# Variance inflation factor

In statistics, the variance inflation factor (VIF) is the quotient of the variance in a model with multiple terms by the variance of a model with one term alone.[1] It quantifies the severity of multicollinearity in an ordinary least squares regression analysis. It provides an index that measures how much the variance (the square of the estimate's standard deviation) of an estimated regression coefficient is increased because of collinearity.

## Definition

Consider the following linear model with k independent variables:

Y = β0 + β1 X1 + β2 X 2 + ... + βk Xk + ε.

The standard error of the estimate of βj is the square root of the j + 1 diagonal element of s2(XX)1, where s is the root mean squared error (RMSE) (note that RMSE2 is a consistent estimator of the true variance of the error term, ${\displaystyle \sigma ^{2}}$); X is the regression design matrix a matrix such that Xi, j+1 is the value of the jth independent variable for the ith case or observation, and such that Xi,1, the predictor vector associated with the intercept term, equals 1 for all i. It turns out that the square of this standard error, the estimated variance of the estimate of βj, can be equivalently expressed as:[2][3]

${\displaystyle {\widehat {\operatorname {var} }}({\hat {\beta }}_{j})={\frac {s^{2}}{(n-1){\widehat {\operatorname {var} }}(X_{j})}}\cdot {\frac {1}{1-R_{j}^{2}}},}$

where Rj2 is the multiple R2 for the regression of Xj on the other covariates (a regression that does not involve the response variable Y). This identity separates the influences of several distinct factors on the variance of the coefficient estimate:

• s2: greater scatter in the data around the regression surface leads to proportionately more variance in the coefficient estimates
• n: greater sample size results in proportionately less variance in the coefficient estimates
• ${\displaystyle {\widehat {\operatorname {var} }}(X_{j})}$: greater variability in a particular covariate leads to proportionately less variance in the corresponding coefficient estimate

The remaining term, 1 / (1  Rj2) is the VIF. It reflects all other factors that influence the uncertainty in the coefficient estimates. The VIF equals 1 when the vector Xj is orthogonal to each column of the design matrix for the regression of Xj on the other covariates. By contrast, the VIF is greater than 1 when the vector Xj is not orthogonal to all columns of the design matrix for the regression of Xj on the other covariates. Finally, note that the VIF is invariant to the scaling of the variables (that is, we could scale each variable Xj by a constant cj without changing the VIF).

${\displaystyle {\widehat {\operatorname {var} }}({\hat {\beta }}_{j})=s^{2}(X^{T}X)^{-1}}$

Now let ${\displaystyle r=X^{T}X}$, and without losing generality, we reorder the columns of X to set the first column to be ${\displaystyle X_{j}}$

${\displaystyle r^{-1}={\begin{bmatrix}r_{j,j}&r_{j,-j}\\r_{-j,j}&r_{-j,-j}\end{bmatrix}}^{-1}}$
${\displaystyle r_{j,j}=X_{j}^{T}X_{j},r_{j,-j}=X_{j}^{T}X_{-j},r_{-j,j}=X_{-j}^{T}X_{j},r_{-j,-j}=X_{-j}^{T}X_{-j}}$.

By using Schur complement, the element in the first row and first column in ${\displaystyle r^{-1}}$ is,

${\displaystyle r_{1,1}^{-1}=[r_{j,j}-r_{j,-j}r_{-j,-j}^{-1}r_{-j,j}]^{-1}}$

Then we have,

{\displaystyle {\begin{aligned}&{\widehat {\operatorname {var} }}({\hat {\beta }}_{j})=s^{2}(X^{T}X)^{-1}=s^{2}r_{1,1}^{-1}\\={}&s^{2}[X_{j}^{T}X_{j}-X_{j}^{T}X_{-j}(X_{-j}^{T}X_{-j})^{-1}X_{-j}^{T}X_{j}]^{-1}\\={}&s^{2}[X_{j}^{T}X_{j}-X_{j}^{T}X_{-j}(X_{-j}^{T}X_{-j})^{-1}(X_{-j}^{T}X_{-j})(X_{-j}^{T}X_{-j})^{-1}X_{-j}^{T}X_{j}]^{-1}\\={}&s^{2}[X_{j}^{T}X_{j}-{\hat {\beta }}_{*j}^{T}(X_{-j}^{T}X_{-j}){\hat {\beta }}_{*j}]^{-1}\\={}&s^{2}{\frac {1}{\mathrm {RSS} _{j}}}\\={}&{\frac {s^{2}}{(n-1){\widehat {\operatorname {var} }}(X_{j})}}\cdot {\frac {1}{1-R_{j}^{2}}}\end{aligned}}}

Here ${\displaystyle {\hat {\beta }}_{*j}}$ is the coefficient of regression of dependent variable ${\displaystyle X_{j}}$ over covariate ${\displaystyle X_{-j}}$. ${\displaystyle \mathrm {RSS} _{j}}$ is the corresponding residual sum of squares.

