5. Factor Analysis Flashcards
The construct of interest is called a l____ variable, or f____
latent, factor
Scale items are directly measured (directly o____), so are called o____ variables
observed, observed
Factor analysis allows us to determine d____
dimensionality
When building a new scale, we create a set of o____ v____. We want to see whether these items capture o____ c____ or s____ r____ c____ in a meaningful way
observed variables
one construct, several related constructs
Exploratory factor analysis:
- No p____ about the number of factors or which i____ belong to what factors
- You can e____ the data
- R____ poor performing items and r____ the factor analysis
- predictions, items
- Explore
- Remove, re-run
Confirmatory factor analysis:
- You have c____ e____ of what the factor structure will look like
- You want to c____ these predictions with our data. You cannot e____ the data
- clear expectations
- confirm, explore
Factor analysis is about making s____ of s____ v____ between variables
sense, shared variance
Communality = the p____ of v____ that an item s____ with other items. In general, larger communalities are d____
proportion, variance, shares, desirable
factor analysis types:
1. M____ r____ e____
2. M____ l____ e____
3. P____ a____ f____
- Minimal residual estimation
- Maximum likelihood estimation
- Principle axis factoring
Preliminary analysis involves:
1. Check i____-i____ c____ (Bartlett’s test of sphericity)
2. test whether an analysis is a____
- inter-inter correlations
- appropriate
Main analysis includes:
1. Factor e____
2. M____ f____
3. Factor r____ and factor l____
4. I____ the factors
- Extraction
- Model fit
- rotation, loadings
- Interpreting
The Kaiser-Mayer-Olkin test of sampling adequacy tests the p____ of v____ between variables that might be c____ v____
proportion variance
common variance
Bartlett’s test of sphericity tests whether the c____ m____ is significantly different from an i____ m____
correlation matrix, identity matrix
Factor extraction = deciding h____ m____ factors are c____ by our i____
how many, captured, items
Extraction methods are based on e____. E____ ARE THE AMOUNT OF T____ v____ explained by each factor. They indicate the i____ of a factor
Extraction methods are based on eigenvalues.
Eigenvalues are the amount of total variance explained by each factor
Eigenvalues indicate the importance of a factor: bigger eigenvalue = more important factor
Extraction methods start by generating the m____ number of factors and inspect their e____
maximum, eigenvalues
Parallel analysis involves comparing your data to r____ g____ data to find out which factors m____ explain variance in your data, beyond r____ noise
randomly generated
meaningfully
random
The point at which the slope changed on the scree plot is called the i____ point
inflexion (We extract however many factors are to the left of the inflexion point (not including the inflexion point itself))
The relationship between a factor and an item can be thought of as the P____ c____ between the two
This relationships is called a f____ l____
Pearson correlation
factor loading
An item’s highest factor loading is its p____ factor loading. An item’s additional, lower factor loadings are its s____ factor loadings. When an item has high loadings on multiple factors, we say it is a c____-l____ item.
primary, secondary, cross-loading
Inter-items of +-__ are too different, and +-__ are too similar
+-.30
+-.90
