Stats Calculator

Math

Compute count, sum, mean, median, mode, min, max, range, variance and standard deviation from a list of numbers for data analysis, reports and stats homework.

10 Count
#Count
10
ΣSum
34
μMean
3.4
MMedian
3
★Mode
3
↓Min
1
↑Max
6
↔Range
5
σ²Variance (population)
2.24
σStd Dev (population)
1.496662955
s²Variance (sample)
2.488888889
sStd Dev (sample)
1.577621275
Σx²Sum of Squares
138

About Stats Calculator

Descriptive statistics is the first step of data analysis — given a set of measurements or samples, you look at central tendency and dispersion before deciding how to model further. This tool computes eleven metrics at once: count, sum, mean, median, mode, min, max, range, variance, standard deviation and sum of squares. Enter numbers separated by commas or spaces. Everything runs locally in the browser, ideal for students checking homework, researchers scanning distributions and PMs reconciling report figures. For example, entering 2,4,4,4,5,5,7,9 yields mean 5, median 4.5, standard deviation about 2.2 and eight more metrics in one go.

Descriptive and inferential statistics answer different questions, and mixing them yields conclusions beyond what the data supports. Descriptive work says what this dataset looks like, such as centre and spread; inferential work generalises from a sample to a population with a measure of uncertainty. When the data is the entire object of study, a description suffices with no significance measures.

Outliers move the mean far more than the median, which decides which to lead with on skewed distributions. A few extremes drag the mean to one side so it stops representing a typical value, while the median depends only on position and resists extremes. On naturally skewed data such as income, duration or page views, report both and explain the gap.

Which denominator for variance depends on treating the data as a population or a sample. As the entire set, divide by the count; as a sample drawn from a larger population, divide by count minus one to correct for using the sample centre in place of the population centre. The same numbers give slightly different results under the two conventions, so state which is in use.

Quantiles have several accepted conventions, so one figure differs slightly between tools. When the count is not divisible by the quantile, an interpolation between neighbouring observations is needed, and the choice of interpolation, plus whether the sorted position is adjusted, changes the result. Compare conventions before debating whether a difference matters.

How missing values are handled is often the real source of variation, not the algorithm. Dropping them shrinks the sample but keeps all information, filling with zero systematically lowers the mean, and mean imputation leaves the mean unchanged while understating the spread. Count the missing share and fix the treatment in advance, documenting it rather than explaining at the result.

Finally, check relationships between the statistics to catch problems in the data or the pipeline. A minimum above the median, a negative standard deviation or a field that is constant all signal an error rather than an unusual distribution. One consistency check before concluding blocks a fair number of interesting findings that were really mistakes.

How It Works Internally

A statistic means little without the distribution: the mean is heavily influenced by extremes while the median is not, and the standard deviation describes spread but assumes roughly symmetric data. A large gap between mean and median usually signals a long tail or outliers, in which case reporting a single figure misleads.

InputIncome data 1, 2, 3, 4, 1000
OutputMean 202 against median 3 — one extreme value inflates the mean, so the median better represents the typical case

Measured on This Page

Core statisticsCount, sum, mean, median, range and sum of squares match statistics on all seven samples
Added hereVariance and standard deviation now come in both population (n) and sample (n−1) forms; the old build computed only population but said just Variance
How far apartFor n=5 it is 2 versus 2.5 (std dev 1.414 vs 1.581, 11.8% apart); for n=2 it is 1 versus 2 (41% apart)
Single valueThe sample form is undefined at n=1 and shows a dash instead of 0; the population form stays 0
Mode ruleWhen every value is distinct the page says there is no mode; ties are all listed rather than picking one

Variance has two forms that differ a lot: dividing by n is the population form (Excel STDEVP, Python statistics.pvariance), dividing by n−1 is the sample form (Excel STDEV.S, R's var). For n=5 the standard deviations differ by 11.8% and for n=2 by 41%, so check which form the other side used before reconciling numbers — otherwise it looks like the wrong one is broken.

How this was measured: Seven samples (the default ten values, consecutive integers, a single value, two values, negatives, decimals, all-identical) were entered one by one and the 13 cards' values and labels read back, then compared field by field against Python statistics — pvariance for the population form and variance for the sample form — with a 1e-9 tolerance, also determining which form the page used. · Measured 2026-10-28

How to Use

  1. Open "Stats Calculator"
  2. Paste numbers separated by commas or spaces
  3. Click "Sample" to try a ready-made dataset
  4. Cards update with mean, median, and more
  5. Copy any single metric from its card

Use Cases

  • Homework check — Paste exercise data to verify hand-calculated mean, variance and stddev and locate step errors.
  • Lab data scan — Quickly inspect the range and dispersion of measurements to judge instrument stability and outliers.
  • Report reconciliation — Compare mean vs median from backend exports to spot skew and outlier inflation.
  • Quality control — Track standard-deviation drift in production samples to catch process shifts early.
  • Rating summary — Use mode and median on review scores to resist extreme ratings skewing the mean.
  • A/B test quick scan — Compute mean and variance for both control and variant groups to gauge significance before running a full test.
  • Performance benchmark — Paste latency measurements to get mean, stddev and range for comparing framework or query performance.

FAQ

Why does mode show "No mode"?

When every value occurs equally often there is statistically no mode, and the tool reports so; if several values tie for most frequent, all are listed.

Is variance population or sample?

This is population variance (divided by n), suited to describing the dataset itself; for inferential work use sample variance (n-1).

Are decimals and negatives supported?

Yes. Any valid number (decimal, negative, scientific notation) is parsed; non-numeric tokens are ignored.

Are quantiles included?

This release focuses on eleven core metrics; quantiles are not yet provided but may be added later.

Mean and median differ a lot — which should I use?

Look at the distribution first. With outliers (a few huge orders, a few very long sessions) the mean is dragged away from the typical case and the median is more representative. Both are reported; the wider the gap, the more skewed the data and the more you should rely on percentiles.

My pasted data is not recognised — what now?

Splitting assumes common separators (comma, space, tab, newline). The usual traps are thousands separators copied from Excel (1,234) and full-width digits; strip thousands separators, convert full-width to half-width, or switch separators. Any non-numeric token also fails the whole parse.

Is the standard deviation sample or population?

Both are provided: the sample SD divides by n−1 (estimating a population from a sample) and the population SD divides by n (when the data is the whole set). Use the sample SD for tests and confidence intervals and the population SD to describe the data itself; mixing them understates spread.

How are empty values or text handled?

Non-numeric tokens are dropped during parsing and their count reported; only the rest enter the calculation. That means a forgotten header or unit row silently shrinks the sample. Confirm the "items used" count matches your expectation before reading the results.