Panel Data Interrupted Time Series
Panel Data Interrupted Time Series Analysis · Also known as: panel ITS, multi-unit ITS, panel ITSA, controlled interrupted time series
Panel Data Interrupted Time Series (panel ITS) is a quasi-experimental method that estimates the causal effect of an intervention using repeated observations from multiple units over time. By exploiting variation across both units and time periods, it provides stronger causal identification than single-unit ITS, detecting changes in the level and slope of the outcome trajectory immediately following a clearly dated intervention.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
+1 more
When to use it
Use panel ITS when you have a naturally occurring, clearly dated policy change or intervention that affects multiple units (regions, hospitals, schools, firms) at approximately the same time, and when you observe each unit over enough time periods before and after the interruption — as a rule of thumb, at least 8 periods per segment is advisable. Panel ITS is preferable to single-unit ITS when you have access to multiple units, because pooling increases statistical power and allows control units to anchor the counterfactual. Do not use panel ITS when the intervention timing varies substantially across units without a clear common date (use staggered DiD instead), when the number of pre-intervention time points is very small (fewer than 6), or when the assumption that the pre-intervention trend would have continued unchanged is implausible.
Strengths & limitations
- Exploits both within-unit time variation and cross-unit variation, yielding more precise estimates than single-unit ITS.
- Separates two distinct causal mechanisms: an immediate level change and a gradual slope change in the outcome trajectory.
- Unit fixed effects absorb all time-invariant confounders that differ between units, improving internal validity.
- The sharply dated interruption provides a clean identification strategy that does not require a parallel-trends assumption across all pre-periods.
- Applicable in policy evaluation contexts where randomisation is infeasible but the intervention date is known and exogenous.
- Requires a clearly identifiable and externally imposed intervention date common (or approximately common) across units; endogenous or staggered timing complicates interpretation.
- The method assumes the pre-intervention trend would have continued linearly absent the intervention; non-linear or seasonally varying baselines need explicit modelling.
- A short pre-intervention window (fewer than 8 time points) leaves the baseline trend poorly estimated, undermining inference.
- Contemporaneous events ('history threat') that coincide with the intervention date and affect all units simultaneously can confound the estimate.
- Autocorrelation in residuals, if ignored, inflates the Type I error rate substantially.
Frequently asked
How is panel ITS different from standard single-unit ITS?
Standard ITS applies to a single aggregate series (e.g., one country or one hospital). Panel ITS stacks data from many units, adds unit fixed effects to control for stable unit differences, and typically yields more precise estimates because the greater sample size reduces uncertainty around the baseline trend and the intervention effect.
How many time points do I need before and after the intervention?
A common guideline is at least 8 time points per segment (pre and post) for adequate trend estimation. With panel data, having more units can partially compensate for a shorter time series, but fewer than 6 pre-intervention periods makes the baseline trend estimate unreliable regardless of the number of units.
Can I include control units that were not exposed to the intervention?
Yes, and doing so is strongly recommended when available. A controlled panel ITS adds unexposed units as a comparison group, which helps rule out history threats: if the outcome shifts in treated units but not in control units at the same date, the intervention is more plausibly responsible.
What if different units received the intervention at different times?
Staggered adoption breaks the single-interruption-point assumption of standard panel ITS. In that setting, use a staggered difference-in-differences estimator (e.g., Callaway-Sant'Anna or Sun-Abraham) or a unit-specific ITS combined in a meta-analytic framework.
How do I handle seasonality in the baseline?
Include harmonic terms (sine and cosine functions of calendar time) or month/quarter dummy variables in the pre-intervention segment regression. Failing to account for seasonality can produce a badly estimated counterfactual and a biased treatment-effect estimate.
Sources
- Lopez Bernal, J., Cummins, S., & Gasparrini, A. (2017). Interrupted time series regression for the evaluation of public health interventions: a tutorial. International Journal of Epidemiology, 46(1), 348-355. DOI: 10.1093/ije/dyw098 ↗
- Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and Quasi-Experimental Designs for Generalized Causal Inference. Houghton Mifflin. ISBN: 978-0395615560
How to cite this page
ScholarGate. (2026, June 3). Panel Data Interrupted Time Series Analysis. ScholarGate. https://scholargate.app/en/causal-inference/panel-data-interrupted-time-series
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Difference-in-DifferencesEconometrics↔ compare
- Interrupted Time SeriesCausal inference↔ compare
- Panel Data Difference-in-DifferencesCausal inference↔ compare
- Panel Fixed EffectsEconometrics↔ compare
- Synthetic Control MethodCausal inference↔ compare