Factor Analysis

Front 1

What is the goal of factor analysis?

Back 1

Factor analysis focuses on the discovering the underlying structure of the variables and reducing those variables to the smallest number of variables possible.

Front 2

Name 2 types of Factor Analysis.

Back 2

-Principle Component Analysis (PCA)
-Factor Analysis (Principle Axis factoring)

Front 3

What is the outcome of Factor Analysis?

Back 3

As many factors as variables as factors are reproduced, however FA in SPSS only retains those factors that contribute to the equation.

Front 4

Communalities what are they?

Back 4

Communities are sum of squared factor loadings, and are specifically a representation of the common variance in a variable.

Front 5

What are the requirements for a FA?

Back 5

-Linear Relationship
-Absence of both multivariate and univariate outliers
- A sufficient number of cases for correlation
- The assumption of normality, assuming that the research want’s to generalize the population.

Front 6

What is the required sample size for a FA?

Back 6

There’s some controversy over this, with some advocating for a 5 or even ten times the participants for the number of variables.

Front 7

7. What is the required factorability for a FA?

Back 7

Generally a correlation of at least .32 is required for a significant correlation

Front 8

What are Eigenvalues?

Back 8

Eigenvalues are the amount of variance that each variable contributes to the overall factor. Kaiser’s rule suggests that retain eigenvalues greater than 1.

Front 9

What are factor loadings?

Back 9

Factor loadings are the correlation between the variable and factor.

Front 10

What are the Rotations?

Back 10

There two kinds of rotations:
-The first kind of rotation is orthogonal, used for uncorrelated factors, and sits at 90% of an angle. Varimax is the most common rotation.
- The second kind of rotation is Oblique, used when the factors are correlated, and does not sit at an 90% angle. Direct oblimin is the most common of the rotations.

Front 11

Explain the difference between PCA and FA

Back 11

Perhaps the most important difference is that PCA measure or takes into account all the total variance, but FA only takes the common variance into account. Secondly, where PCA uses the measured variables to create components, FA uses the Factors to estimate measured variables. We use PCA if we merely want to reduce the number of variables and FA if we want to examine a construct through the use of variables.

Front 12

What are Factor or component scores and what are they used for?

Back 12

Factor scores are the ability to estimate someone scores on factor through their position on the variable that makes up the factors. They can be used to check for collinearity and for further analysis, such as t-tests.

Front 13

Can FA analysis be generalized?

Back 13

Some types of factor analysis cannot be generalized and all the score must adhere to the assumption of normality

Front 14

Split the Variables

Back 14

Total Variance can be split into 3 groups: Unique, common and Error

Front 15

What is the Required Split off point for KMO and should Bartlett’s test of Sphericity be significant?

Back 15

KMO > 0.6 and Bartlett’s test of Sphericity must be significant

Front 16

What is multicollinearity and how do we tell if we have it?

Back 16

Multicollinearity is where variables are too highly correlated with each other, where it has become impossible to separate the variables. Look at the determinant.

Front 17

What is required in APA write-up?

Back 17

Factors
Amount of variance accounted for
outliers
KMO
Number of participants

Front 18

What are Residuals?

Back 18

They are difference between the original Correlation matrix and the reproduced orrelation matrix.

Front 19

What are the two oblique rotations and which one we usually interpret?

Back 19

Pattern and Structure Matrix, but Pattern matrix is the one we usually interpret.