Synthetic control builds a weighted combination of donor units to mirror the ______ period of the treated unit.
forecast
future
post‑test
pre‑intervention
Matching pre‑period trends ensures any divergence after treatment is attributable to the intervention.
Weights in synthetic control are chosen to minimise ______ error.
post‑period MAPE
p‑value
pre‑period RMSPE
lift
Root‑mean‑squared‑prediction‑error measures how closely the synthetic matches the treated before exposure.
The causal effect is computed as treated minus ______ outcome after intervention.
control group average
global mean
time trend
synthetic
This difference isolates the impact by comparing reality to the counter‑factual synthetic path.
Placebo tests iterate each donor unit as if treated to create a ______ distribution.
posterior
uniform
gamma
null
Comparing the actual gap to placebo gaps yields an empirical p‑value for significance.
Synthetic control is well‑suited when you have ______ treated unit(s) and many controls.
a single
zero
hundreds
clustered
The method shines with scarce treated entities where traditional panel regression lacks degrees of freedom.
Predictor variables often include lagged outcomes and relevant ______ factors.
creative IDs
cookie IDs
hashes
exogenous
Adding covariates like GDP or weather improves pre‑period fit beyond simple lag matching.
BSTS (Bayesian structural time series) offers a probabilistic alternative but shares the goal of building a ______ counter‑factual.
polynomial
synthetic
frequency‑domain
linear
Both methods construct an expected series absent treatment; BSTS adds posterior uncertainty intervals.
The open‑source **SynthPy 2.0** library released in 2025 adds automated ______ selection.
creative
geo
variable and weight
cookie
Regularisation routines choose optimal predictors and donor weights, reducing manual tuning.
A permutation p‑value is the share of placebo gaps that exceed the treated unit’s ______.
CTR
ad spend
variance
post‑period RMSPE
This forms a distribution‑free significance test of the observed effect size.
Synthetic control replaces parallel‑trends with a requirement of good pre‑period ______.
fit
color
p‑value
heteroskedasticity
Because weights replicate the treated path before treatment, we no longer need the strict linear‑trend assumption.
Starter
Review the basics.
Solid
Good job—refine for mastery.
Expert!
Outstanding performance.
