Checking for multicollinearity

What is the Variance Inflation Factor (VIF)?

The Variance Inflation Factor (VIF) is a tool you can use to check for multicollinearity in your regression model. Multicollinearity occurs when two or more independent variables in a regression model are highly correlated with each other, which can lead to inaccurate and unreliable estimates of the regression coefficients.

Before you can carry out the VIF test, you must:

  • Have created a linear regression model.
  • Have multiple independent variables that you think might be correlated with each other.

How is the test carried out?

Implementation in R with an example

In R, you can calculate the VIF using the car package:

# Paket installieren und laden
install.packages("car")
library(car)

# Fiktiver Datensatz
set.seed(123)
x1 <- rnorm(100)
x2 <- x1 + rnorm(100, sd=0.5)
y <- 3 + 2*x1 + 0.5*x2 + rnorm(100)
daten <- data.frame(y, x1, x2)

# Regressionsmodell erstellen
modell <- lm(y ~ x1 + x2, data=daten)

# VIF berechnen
vif(modell)

This code returns the VIF value for each independent variable in your model.

Implementation in SPSS

In SPSS:

  1. Run a linear regression: “Analyze” > “Regression” > “Linear”.
  2. Add the dependent variable and the independent variables.
  3. Under “Statistics”, select “Collinearity diagnostics”.
  4. Click “OK”.

In the results, you will find the VIF value for each independent variable.

Implementation in JASP

In JASP:

  1. Select “Regression” > “Linear Regression”.
  2. Add the dependent variable and the independent variables.
  3. Enable the “Collinearity” option.
  4. The “Collinearity” table shows the VIF for each variable.

How is the Variance Inflation Factor interpreted and reported?

Suppose you obtain a VIF value of 5 for x1 and a value of 4.5 for x2.

A VIF value of 1 indicates no multicollinearity. In general, values above 10 are often a cause for concern, although some experts advise caution even with values above 5. In our example, the values of 5 and 4.5 would indicate that there may be multicollinearity between x1 and x2.

In APA format, you would write: “A test of multicollinearity revealed VIF values of 5 for x1 and 4.5 for x2, indicating possible multicollinearity between these variables.”

Conclusion

The variance inflation factor (VIF) is an indispensable tool when you want to ensure the reliability of your regression model. Multicollinearity can distort the results and lead to incorrect conclusions. It is therefore crucial to check for it and correct it where necessary. Whether you are working in R, SPSS, or JASP, the VIF gives you clarity about the presence of multicollinearity in your model.