Skip to content
Omnikalc

Variance inflation factor calculator

Enter the R-squared of a predictor regressed on the other predictors and the sample size to get the VIF, the tolerance, how much the standard error is inflated and the effective sample size.

Updated on
Your inputs
%

Value out of range

Value out of range

Results

Variance inflation factor (VIF)

5.00

Moderate to high multicollinearity
Tolerance
0.200
Standard error inflation
2.24 ×
Wider confidence interval
124 %
Effective sample size
40

The variance inflation factor calculator gives the VIF, the tolerance, the inflation of the standard error and the effective sample size of a predictor in a multiple regression, from its R-squared on the other predictors.

How the calculation works

  • Enter the R-squared obtained by regressing the predictor on all the others.
  • Enter the sample size.
  • The calculator gives the VIF, the tolerance, the standard error inflation, the confidence interval widening and the effective sample size.

Example

If the other predictors explain 80 % of a predictor, its VIF is 1 / (1 − 0.8) = 5.00. The tolerance is 0.200, the standard error is 2.24 times what it would be without correlation, and the confidence interval is 124 % wider. A sample of 200 observations then carries the information of about 40.

Why it matters

The calculator helps students, researchers and analysts who check a regression for multicollinearity and want to explain what a given VIF means in practice.

Good to know

The VIF is computed for each predictor separately, so run the auxiliary regression for each one, or read the values from your software. Dummy variables created from the same category and interaction terms are often highly correlated by construction, and centering the variables can reduce the VIF of interaction terms. A high VIF affects the coefficients of the correlated predictors only, not the other ones nor the overall fit. If your goal is prediction rather than interpretation, multicollinearity matters less. The tolerance is the share of the predictor that is not explained by the others, so a tolerance below 0.2 matches a VIF above 5. In practice, look at the correlation matrix first to spot pairs of near-duplicates, then at the VIF to catch groups of predictors that are redundant together.

The linear regression calculator fits a line, the R-squared calculator measures fit, the partial correlation calculator removes the influence of a third variable and the correlation significance calculator tests a correlation.

Frequently asked questions

What does the VIF measure?

It measures how much the variance of a regression coefficient is inflated because the predictor is correlated with the others. If a predictor is regressed on all the other predictors with an R-squared of R², its VIF is 1 / (1 − R²).

Which thresholds are used?

A VIF of 1 means no correlation. Many authors worry above 5 and act above 10, which corresponds to R² of 80 % and 90 %. These are conventions: with a large sample, a VIF of 10 may still leave precise estimates, while with a small one a VIF of 3 can hurt.

What can I do about a high VIF?

Remove or combine the redundant predictors, for example by averaging them or using a principal component, collect more data, or use a regularized method such as ridge regression. Multicollinearity does not bias predictions, but it makes individual coefficients unstable.

Was this calculator helpful?

Thanks for your feedback!

Results are estimates provided for information purposes. They do not replace professional advice.