotation
This procedure evaluates difference between observed COV & hypothesized COV matrices
CFA/SEM
The problem with this statistic is that it is very sensitive to N
Chi-Square
When examining univariate results for each DV where IV has three levels, "post-hoc" tests control for what type of error?
Type I
The DV for Discriminant Analysis (& Chi-Sq/Logistic Regression) is measured on what scale?
nominal/categorical/binary
Proportion of variability in a measured variable accounted for by the factors as-a-set
post-extraction communality
True or False? In CFA, we want the RMSEA to be high (near 1.00) and the GFI to be low (near 0)
False
A way to adjust for the sensitivity of this test is to
Divide the Chi-Sq by df
What is the multivariate effect size measure? How do we calculate?
1 – Wilks’ lambda
Which procedure–orthogonal or oblique–produces additional factor loading matrices
Oblique
The goal of which "data reduction" technique is to achieve a non significant result?
CFA/SEM
The Chi-Square test compares
Observed (r matrix) to Expected (r matrix)