Panel data combines time-series and cross-sectional data by observing the same variables from the same sample over multiple time periods. This allows researchers to answer questions that cannot be addressed using only cross-sectional or time-series data. There are two main approaches to estimating panel data models: fixed effects and random effects. The fixed effects model controls for time-invariant characteristics of cross-sectional units by including dummy variables for each unit, avoiding omitted variable bias. However, it uses up many degrees of freedom. The random effects model assumes cross-sectional differences are random variables, conserving degrees of freedom but requiring those variables be uncorrelated with regressors. Researchers use tests like Hausman to determine whether fixed or random effects is preferred