Mean vs Median Tap Intervals

Reviewed concrete example

Two centers answer different questions

Mean vs median tap intervals is not a contest with one permanent winner. The arithmetic mean distributes the total duration across every gap. The median identifies the center of the sorted list. One extreme affects the mean strongly but may leave the median near the typical cluster.

Both require a declared unit and list. Preserve milliseconds before converting to BPM.

Arithmetic mean

Add all intervals and divide by count. For 490, 500, 510 milliseconds, mean is (490 + 500 + 510) / 3 = 500. Converting that duration gives 120 BPM.

The mean uses every value and relates directly to total elapsed interval time. It can be useful when the sequence is coherent, but one missed tap can dominate a short list.

Median

Sort values. For an odd count, take the middle. For an even count, this site averages the two central durations. The original displayed order remains unchanged; sorting is only a calculation copy.

For 480, 500, 520, 1,000, sorted order is the same and median is (500 + 520) / 2 = 510 milliseconds, about 117.65 BPM. Mean is 625 milliseconds, or 96 BPM when converting the mean duration.

The difference flags an asymmetric list. It does not reveal why 1,000 occurred.

Duration centers and BPM centers

Three related calculations can be confused:

• Convert mean duration: 60000 / mean milliseconds.
• Convert median duration: 60000 / median milliseconds.
• Calculate each interval BPM, then average those rates.

Because reciprocal conversion is nonlinear, the first and third generally differ. For 480 and 520 milliseconds, mean duration is 500, converting to 120 BPM. Individual rates are 125 and about 115.38; their mean is about 120.19.

Label the operation rather than calling all of them “average BPM.” BPMCounter.click exposes per-interval mean and median rates while also showing duration centers.

Effect of one long interval

Consider 500, 498, 502, 501, and 1,000 milliseconds. Mean is 600.2; median is 501. The mean-duration rate is about 99.97 BPM; median-duration rate is about 119.76.

Possible explanations include a missed activation, intentional half-rate event, pause, source change, or transcription error. Statistics cannot choose among them. Inspect position and context.

Do not automatically delete the long gap simply because median looks more plausible. If a protocol allows removal, state it and compare raw with trimmed results.

Even counts and software differences

Some systems define median or quartiles differently, especially for even counts. This site averages the two middle values for median. A report should name that when reproducing results elsewhere.

Rounding also matters. Keep unrounded values for arithmetic, then display sensible precision. Do not round each interval BPM before calculating the mean.

Which should you report?

Report both when spread or outliers matter. Add interval count, min–max, and a note about unusual positions. If a workflow requires one center, choose the rule before seeing the desired result and explain why it matches the question.

For steady metronomic input with a symmetric cluster, mean and median may agree closely. For drift, neither single center describes direction; use ordered windows.

Limits

Neither center verifies pulse level, meter, source identity, or performance. A consistent 60-BPM half-time tap can disagree with a 120-BPM grid while having tiny spread. Median robustness does not make it ground truth.

Browser events and human input introduce uncertainty that extra decimals cannot remove.

Visualizing both centers

Place the ordered intervals on a number line. The median sits at the positional center, while the mean acts like a balance point influenced by distance. A far-right long interval pulls the balance point toward it without necessarily moving the middle position. This mental model explains robustness without calling the long value irrelevant.

For a drifting sequence, calculate centers for adjacent windows and retain their order. A full-session mean may land in the middle of an acceleration while corresponding to no actual local passage. A median likewise compresses history. Center plus sequence is the useful combination.

A reporting pair

Write duration center first and rate conversion second: “median interval 501 ms, equivalent rate 119.76 BPM.” Add mean, count, and range. This ordering keeps the observed unit visible and makes reciprocal calculations easier to audit across tools. Keep that label in every shared note.

Frequently asked questions

Paste one coherent list into the interval analyzer, compare duration and rate definitions, locate extremes in original order, and report the chosen convention.