Panel Data, where cross-section meets time series
Panel data follow the same entities over time, splitting variation into within and between, so each entity can act as its own control through the unobserved-effects model.
A 5.5 minute lesson on panel data, the workhorse of applied project econometrics. Built for ECON3006 and FIN301 students at Western Sydney University, worked in Stata.
Panel data follow the same N entities across T time periods, so every observation carries an entity and a time index. The video splits the variation into between (how firms differ from each other) and within (how one firm changes over time), and introduces the unobserved-effects model, where a time-invariant term captures everything stable about a firm that you never measured.
That is why panel data dominate empirical projects. Watching a firm across years lets it act as its own control, holding fixed the unmeasured quality a single cross-section leaves trapped in the error. Pair the video with the Atlas concept page for the formulas, xtsum and xtdescribe in Stata, a quiz, and citations.