Spectral flatness: what it means and how to read it
In the report: spd — The geometric mean divided by the arithmetic mean of the shares x² / Σ x².
How to read it
From 0 to 1: close to 1 — the energy (x²) is spread evenly over the values, like noise; close to 0 — a few values carry most of it. A measure from signal analysis, rarely needed for survey data.
In plain words
A hobbyist recorded how loud a sound was at 8 moments and wanted to know whether it was a hiss or a tone, without listening. He asked for spd, the spectral flatness. It came out 0.9. A number close to 1 means the loudness is spread evenly over all the moments, like the hiss of a radio between stations. A tone with a few very loud moments would give a number close to zero. Now he can tell hiss from a tone by one number. It comes from the study of sound and radio signals, and you will rarely need it for survey answers.
Where you see it
STATS_DESCRIPTIVE— Column-wise descriptive statistics with flexible metric selection and optional column filtering.
Related values
- Slope from the first to the last value — (last value − first value) / (n − 1), in row order; empty for fewer than 2 values, 0 for a constant.
- Sum of squared values — The sum of the squared values: Σ x²; empty for a column without a single number.
- Sum of squared deviations — The sum of the squared deviations from the mean: Σ (x − x̄)²; empty for a column without a single number, 0 for one value.
- Inverse-variance weight — n / variance; empty for a column without spread or with fewer than 2 values.
- z-score of a value — The z-score of the value v: (v − mean) / std; 0 for a column without spread; empty without v.
Get it in your sheet
One formula in a cell gives the whole report, Spectral flatness included. ALS Stats is a free add-on for Google Sheets™.
Install ALS Stats from Google Workspace Marketplace (opens in a new tab)