GlossaryStatistics and interpretation
Trend
A trend describes a longer pattern across comparable observations. Learn how seasonality, selected dates, and reporting changes affect interpretation.
A trend is a longer-term pattern in a series of comparable observations. It can indicate a general upward or downward direction despite shorter fluctuations. The period, measure, and method determine what a claimed trend means; a change between two isolated values does not fully describe the intervening pattern.
Read the whole series
A chart can contain a gradual movement, recurring seasonal variation, unusual peaks, and changes in variability at the same time. Reducing all of those features to “up” or “down” can hide information relevant to residents.
NIST's time-series guidance distinguishes long-term trend from other characteristics and discusses removing a trend before some further analyses. Its run-sequence guidance also emphasizes viewing ordered observations for changes and unusual values. NIST time-series discussion.
Example
Hypothetical example: Comparable annual call totals are 900, 930, 970, 950, and 1,010. The series has an overall upward direction, with a decline in the fourth year. It would be inaccurate to say calls increased every year.
Selecting only years three and four would show a decrease while omitting the longer pattern. Selecting only the first and last year would show the net increase but hide the intervening variation. A useful description names the period and both features.
Common confusion
A trend is not necessarily statistically established merely because a fitted line slopes upward. A line can be fitted to very few points or to observations with incompatible definitions. Statistical tests and models require their own assumptions and explanation.
A trend also does not prove a cause or forecast continuation. A rising call series might reflect demand, broader reporting, new service responsibilities, or a combination. Historical direction alone does not establish what will happen next year.
Keep the measurement comparable
Use the same counting unit, date basis, geography, and categories over time. Mark breaks when a system or classification changes. A sudden step caused by adding an agency should not be presented as if the original coverage experienced the entire increase.
Show enough history to reveal recurring patterns and unusual reference periods. If a rolling average or model is used, explain it and preserve access to the underlying observations when practical. Recent periods can be incomplete, so a downward tail may change after delayed reports arrive. A careful trend statement connects the claimed direction to a defined series and identifies material limits on that interpretation.