Flashcards in Chapter 21 Factor Analysis Deck (16)

Loading flashcards...

1

## Communality

### Variability in an item explained by all the identified factors within a solution

2

## Eigenvalue

### Measure of the total variance in the variables accounted for by one factor

3

## Factor analysis (FA)

### Method of extracting factors which account for correlations (technically co-variances) between several variables (e.g. items on psychological scales, experimental tests)

4

## Exploratory factor analysis

### Use of factor analysis to identify latent (underlying) factors which explain the variance in correlations. Usually performed in a conceptual area where there is as yet no known well-supported factor structure.

5

## Confirmatory factor analysis

### Factor analysis performed to support an already identified factor structure. Might be with a larger data set or on a different population from the original.

6

## Factor extraction

### Stage in factor analysis when an initial set of factors is developed to explain the correlations between variables

7

## Factor loading

### Degree to which an item is associated with a factor in the analysis. A form of partial correlation.

8

## Factor matrix

### Table produced by SPSS showing loadings of each factor on each item/variable.

9

## Factor rotation

### Adjustment of the factor solution so that factors tie up more closely with the original variables. Can be orthogonal or oblique.

10

## Initial solution

### This is obtained in the first of two major steps in a factor analysis. This will provide the information needed for data checking and for deciding on the number of factors to extract.

11

## Oblique factors

### Factors which are allowed to correlate with each other.

12

## Orthogonal factors

### Factors which are not allowed to correlate with each other; geometrically at right angles to one another.

13

## Pattern matrix

### Table provided by SPSS which shows the loadings, after rotation, of each factor on each item.

14

## Principal component's analysis (PCA)

### A method of data reduction, which can be used to identify groups of indicators (e.g., items on a scale or experimental tests) that are correlated but are not expected to be caused by an underlying factor. Thus, PCA does not find 'latent' factor structure.

15

## Scree test/plot

### Plot of factors against their eigenvalues that can be used to assist with identifying the number of factors to extract

16