Week 2: Psychometrics & trait theory Flashcards
What is factor analysis?
Factor analysis examines the relationship between observed correlating variables to determine whether or not those variables combined represent latent unobserved variables, thus reducing the number of variables to a minimum number of combined factors.
What’s the difference between an item and a factor or component?
An item is an observed, measured element or expression of an attribute (eg a Q psychometric questionnaire or a response to a behavioural task).
A factor or component is a combination of items which form a latent dimension (construct).
What is factor loading, and what is the value used to determine that each item loads on to a factor?
Factor loading is a measure used to determine the statistical relationship between an item and a factor-> it’s basically the correlation coefficient for the item and the factor it’s loading on to.
Generally if the factor loading is >|.40| (medium correlation) then it is considered to belong to, or make up that factor.
For a factor to be considered a meaningful psychological construct, the items which load onto it need to be both:
The variables need to be both qualitative and quantitative
What are the two types of factor analysis (FA), EFA and CFA? Describe their differences.
The two types of factor analysis are analysis are exploratory factor analysis (EFA) and confirmatory factor analysis (CFA).
EFA is used to identify possible latent construct/s whereas CFA is used to reveal already hypothesised/ theorised or empirically tested latent constructs.
PCA (principle component analysis) is the simplest and most commonly used EFA (exploratory factor analysis).
What are it’s assumption in regards to common and unique variance? Use the construct of neuroticism as an example
PCA assumes that all the common and unique variance in the items are attributed to the same latent construct eg neuroticism
What’s the difference between orthogonal and oblique factors/ components?
Orthogonal factors are dimensions which are independent of one another, whereas oblique factors are related to one another.
Are neuroticism and extraversion orthogonal or oblique?
Neuroticism and extraversion are considered orthogonal, they are not related to one another.
Are Gf (general fluid) and Gc (general crystalised) intelligence factors considered to be orthogonal or oblique constructs/ dimensions?
Gf and Gc are considered to be oblique constructs; they are related to one another.
The suitability of an item (or a psychometric test dataset) to be included in a FA model, is considered to be that item’s _______
factorability
Items which form distinct factors, or item clusters, and exhibit high levels of convergence and discriminant validity, are known to belong to which type of factor analyses structure?
Items which cluster together and show high discriminant validity from the other clusters are forming a simple structure.
Name the rotation type most commonly used in psychology, used to rotate orthogonal factors in such a way that the variability they each explain is maximised.
Varimax rotation (from variance max rotation).
A rotation type used for oblique factors is called _____, or direct rotation. This twists the space, and is avoided in psychology, because it can distort the conceptual space.
(Direct) Oblimin rotation (Oblique rotation)
What is the maximum possible number of factors?
The maximum possible number of factors is equal to the number of dimensions being measured
The standardised variance of all the items that is explained by a single factor is known as the ______, or characteristic roots. Explain in different words what it represents.
eigenvalue
Eigenvalue is a numerical index used to express the portion of the total variance among several correlated variables which contributes to a single factor.