Outlier Visual Detection Tool — Free IQR & Z-Score Finder

This tool finds outliers — values far from the rest of your data — using two standard rules, and draws your values as a scatter plot so you can see them. The IQR rule looks at the middle fifty percent of your data: anything further than a set multiple of that spread beyond the quartiles is flagged. The z-score rule measures how many standard deviations each value sits from the mean. Red dots on the plot are IQR outliers. An outlier is not automatically an error. It may be a typo, a sensor glitch, or the most interesting observation in your study. The tool flags; you decide.

How to use it

Paste your numbers, separated by commas or new lines.
Set the IQR multiplier (1.5 is standard) and the z-score cutoff (3 is standard).
Click Find outliers and read the fences and the two outlier lists.
Check the plot: red dots are IQR outliers.
Investigate flagged values in your source data before deleting anything.

Worked example

Ten reaction times read 12, 14, 13, 15, 12, 14, 13, 120, 14, and 13 milliseconds. The IQR fences come out around 10 and 18, so 120 is flagged by both rules while every other value passes. Before deleting it, check the lab log: was the trial interrupted, did the equipment misfire, or did the participant genuinely hesitate? If the log shows an interruption, exclusion is defensible and should be reported. If nothing explains it, keep the value and use a robust analysis instead of quietly dropping your most unusual data point.

Honest limits

Both rules assume roughly bell-shaped or at least single-peaked data. On strongly skewed data, the IQR rule flags legitimate high values, and on tiny samples the z-score rule is unstable because one extreme value inflates the standard deviation itself. The plot is a simple strip plot, not a full box plot, and it cannot show subgroups: an outlier in the pooled data may be perfectly normal inside its own group. Always plot and split by group before excluding, and always report how many values you excluded and why.

Who it is for

This tool is for students and researchers doing exploratory data analysis, lab scientists checking a run of measurements, and analysts screening a column before modelling. It is a first-pass screen, not a verdict. Use it to start an investigation into strange values, and let your lab notes, not the red dot alone, make the final call.

Going further

When outliers are genuine values rather than errors, robust statistics are often the honest answer. The median and interquartile range describe the center and spread without letting one extreme value drag them around, unlike the mean and standard deviation. A trimmed mean, which drops a fixed percentage from each tail before averaging, is a middle path. Reporting both the full-sample result and the result after excluding flagged values is even better: if the conclusion survives both, it is sturdy; if it flips, the outlier deserves a paragraph of its own. Whatever you do, pre-specify the rule before you look at the outcome. Deciding to exclude a value because it spoils a significant result is not cleaning, it is cherry-picking, and reviewers who spot a missing data point will ask about it. Keep an exclusion log with the value, the reason, the evidence from source records, and who approved it. That log turns a suspicious deletion into a defensible decision. A related check is to look for outliers in pairs of variables, not just one column at a time. A value can be unremarkable in every single column yet impossible in combination — a six-year-old with a university degree, a resting heart rate of 200 with no recorded exertion. Scatter plots of two variables at once catch those cases, and domain rules catch more: age ranges, physical limits, and logical constraints that no statistical fence knows. Build a short list of such rules for your dataset before analysis, run them alongside the IQR and z-score screens, and treat any violation as a data-quality ticket to resolve with the source, not as a number to delete quietly.

Embed this tool on your website — free

Copy the code below to add this working tool to your own page. A small credit link stays with it — that’s all we ask.