Sampling, Estimation and Statistical Inference
EVIDENCE LAB · 05
Sampling
& Inference.
We rarely observe an entire economy. Learn how a carefully selected sample can estimate population characteristics—and how uncertainty must accompany every estimate.
SAMPLEESTIMATEINFER
01 · THE INFERENTIAL JOURNEY
From observed sample
to unknown population.
The target
Every household, worker, firm or region about which the research question seeks knowledge.
→
The evidence
A subset selected through a transparent design that aims to represent the target population.
→
The conclusion
An estimate accompanied by a quantified statement of sampling uncertainty and assumptions.
02 · TWO SOURCES OF UNCERTAINTY
More observations help.
But they do not cure bias.
Chance variation
Different random samples produce different estimates. Larger samples generally reduce this variation at the square-root rate.
Address with probability sampling, standard errors and confidence intervals.
Systematic distortion
Coverage gaps, nonresponse, faulty questions, measurement error or processing mistakes can bias even a very large sample.
Address through design, validation, weighting, documentation and sensitivity analysis.
INTERACTIVE SAMPLING SIMULATOR
Change the sample.
Watch precision respond.
Set a sample estimate, variability, sample size and confidence level. ECONORIA calculates the standard error, margin of error and confidence interval.
INFERENTIAL READINGThe 95% confidence interval extends from 47.06 to 52.94.
03 · THE SQUARE-ROOT RULE
To halve the standard error,
quadruple the sample.
Baseline precision
Half the standard error
One-quarter of the standard error
A 95% confidence procedure captures the true parameter in 95% of repeated samples under its assumptions. It does not assign a 95% probability to a fixed parameter after this interval is observed.
