An estimate is a random variable (the key idea of inference)
This may be the most important lesson of the whole course — keep it in mind through all the following ones. All we have is a sample, and from it we ESTIMATE something: a mean, a share, a difference. The key insight: any such estimate is itself a random variable. Take another sample — you get another number. And this "wandering" has its own distribution, on which all of statistical inference rests.
Whichever estimate you pick — every sample has its own, i.e. it is itself random. Its spread falls as n grows. That is the sampling distribution — the object the CLT, intervals and hypothesis testing all stand on.
Let's separate two things that are easy to confuse. The truth (the estimand) is a number about the whole population: the true mean μ, the true share p. It is single and fixed, but unknown to us. The estimate (the estimator) is what we compute from a sample: x̄, p̂. It is known to us, but changes from sample to sample. The yellow line is the truth (one); the teal histogram below is how the estimate dances.
What it means
The idea that every estimate is random changes how you read any number in a report. "Conversion is 12%" sounds like a fact, but it is not the truth — it is one realization of a random estimate: collect the data for a different week and you would get 11% or 13%, while the real conversion never moved. So the number alone, without a sense of its spread, is nearly useless: 12% from a sample of 50 and 12% from a sample of 50,000 are two completely different degrees of confidence, even though the figure is the same. The right question to ask of any estimate is not "what is it?" but "how widely could it have scattered?".
Two practical conclusions follow. First: an estimate is judged as an instrument — a good one is unbiased (on average it hits the truth rather than missing to one side every time) and narrow (it varies little from sample to sample). Second: that spread is under your control — the larger the sample, the more tightly estimates cluster around the truth. So "how much data to collect" is not a technical question but the main lever of precision.
Where it shows up
Polls, ratings, quality measurements, product metrics — everywhere you hold only a sample-based estimate. A professional automatically keeps its sampling distribution in mind and never confuses "the number we got" with "the truth".
That is why serious reports show not a bare point but an estimate with an interval — an honest admission that it is a random variable, not a verdict.
Definitions
When the method lies (assumptions)
The formula for the standard error, σ/√n (meet it properly in the next lesson), assumes independent observations. If rows are clustered (several visits by one person, cities, ad campaigns), the effective n is smaller than the row count — the real error is larger than the formula says.
The standard error describes only random scatter. Systematic error — an unrepresentative sample, a biased measurement — is not cured by √n: as n grows you converge ever more confidently to the wrong answer.
Deep dive: the math and the mechanism (optional)
Formally an estimate is a function of a random sample, so it is itself a random variable. It has an expectation (where it aims on average) and a variance (how much it wanders). Unbiasedness means "expectation of the estimate = the thing being estimated"; efficiency means "smaller variance at the same sample size".
We will see next that for the mean this distribution has width σ/√n and tends to the normal one (the CLT) — but the idea itself, that an estimate is random and has a distribution, holds for any statistic, even when there is no formula for it (that is when the bootstrap comes to the rescue).