Mean

The mean is the sum of a set of values divided by how many values there are — the arithmetic average, and the most commonly used measure of central tendency in statistics. It's the number most people reach for first when asked to summarize "what's typical" about a dataset, and for data that's roughly symmetric and free of extreme outliers, it's usually a reasonable choice — but that reasonableness depends entirely on the shape of the underlying distribution, a dependency that's easy to overlook precisely because computing a mean takes the same single formula regardless of whether that assumption actually holds.

The mean's defining mathematical property is also its primary weakness: it's sensitive to every value in the dataset equally, including extreme ones, which means a small number of unusually large or small values can pull the mean substantially away from where most of the data actually sits. Household income is the textbook example: a neighborhood where most households earn around $60,000 but one household earns $5 million will show a mean income dramatically higher than what nearly every actual household earns, because that single extreme value is weighted exactly as heavily in the sum as every typical value — which is precisely why income statistics are more often reported as a median than a mean, and why checking whether a dataset is skewed or contains outliers is a standard first step before trusting a reported mean as representative of the typical case.

The population mean (denoted μ) and the sample mean (denoted x̄) are computed with the identical formula but serve different conceptual roles: μ represents the true, usually unknowable average of an entire population, while x̄ is an estimate of that true value computed from a sample — and a sample mean is itself a random variable, meaning a different random sample from the same population would almost certainly produce a slightly different x̄, which is the entire basis for the standard error and confidence intervals used in inferential statistics to express how much a sample mean might differ from the true population mean it's estimating.

Variants of the basic arithmetic mean exist specifically for data where a straight average would be misleading: the weighted mean assigns different importance to different values (a grade point average weighting each course by its credit hours, rather than treating a one-credit and a four-credit course as equally important), the geometric mean is used for data that compounds multiplicatively rather than adds (average investment growth rates across several years, where a straight arithmetic mean would overstate true compounded growth), and the trimmed mean removes a fixed percentage of the most extreme values from both ends before averaging, specifically to reduce the outlier sensitivity that makes the ordinary mean unreliable on skewed data without abandoning the mean's basic logic entirely.

Read the mean as the most common but also the most outlier-sensitive measure of central tendency: its exact equal-weighting of every value is what makes it mathematically simple and tractable for further statistical theory, but that same property is exactly what lets a handful of extreme values distort it away from representing the typical case, which is why checking a distribution's shape — and reaching for the median or a trimmed mean when that shape is skewed — is standard practice rather than an afterthought in careful statistical reporting.