When less conditioning provides better estimates: overcontrol and endogenous selection biases in research on...

Area of Science:

• Social Sciences
• Sociology
• Causal Inference

Background:

• The counterfactual approach is dominant in social science causality.
• Omitted variable bias is known, but overcontrol and endogenous selection biases are less recognized.
• Common practices in intergenerational mobility research are susceptible to these biases.

Purpose of the Study:

• To review common practices in intergenerational mobility research through the lens of the counterfactual approach.
• To identify and explain the sources of overcontrol and endogenous selection biases in these practices.
• To demonstrate the practical impact of these biases on research findings.

Main Methods:

• Review of four common practices in intergenerational mobility research.
• Application of the counterfactual framework to analyze potential biases.
• Empirical analysis using data from the German Socio-Economic Panel Study (SOEP).

Main Results:

• Widely used methods in intergenerational mobility research are shown to be vulnerable to overcontrol and endogenous selection biases.
• These biases can significantly distort estimates of causal effects.
• Empirical evidence from the SOEP illustrates the practical consequences of these biases.

Conclusions:

• Future research on intergenerational mobility must carefully consider the potential for bias from conditioning on variables.
• Researchers should be more aware of and actively address overcontrol and endogenous selection biases.
• Rethinking methodological practices is crucial for accurate causal inference in this field.