Factor Analysis SPSS Workshop

This set of vocabulary flashcards covers the fundamental concepts, statistical tests, and procedural steps of Factor Analysis in SPSS as outlined by Dr. Maher Khelifa.

Factor Analysis

An exploratory and confirmatory multivariate technique used for data reduction, scale development, psychometric evaluation, and assessment of variable dimensionality.

Constructs

Unobservable latent variables, such as egoism or teaching effectiveness, that are derived from the measurement of other, directly observable variables.

Data-Reduction Technique

An approach in factor analysis that reduces a large number of overlapping variables to a smaller set of factors reflecting specific constructs or dimensions.

Correlation Matrix

The first step in factor analysis where the correlation for all variables is computed; coefficients greater than $0.3$ in absolute value are typically indicative of acceptable correlations.

Bartlett Test of Sphericity

A test used to determine if the correlation matrix is an identity matrix (where all diagonal terms are $1$ and off-diagonal terms are $0$).

Kaiser-Meyer-Olkin (KMO) Measure of Sampling Adequacy

An index comparing the magnitude of observed correlation coefficients to partial correlation coefficients; values of $0.8$ and higher are considered great, while less than $0.5$ is unacceptable.

Factor Extraction

The second step in factor analysis aimed at determining the number of factors underlying a set of measured variables.

Principal Components Analysis (PCA)

The most commonly used extraction method which forms linear combinations of observed variables to explain the largest amount of variance.

Eigenvalues

A statistical criterion used to decide how many factors to represent the data; factors with a variance less than $1$ are generally not included.

Scree Plot

A visual representation showing the total variance associated with each factor, where a steep slope identifies the large factors to be selected.

Factor Rotation

The third step in factor analysis where factors are adjusted to make them more meaningful, interpretable, and associated with a minimal number of factors per variable.

Varimax Rotation

The most popular orthogonal rotation method that yields uncorrelated factors and attempts to minimize the number of variables with high loadings on a single factor.

Oblique Rotations

A rotational method, such as Promax, that yields correlated factors and is less frequently used because the results are more difficult to summarize.

Identity Matrix

A matrix in which all diagonal terms are $1$ and all off-diagonal terms are $0$, indicating that variables are not related.

Factor Loading

Values in a component matrix that represent the relationship between a measured variable and an underlying factor.