Concept Architecture
Concept
Theoretically, Variance Inflation Factor (VIF) is a diagnostic measure that quantifies the extent to which multicollinearity inflates the variance of a regression coefficient estimate. It measures how strongly a predictor variable is linearly related to the remaining predictors in a regression model. The concept is founded on multiple linear regression theory and is used to assess the stability and precision of estimated regression coefficients.
Mathematically, the Variance Inflation Factor is calculated from the coefficient of determination obtained by regressing a predictor variable on all remaining explanatory variables. As the explanatory power of this auxiliary regression increases, the VIF increases, indicating greater multicollinearity and a corresponding increase in the variance of the estimated regression coefficient. A VIF of 1 indicates no multicollinearity, while progressively larger values indicate increasing variance inflation.
In practice, VIF is routinely calculated during regression diagnostics before interpreting model coefficients. Thresholds of 5 or 10 are commonly used as practical indicators of problematic multicollinearity, although interpretation depends on the modelling context. In health economics, VIF is frequently applied when modelling healthcare costs, quality-adjusted life-years, utilisation and clinical outcomes to ensure that correlated explanatory variables do not compromise inference or prediction.
Purpose
Used to diagnose multicollinearity among explanatory variables by quantifying the increase in the variance of regression coefficient estimates attributable to linear dependence between predictors.
Mathematical Formulae
Primary Formula
VIF? = 1 / (1 ? R�?)
where:
- R�? = coefficient of determination obtained by regressing predictor X? on all remaining predictors.
Supporting Formulae
Tolerance:
Tolerance? = 1 / VIF? = 1 ? R�?
Variance of the estimated regression coefficient:
Var(???) = �� ? VIF? / SXX?
where:
- �� = residual variance
- SXX? = ?(x? ? x??)�
Related Mathematical Methods
- Multiple Linear Regression
- Collinearity Diagnostics
- Tolerance
- Ordinary Least Squares
- Coefficient of Determination (R�)
- Condition Index
- Eigenvalue Analysis
Example
A health economist develops a regression model predicting annual healthcare costs using age, comorbidity score, body mass index and number of chronic conditions. Regressing body mass index on the remaining predictors yields:
R� = 0.80
The Variance Inflation Factor is:
VIF = 1 / (1 ? 0.80)
= 1 / 0.20
= 5.0
This indicates that the variance of the estimated regression coefficient for body mass index is inflated five-fold because of multicollinearity with the other explanatory variables.
Excel Implementation
| Function | Example Formula | Health Economics Application |
|---|---|---|
| RSQ | =RSQ(B2:B101,C2:C101) | Calculate R� for simple exploratory assessment of predictor relationships. |
| LINEST | =LINEST(B2:B101,C2:E101,TRUE,TRUE) | Estimate auxiliary regression models used to obtain R� values. |
| POWER | =1/(1-F2) | Calculate the Variance Inflation Factor from the auxiliary regression R� stored in cell F2. |
| IF | =IF(G2>5,"Investigate multicollinearity","Acceptable") | Flag predictors with potentially problematic multicollinearity. |
VBA (Optional)
Automate calculation and reporting of Variance Inflation Factors for all explanatory variables within regression models used in health economic analyses.
Sources
- Montgomery DC, Peck EA, Vining GG. Introduction to Linear Regression Analysis.
- Kutner MH, Nachtsheim CJ, Neter J, Li W. Applied Linear Statistical Models.
- Belsley DA, Kuh E, Welsch RE. Regression Diagnostics: Identifying Influential Data and Sources of Collinearity.
- Draper NR, Smith H. Applied Regression Analysis.
- Drummond MF, et al. Methods for the Economic Evaluation of Health Care Programmes.
- Briggs A, Claxton K, Sculpher M. Decision Modelling for Health Economic Evaluation.
Related Concepts (3)
Library
Publications
1
Statistical Analysis of Cost-Effectiveness Data — Willan & Briggs, 1st Edition ed., 2006 (John Wiley & Sons)
A synthesis of statistical methods for analysing cost-effectiveness data, including net-benefit regression, confidence intervals for the ICER, cost-effectiveness acceptability curves, and covariate adjustment. Part of the Wiley Statistics in Practice series.
BookView source →
Frequently Asked Questions (6)
What is the variance inflation factor?
A regression diagnostic quantifying how much an estimated coefficient's variance is inflated by correlation with other predictors, higher values meaning worse multicollinearity.
Source: Belsley, Kuh & Welsch 1980
What does the variance inflation factor quantify about a coefficient?
The variance inflation factor quantifies how much the variance of an estimated regression coefficient is enlarged because its predictor is correlated with the others. A value of one means no inflation, the predictor being independent of the rest, while large values mean collinearity has swollen the uncertainty around that coefficient, making it unstable and hard to interpret. It is the reciprocal of the tolerance, and values above about five or ten are commonly taken as warnings. Measuring collinearity's toll on a coefficient is its purpose. Kirkwood and Sterne (2003) describe this diagnostic.
Source: Kirkwood & Sterne 2003
How is the variance inflation factor calculated?
The variance inflation factor for a predictor is calculated as one divided by its tolerance, where the tolerance is one minus the coefficient of determination from regressing that predictor on all the others; equivalently, it is one divided by one minus that coefficient of determination. So the variance inflation factor is calculated from how well a predictor is explained by the other predictors, with a strong correlation giving a high coefficient of determination, a low tolerance, and hence a high variance inflation factor, which is why it directly reflects the degree of multicollinearity affecting that predictor's coefficient.
Source: Belsley, Kuh & Welsch 1980
How is the variance inflation factor interpreted?
The variance inflation factor is interpreted by its size, with higher values indicating greater inflation of the coefficient's variance due to multicollinearity; a value of one indicates no correlation with the other predictors, while rough guidelines flag values above five or ten as indicating problematic collinearity. So the variance inflation factor is interpreted as how many times larger the coefficient's variance is because of correlation with the other predictors, with large values signalling that the estimate is unstable and imprecise, which is why high variance inflation factors prompt investigation of the multicollinearity, though the thresholds are conventions rather than strict cut-offs.
Source: Belsley, Kuh & Welsch 1980
What does a high variance inflation factor indicate?
A high variance inflation factor indicates that a predictor is strongly correlated with the other predictors, so its coefficient's variance is greatly inflated, making the estimate unstable and imprecise and its individual effect hard to separate from the others. So a high variance inflation factor signals problematic multicollinearity for that predictor, warning that its coefficient may be unreliable and sensitive to small changes in the data, which is why it prompts the analyst to consider remedies such as removing or combining predictors, though whether action is needed depends on whether the interest is in the individual coefficients or only in prediction.
Source: Belsley, Kuh & Welsch 1980
How is the variance inflation factor used?
The variance inflation factor is used to detect and quantify multicollinearity in a regression by computing it for each predictor and identifying those with high values, whose coefficients are most affected by correlation with the others. So the variance inflation factor is used as a routine diagnostic when fitting regressions with several predictors, flagging collinearity that could undermine the reliability of individual coefficient estimates, which allows the analyst to decide whether to address it, and it is a standard tool alongside tolerance, condition indices, and examination of the correlations for assessing multicollinearity.
Source: Belsley, Kuh & Welsch 1980
Trust Record
Verified by Dr Darrin Baines
British health economist
Professional identity: darrinbaines.org
Verification date: 26 Dec 2025
Content version: 1.0.0
Canonical Identity
- Term code
- HE-ES-SA-233
Stable URI · Machine-readable · Resolvable · CC BY 4.0