A statistic answers a defined question. Before comparing results, check data, units, order and conventions. Different valid definitions can give different answers for a short list.
Inspect distributions with explicit bins
The histogram uses equal-width intervals. An interior boundary belongs to the bin on its right; the final bin includes the maximum. Changing the bin count can change the appearance of a distribution. A striking grouping does not prove a separate population or a data error.
Respect time order
A trailing moving average uses the entered row order and only complete windows. It does not sort date labels or fill missing values. Smoothing helps inspection but lags changes and does not itself predict the next observation. Blank early averages mean insufficient rows, not zero.
Match definitions before comparing
Sample and population denominators, quartile interpolation and weighted-median ties matter. Regression describes an association under a model, not causation. Record the definition, inspect unusual points rather than automatically deleting them, and avoid comparing statistics from differently filtered data.
Worked example
Scroll the table horizontally on small screens to see every column.
| Period | Value | Mean |
|---|---|---|
| Jan | 10 | Not enough rows |
| Feb | 20 | Not enough rows |
| Mar | 30 | 20 |
| Apr | 40 | 30 |
The first two averages are blank because a complete three-row window is unavailable.
Sources and checks
Examples are illustrative and can be checked with the linked tools. Read each tool’s assumptions before applying its result. How we verify methods.