Omitted Variable Bias Quant Interview Guide
Omitted variable bias quant interview guide for missing drivers, distorted coefficients, causal caveats, examples, and mitigation.
Candidates discussing causal interpretation and regression caveats.
Omitted variables distort interpretation
Omitted variable bias appears when a missing driver affects the outcome and is related to an included predictor. The included coefficient can absorb part of the missing variable effect.
Bias needs two ingredients
A missing variable matters most when it is relevant to the outcome and correlated with the variable you are interpreting. Stating both ingredients makes the interview answer precise.
Concrete example
If a model links a signal to returns but omits liquidity, the signal coefficient may partly reflect liquidity exposure if signal strength and liquidity are related in the sample.
Mitigation is evidence-dependent
Possible responses include adding controls, changing the design, using instruments when justified, or avoiding causal language. The right answer depends on what data and assumptions the prompt allows.
Common mistakes
Candidates often say add more variables without asking whether the new controls are measured correctly or introduce new leakage. Better answers focus on the specific missing driver.
Practice the pattern
Use the LeetQuidity curriculum and calibration to turn this topic into a focused practice plan.