Psychometrics Final

optional range for factor loadings

.40 - .75

suggest they are probably NOT different factors

Factor loadings ABOVE .75

Splitter items could be indicative of

(1) a double-barreled item or (2) an item that taps into a more broader, general construct – which not useful in a factor (i.e., subscale), but which may be useful as a single-factor

300*

unless it’s a super rare population/low base rate, like schizophrenia

minimum sample size for factor analysis

Principle Components Analysis

Used to reduce the number of variables (i.e., more of a data reduction technique) 

Principal Factor Analysis

identify a latent variable factor 

Exploratory Factor Analysis

let the items fall on whichever factors they fall on. creates 2 kinds of fit indices

Fit indices from EFAs

absolute & relative

absolute fit indices

assess whether the model fits the data well or not (not comparing) & can be used by themselves

Relative Fit Indices

only meaningful when comparing with other models​

Test Validity Important Articles

Loevinger, 1957; Clark & Watson, 1995;