Confirmatory Factor Analysis Calculator

This confirmatory factor analysis calculator lets you specify and estimate a CFA model online. Use CFA when theory or previous research already determines which measured variables should be indicators of each latent factor. The calculator estimates the measurement model and reports factor loadings, factor covariances, model-fit indices, confidence intervals, and standardized estimates.

How to calculate a CFA

1. Enter your data in the table and select the indicator variables.
2. Select the indicators that belong to one construct and choose + Create factor, or add a factor with + Factor.
3. Rename the factor so that it represents the latent construct being measured.
4. Connect the factor to its indicators and set each connection to Loading.
5. Add a covariance between latent factors when the factors are expected to correlate.
6. Choose the estimator and factor-scaling method, and then select Run model.

Each factor needs a scale. By default, the calculator fixes the first loading to 1. You can instead select Factor variance = 1. The first indicator is therefore meaningful: check the loading order in the diagram or in the model syntax before running the CFA.

How to read the CFA results

The loading table shows how strongly each indicator is related to its factor. The standardized loading is useful for comparing indicators that use different units. Standard errors, p-values, and confidence intervals describe the uncertainty of each estimated loading.

Overall model fit is summarized by chi-square, CFI, TLI, RMSEA, its 90% confidence interval, and SRMR. These measures should be considered together and in relation to the model, sample, and research context. They are not automatic pass-or-fail rules. Modification indices show possible omitted loadings or covariances, but a change should only be made when it is theoretically defensible.

CFA with ordinal questionnaire items

If Likert-type indicators should be treated as ordinal, set their scale level to ordinal in the data table. The calculator then models their thresholds and automatically uses WLSMV. If the indicators should instead be treated as continuous, keep them at the metric scale level and choose ML or MLR.

CFA compared with exploratory factor analysis

CFA tests a factor structure that you specify in advance. Exploratory factor analysis searches for a factor structure without requiring the loading pattern to be completely defined beforehand. If you do not yet know which items belong to which factors, use the exploratory factor analysis calculator first. To combine a measurement model with directional paths between constructs, use the main structural equation modeling calculator.