Structuring, Checking and Preparing ESM Data

Description

Put experience-sampling data in long format, run design and plausibility checks, apply response-delay and compliance rules, then build centred and lagged variables and plot individual series.

Intensive longitudinal data need preparation beyond survey cleaning. Data go in long format, one row per prompt, with subject, day, beep, scheduled time and response time. Checks: design and time variables never missing; subject-level variables constant within a person; no out-of-range values; no duplicated beeps; beep times inside their scheduled blocks; branching items consistent. Rules then handle response delay (for example a 15-minute limit), implausibly short completion times as careless responding, and a minimum response count per person, each tested with sensitivity analyses because exclusion thresholds are arbitrary. Scale scores use a missing-item rule; person means and person-mean centred values are computed; lagged variables never cross persons or nights. Individual time series and per-person slopes are plotted before modelling. Collapsing to subject level gives descriptive statistics.

Attributes

Loss type
Unmigrated
Effort
Days

Steps

  1. Arrange the data in long format with design and timing variables (subject, day, beep, scheduled and actual times).
  2. Check design variables for missing values, subject variables for constancy, all variables for range and duplicates, beep times against blocks, and branching consistency.
  3. Inspect response delay and completion time; exclude responses beyond the delay limit (for example 15 minutes) or implausibly fast.
  4. Compute compliance per person; apply any minimum and test it with sensitivity analyses.
  5. Compute scale scores with a stated missing-item rule.
  6. Compute person means, person-mean centred values and lagged variables, setting the first beep of each day and person to missing.
  7. Plot individual time series and per-person predictor-outcome slopes before modelling.

Sources

  • The Open Handbook of Experience Sampling Methodology (Myin-Germeys & Kuppens, 2022)

Also known as

  • ESM data preparation
  • Intensive longitudinal data cleaning

Categories

  • Analytics & Behaviour
  • Statistics & Modelling