Test Construction: Factor Analysis

Question 1

Factor Analysis: Overview

Answer 1

One Group
Several Tests

Create a correlation [R] matrix of all item scores on all tests, and convert into Factor matrix

‘Rotate’ factors to simplify interpretation, interpret and name factors

Question 2

Factor Analysis: Factor Loading

Answer 2

Correlation coefficients = association between each test and each factor

*Square the factor loading to determine variance accounted for

Question 3

Factor Analysis: Orthogonal Rotation

Answer 3

Resulting factors are uncorrelated, independent

Attribute measured by one factor is independent from attribute measured by another factor

Question 4

Factor Analysis: Oblique Rotation

Answer 4

Resulting factors are correlated

Attributes measured by factors are NOT independent

Question 5

Factor Analysis: Main Takeaways for Exam

Answer 5

• Squared factor loading provides a measure of shared variability
• Comunality determined from sum of squares of orthogonal factors
• Orthogonal: uncorrelated
• Oblique: correlated