Expected value vs sample mean: parameter vs data

Expected value vs sample mean: the short answer

Both are averages, and both are often written with the word mean, which is why they blur together. The difference is what you average and whether the result is fixed or variable.

The expected value is a parameter: the theoretical long-run average of a probability distribution, written $E(X)$ or $\mu$ (mu). You compute it from probabilities, not data, by weighting each possible value by how likely it is. It is a single fixed number for a given distribution.

The sample mean, written $\bar{x}$ (x-bar), is a statistic: the ordinary average of the values you actually observed, computed by adding them up and dividing by the count. It changes every time you collect a new sample. The sample mean is your best estimate of the expected value, and by the law of large numbers it drifts toward the expected value as the sample grows.

What the expected value is

The expected value, or mean, of a probability distribution is a parameter denoted $E(X)$ or $\mu_X$. For a discrete random variable $X$, it is

where $x_i$ is a possible value of the variable and $P(x_i)$ is its probability. You multiply each value by its probability and add across all the values. The result is interpreted as the long-run average outcome of the random variable: what the average would settle toward over endless repetitions.

Because it comes from the probability distribution, the expected value is fixed and theoretical; no data collection is involved. A striking feature is that the expected value need not be a value the variable can actually take. A fair die has expected value 3.5, which is never rolled. That is fine, because it describes the long-run average, not any single outcome.

What the sample mean is

The sample mean is the everyday average of observed data. For $n$ observations $x_1, x_2, \ldots, x_n$, it is

the sum of the values divided by how many there are. It uses data, not probabilities, and it is a statistic, meaning a number computed from a sample.

Because it depends on which sample you happened to draw, the sample mean varies from sample to sample; that variability is what sampling distributions describe. The sample mean is an unbiased point estimator of the expected value: on average, across all possible samples, it neither overshoots nor undershoots $\mu$. Its own sampling distribution is centered at $\mu$ with standard deviation $\frac{\sigma}{\sqrt{n}}$, where $\sigma$ is the population standard deviation. For the vocabulary of a statistic estimating a parameter, see parameter vs statistic.

The differences side by side

Every row is the same split: the expected value is the fixed theoretical target, and the sample mean is the data-based estimate that aims at it.

When to use which

Use the expected value when you have a probability distribution and want its theoretical center: the fair price of a game, the long-run average payout, or the mean of a binomial or other random variable. Any time each outcome comes with a probability, weight the values by those probabilities. The expected value calculator handles the weighting for a discrete distribution.

Use the sample mean when you have actual data and want to summarize it or estimate an unknown population mean. Test scores, measured heights, and survey responses all call for adding and dividing. In inference, the sample mean $\bar{x}$ is the statistic you compute to estimate the population mean $\mu$, then surround with a margin of error. The short rule: probabilities in hand means expected value; data in hand means sample mean.

The classic mix-up and how to avoid it

The mix-up starts with the word mean, since both are called means, and deepens when a sample mean is treated as if it were the fixed truth. A sample mean of 3.2 from a few die rolls does not overturn the expected value of 3.5; it is just one estimate, and a different handful of rolls would give a different $\bar{x}$.

Two checks keep them apart. First, look at the inputs: if you multiplied values by probabilities, you found an expected value (a parameter); if you added data and divided by the count, you found a sample mean (a statistic). Second, watch the symbols: $\mu$ or $E(X)$ is the fixed target, while $\bar{x}$ is the estimate that varies and, by the law of large numbers, converges to $\mu$ as the sample grows. The connection between how the average settles and how its distribution takes shape is drawn out in law of large numbers vs central limit theorem.