Skip to content

Comparative Historical Timeline

Analytical artifact — instantiates Structured Comparative Case Design

Lines up the sequence of events across cases on one shared clock so you can see whether the supposed cause actually came before the effect in each.

A cross-case pattern can survive every check except the oldest one: did the cause come before the effect, or did it merely accompany it? The Comparative Historical Timeline is the artifact that answers this by plotting each case's key events on a shared, aligned time axis. Its distinguishing purpose is temporal order across cases — establishing that the putative cause preceded the outcome in every case (ruling out reverse causation and coincidence) — and, to make that comparison legitimate, aligning the cases' clocks so that "before" and "after" mean the same thing in each. It is the shallow, cross-case sequence view, complementary to the deep single-case tracing that examines one mechanism in detail.

Example

Two central banks meet a similar inflation shock; one contains it, the other does not. Did earlier rate hikes cause the difference — or did each bank hike earlier because its inflation happened to peak sooner? A calendar comparison cannot tell. The timeline aligns both cases not on calendar dates but on months from the shock, and plots each move: shock onset, first hike, communication shift, inflation peak, return to target.

Aligned this way, the ordering separates the cases: one bank's first hike lands two months before its inflation peak, the other's lands after it — a difference a side-by-side of raw dates would have buried. Getting there forced one equivalence decision that the analysis had to defend: what counts as "t = 0," the shock, defined identically for both. The output is a shared timeline showing the candidate cause preceding the effect in one case and lagging it in the other — the temporal spine the rest of the design hangs on.

How it works

  • Choose a common anchor (t = 0) and define it identically across cases — the equivalence step that makes alignment honest.
  • Plot each case's cause-events and outcome-events on the shared axis at the same resolution.
  • Read off temporal order: did the candidate cause precede the outcome in every case, or only some?
  • Flag sequence anomalies — effect-before-cause, or simultaneity — and hand them to the rival table as live alternatives.

Tuning parameters

  • Anchor choice — what defines t = 0 (a shock, a decision, an onset); the anchor decides what "aligned" means and can make or break the comparison.
  • Temporal granularity — days versus quarters versus eras; finer resolution reveals order but demands finer-grained data.
  • Event-inclusion threshold — how much detail per case; more events show mechanism but clutter the contrast.
  • Alignment mode — calendar time versus event-relative time (months-from-onset); event-relative alignment is what exposes ordering differences hidden by the calendar.

When it helps, and when it misleads

Its strength is that it makes temporal precedence — a near-non-negotiable requirement for causation — visible and comparable, and it catches reverse causation that a static side-by-side misses entirely. Its failure mode is that aligning clocks forces judgment calls (when did the "cause" truly begin?) that can manufacture or erase an ordering, and event-equivalence across genuinely different cases is often strained. The classic misuse is sliding the anchor until the sequence supports the desired story. The discipline that guards against this is to fix the anchor and the event definitions before plotting, to hand every before-the-cause anomaly to the rival table, and to treat cross-case event equivalence as a claim to defend rather than assume.[1]

How it implements the components

  • temporal_order_and_process_trace — its core: the cross-case sequence of cause- and outcome-events on one axis, establishing what preceded what in each case.
  • measurement_equivalence_check — the timeline's slice of equivalence: defining the anchor and the event categories identically across cases, so their clocks are actually comparable.

It supplies cross-case sequence, not the within-case causal mechanism (that's within-case process tracing), and it checks only temporal and event equivalence, not the full construct equivalence of every measure in the study (that's the measurement equivalence audit).

References

[1] Temporal precedence — the requirement that a cause occur before its effect, among the least negotiable criteria of causal inference (and one of Bradford Hill's considerations). Making it visible across cases is exactly what a comparative timeline is for. ↩