Psychology Reader

No Time Like the Present: Discovering the Hidden Dynamics in Intensive Longitudinal Data

Ellen L. Hamaker and Marieke Wichers · 2017 · Current Directions in Psychological Science

Role
Foundational argument · Methodological contribution
Difficulty
Moderate
Full work length
6 pages
Type
Journal article

Access

Psychology Reader links to legitimate sources and does not host article PDFs.

Reading focus

Syllabus: How Can We Study Change Within People?

Track the distinction between an individual time series and a model that allows dynamics to vary across people.

Why you’re reading it. Develop a vocabulary for persistence, change and heterogeneity before interpreting a temporal model.

A question to consider. For sleep and stress, write one between-person question and one within-person question. What observations would each require?

Why this reading

Repeated observations can reveal how an experience persists or changes within a person. Hamaker and Wichers introduce ways to study those dynamics while allowing people to differ. Their short review connects the question being asked to the observations and model it requires.

Before you read

A time series records observations in temporal order. Autocorrelation is association with earlier values of the same variable; a cross-lagged association connects different variables across occasions. Multilevel models can represent both processes within people and differences between them.

Read for

  1. Distinguish a person’s usual level from fluctuations around that level and from the persistence of those fluctuations.
  2. Follow how sampling intervals, irregular timing and changes in the process challenge interpretation.

Consider afterwards

  1. Two people have the same average stress rating. What temporal patterns could make their experiences quite different, and what observations would reveal those patterns?

Editorial context

This is a methodological perspective illustrated with clinical research, not evidence that an individual time-series model identifies a treatment target. Temporal prediction needs additional assumptions to support a causal claim. Person-specific estimates can also be uncertain; collecting repeated observations does not eliminate measurement error or selection.