A daily count falls sharply, but the line in the news chart barely moves. That is not necessarily an error. The line may show a rolling average, meaning each plotted point summarises several days rather than displaying only the newest observation.
Seven days inside one point
For a simple trailing seven-day average, add the count for a day and the six preceding days, then divide by seven. On the next day, add the new count and remove the oldest one. The window moves forward by one day.
Consider an invented series in which every day has a count of ten. The average is ten. Now one day records eighty: the seven-day total becomes 140, and the average becomes twenty. If the next day returns to ten while another ten leaves the window, the average remains twenty.
The unusual day is still inside the calculation. It will keep influencing the line until it leaves the window. A high average therefore does not establish that the most recent raw count was high too.
Read the chart’s small print
Check whether the average is trailing or centred, how many days it covers and how missing values are handled. A centred average assigns surrounding observations to a middle date; it should not be interpreted exactly like the trailing example above.
Also distinguish the date an event occurred from the date it was reported. Smoothing a series does not correct every reporting delay, revised count or change in definition. A neat line can still summarise uneven data.
When writing a note about the chart, name the measure: “The seven-day average rose” is more precise than “Today’s count rose” if only the average is shown. If raw observations are available, look at them alongside the smoothed line.
The average helps reveal a broader pattern by reducing day-to-day variation. Its usefulness depends on remembering what was averaged. Before interpreting a sudden turn, inspect the dates entering and leaving the calculation and the source’s explanation of the data.