Definition
Convergence describes an estimator or sequence approaching a target as sample size, simulation count or another index grows. The mode matters: convergence in probability, almost surely, in distribution and in mean are different claims. A convergence statement is meaningful only with its assumptions and limiting target.
How it is calculated
Theta-hat sub n converges in probability to theta: for every positive epsilon, the probability of absolute error above epsilon tends to zero.
| Symbol | Meaning | Unit |
|---|---|---|
| theta | limiting target | target units |
| n | sample-size index | count |
| epsilon | positive tolerance | target units |
Interpretation
Convergence is a statement about a sequence of experiments, not a promise about one finite sample. It tells you what happens as information or computation increases under the model. The rate can be slow, and the assumptions can fail in the setting that matters.
Assumptions and conventions
-
The stated regularity conditions hold as the index grows.
-
The limit and mode of convergence are explicit.
Estimator and target share units; epsilon has target units.
Limitations and common misreadings
-
Convergence does not say how large a finite sample is enough.
-
Convergence in distribution is not the same as convergence of numerical values.
References
- Larry Wasserman. (2004). All of Statistics: A Concise Course in Statistical Inference. Springer Texts in Statistics. Chapter 5, Convergence of Random Variables. doi:10.1007/978-0-387-21736-9. Source

