Exploratory Factor Analysis for Scale Development (EFA)
Exploratory Factor Analysis for Scale Development · Also known as: Açımlayıcı Faktör Analizi — Ölçek Geliştirme (EFA), psychometric EFA, scale construction factor analysis
Exploratory Factor Analysis for Scale Development is the psychometric application of EFA in which an item pool is administered and the resulting response data are analysed to discover the latent factor structure underlying the items. Originating with Spearman's (1904) factor theory and formalised for applied scale construction by Costello and Osborne (2005) and Fabrigar and colleagues (1999), this variant imposes a stricter sample requirement (n ≥ 100, subject-to-item ratio ≥ 5) and a higher loading threshold (≥ 0.40) than general EFA, and it treats the recovered factor structure as a draft to be subsequently validated by confirmatory analysis.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
This method is appropriate in the early stage of a new scale construction project, when the researcher has assembled a pool of candidate items but has not yet established how many subscales the measure should have or which items belong to each. The data must meet three preconditions: a sample of at least 100 respondents with a subject-to-item ratio of at least five; a KMO index of at least 0.70; and a significant Bartlett test. Items should be measured at an ordinal or interval level; polychoric correlations are required for ordinal Likert-type responses. If a theoretically grounded factor structure already exists from prior work, confirmatory factor analysis on a new sample is the appropriate choice rather than another EFA.
Strengths & limitations
- Provides a data-driven map of the latent structure of an item pool without requiring the researcher to pre-specify the structure.
- The 0.40 loading threshold and the subject-to-item ratio rule impose a scale-development-specific discipline that reduces the chance of retaining poor items.
- Oblique rotation yields a pattern matrix and a structure matrix, giving a richer picture of item-factor relationships than orthogonal rotation.
- Serves as a principled first step in a two-stage validation workflow that culminates in confirmatory factor analysis.
- The extracted factor solution is sample-specific and may not replicate in a new sample, particularly when n is close to the minimum.
- The number of factors to retain remains a judgment call even with parallel analysis, and different extraction methods can produce different solutions.
- Oblique rotation complicates the reporting of loadings because the pattern matrix and the structure matrix must both be considered.
- EFA alone cannot confirm the scale structure; a confirmatory study on an independent sample is required before the scale can be used in substantive research.
Frequently asked
How does psychometric EFA differ from general EFA?
They apply the same statistical model, but psychometric EFA for scale development imposes stricter quality gates: a minimum sample of 100 (rather than 50), a subject-to-item ratio of at least five, a KMO threshold of at least 0.70 (rather than 0.60), and a factor-loading criterion of at least 0.40. It also treats the result as a preliminary structure that must be validated with confirmatory factor analysis on a new sample, rather than an end in itself.
Why is oblique rotation preferred in scale development?
Psychological constructs — and the subscales of most instruments — are rarely uncorrelated in practice. Oblique rotations such as promax or oblimin allow the recovered factors to correlate, which usually yields a simpler, more interpretable pattern of loadings. Forcing orthogonality through varimax artificially constrains the solution and can obscure the true structure of the items.
What should I do with a cross-loading item?
An item that loads at or above 0.40 on two or more factors is theoretically ambiguous: it does not clearly belong to a single subscale. The standard procedure is to examine the item wording, consider whether it could be rewritten to align more cleanly with one factor, and if not, exclude it from the final scale before moving to confirmatory validation.
Can I run EFA and CFA on the same sample?
Technically you can, but it is methodologically problematic. An EFA solution is optimised for the sample on which it was estimated, so a CFA run on the same data is almost guaranteed to fit well regardless of the true structure. Best practice is to split the data (if the total sample is large enough) or collect a separate confirmatory sample, then test the EFA-derived structure with CFA on the new data.
Sources
- Costello, A. B. & Osborne, J. W. (2005). Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical Assessment, Research & Evaluation, 10(7), 1–9. link ↗
- Fabrigar, L. R., Wegener, D. T., MacCallum, R. C. & Strahan, E. J. (1999). Evaluating the use of exploratory factor analysis in psychological research. Psychological Methods, 4(3), 272–299. DOI: 10.1037/1082-989X.4.3.272 ↗
How to cite this page
ScholarGate. (2026, June 1). Exploratory Factor Analysis for Scale Development. ScholarGate. https://scholargate.app/en/psychometrics/efa-psychometric
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Confirmatory factor analysisPsychometrics↔ compare
- Cronbach's AlphaStatistics↔ compare
- EFAStatistics↔ compare
- Principal Component AnalysisMachine learning↔ compare
- SEMStatistics↔ compare