Descriptives Calculator
Paste numbers and get the full descriptive rundown: mean, SD, median, quartiles, confidence interval, skewness, outlier flags, and a picture. Nothing leaves your browser, and the page writes the APA-style summary sentence for you.
The three conventions give three answers to the same exam question. Textbook splits the sorted values at the median and takes the median of each half, leaving the middle value out of both halves when n is odd. That is the hand method Moore, McCabe and Craig teach, and the one a paper exam expects. SPSS uses the weighted average that its Frequencies menu produces by default. R/Python interpolates the way R, NumPy, pandas and Excel's QUARTILE.INC all do, and it stays the default here. The switch moves Q1, Q3, the IQR, the fences, the outlier flags and the boxplot with it; the median never moves, because all three agree on that one.
What to check before you trust the mean
- Skewness far from 0? The mean is being dragged toward the tail, so report the median too. (Describing Data)
- Outliers flagged? Don't delete them by reflex; find out what they are first. One typo can move a mean more than ten correctly entered values. (ANOVA & t-Test Assumptions)
- Small n? The 95% CI here uses the t-distribution, so it widens correctly for small samples. No formula makes n = 5 precise, though. (Confidence Intervals)
- Histogram looks different from your software's? Bin counts are a choice, not a fact. This page uses the square-root rule (⌈1.5√n⌉, capped at 30), and SPSS, R and Excel each choose differently. Visualizing Data applies Sturges' rule, Scott's rule and Freedman–Diaconis to the same data to show how different the results are.
Privacy note: everything on this page runs locally in your browser. Nothing you paste ever leaves this machine, so paste away.
Two variables instead of one? The Correlation & Regression Calculator plots X against Y and gives you r, the best-fit line, and an outlier check. No data yet? Grab a practice dataset.