Stationarity versus non-stationarity

Stationarity versus non-stationarity

This is the landing page for topics related to applying stationary and non-stationary models to your intensive longitudinal data.

Considering whether a process is stationary or non-stationary is essential because it affects how patterns like means, variances, and autoregressions behave over time. Stationary models assume these properties are constant, which simplifies modeling but may misrepresent dynamic systems where change occurs. Using non-stationary models when appropriate allows for a more accurate capture of evolving processes, such as trends, cycles, or shifts in behavior.

Below we have specified multiple articles where you can read more about stationary and non-stationary modeling.

Think more about stationary versus non-stationary models

Types of stationarity

The literature distinguishes between different types of stationarity.

[Mean versus trend stationarity]
[Unit root]

Testing for stationarity

There are different tests available to test for the different types of stationarity.

[Stationarity tests]
[Auto-ARIMA]

Accounting for non-stationarity

There are some techniques that can be used to account for non-stationarity in your process.

[Differencing]
[ARIMA]
[(G)ARCH]