Exploratory factor analysis
Look for a small number of factors behind a set of items.
When to use it
Use exploratory factor analysis when you do not yet want to fix which items belong to which factor. You ask how many factors the correlations suggest, then read the loadings. If you already have a model, use confirmatory factor analysis.
Assumptions
Variables are numeric and are standardized before fitting. Rows with any missing value are dropped. Tensr extracts factors by maximum likelihood. It does not rotate them, and the request has no rotation or extraction menu. There is no KMO or Bartlett test.
Running it in Tensr
An empty factor count keeps up to five factors, and never more than the number of variables or one less than the number of complete rows.
Options
Prop
Type
Reading the output
One factor extracted from the six items. Loadings should be moderate, and the sixth item was built weaker than the first five.
| Variable | Factor1 |
|---|---|
| item1 | 0.456 |
| item2 | 0.549 |
| item3 | 0.58 |
| item4 | 0.562 |
| item5 | 0.497 |
| item6 | 0.363 |
| Factor | Eigenvalue | Variance % | Cumulative % |
|---|---|---|---|
| Factor1 | 1.54 | 25.7% | 25.7% |
EFA on 6 variables, n = 120. EFA on 6 variables, n = 120. Metrics: Factors = 1; Cases = 120.
Reporting (APA 7)
EFA on 6 variables, n = 120. Report the estimate in the table. This procedure is not summarised by one p-value.
Coming from SPSS
Analyze → Dimension Reduction → Factor. SPSS asks for an extraction method and a rotation. Tensr always uses maximum likelihood and does not rotate.