## Calculation and analysis

We can calculate k different VIFs (one for each Xi) in three steps:

### Step one

First we run an ordinary least square regression that has Xi as a function of all the other explanatory variables in the first equation.
If i = 1, for example, equation would be

${\displaystyle X_{1}=\alpha _{0}+\alpha _{2}X_{2}+\alpha _{3}X_{3}+\cdots +\alpha _{k}X_{k}+e}$

where ${\displaystyle \alpha _{0}}$ is a constant and e is the error term.

### Step two

Then, calculate the VIF factor for ${\displaystyle {\hat {\alpha }}_{i}}$ with the following formula :

${\displaystyle \mathrm {VIF} _{i}={\frac {1}{1-R_{i}^{2}}}}$

where R2i is the coefficient of determination of the regression equation in step one, with ${\displaystyle X_{i}}$ on the left hand side, and all other predictor variables (all the other X variables) on the right hand side.

### Step three

Analyze the magnitude of multicollinearity by considering the size of the ${\displaystyle \operatorname {VIF} ({\hat {\beta }}_{i})}$. A rule of thumb is that if ${\displaystyle \operatorname {VIF} ({\hat {\beta }}_{i})>10}$ then multicollinearity is high[4] (a cutoff of 5 is also commonly used[5]).

Some software instead calculates the tolerance which is just the reciprocal of the VIF. The choice of which to use is a matter of personal preference. .

## Interpretation

The square root of the variance inflation factor indicates how much larger the standard error increases compared to if that variable had 0 correlation to other predictor variables in the model.

Example
If the variance inflation factor of a predictor variable were 5.27 (√5.27 = 2.3), this means that the standard error for the coefficient of that predictor variable is 2.3 times larger than if that predictor variable had 0 correlation with the other predictor variables.

## Implementation

• vif function in the car R package
• ols_vif_tol function in the olsrr R package
• PROC REG in SAS System
• variance_inflation_factor function in statsmodels Python package
• estat vif in Stata

## References

1. James, Gareth; Witten, Daniela; Hastie, Trevor; Tibshirani, Robert (2017). An Introduction to Statistical Learning (8th ed.). Springer Science+Business Media New York. ISBN 978-1-4614-7138-7.
2. Rawlings, John O.; Pantula, Sastry G.; Dickey, David A. (1998). Applied regression analysis : a research tool (Second ed.). New York: Springer. pp. 372, 373. ISBN 0387227539. OCLC 54851769.
3. Faraway, Julian J. (2002). Practical Regression and Anova using R (PDF). pp. 117, 118.
4. Kutner, M. H.; Nachtsheim, C. J.; Neter, J. (2004). Applied Linear Regression Models (4th ed.). McGraw-Hill Irwin.
5. Sheather, Simon (2009). A modern approach to regression with R. New York, NY: Springer. ISBN 978-0-387-09607-0.
• Allison, P. D. (1999). Multiple Regression: A Primer. Thousand Oaks, CA: Pine Forge Press. p. 142.
• Hair, J. F.; Anderson, R.; Tatham, R. L.; Black, W. C. (2006). Multivariate Data Analysis. Upper Saddle River, NJ: Prentice Hall.
• Kutner, M. H.; Nachtsheim, C. J.; Neter, J. (2004). Applied Linear Regression Models (4th ed.). McGraw-Hill Irwin.
• Longnecker, M. T.; Ott, R. L. (2004). A First Course in Statistical Methods. Thomson Brooks/Cole. p. 615.
• Marquardt, D. W. (1970). "Generalized Inverses, Ridge Regression, Biased Linear Estimation, and Nonlinear Estimation". Technometrics. 12 (3): 591–612 [pp. 605–7]. doi:10.1080/00401706.1970.10488699.
• Studenmund, A. H. (2006). Using Econometrics: A Practical Guide (5th ed.). Pearson International. pp. 258–259.
• Zuur, A.F.; Ieno, E.N.; Elphick, C.S (2010). "A protocol for data exploration to avoid common statistical problems". Methods in Ecology and Evolution. 1: 3–14. doi:10.1111/j.2041-210X.2009.00001.x.