Sampling Schemes and the Recovery of Within-Person Dynamics in Intensive Longitudinal Data: A Two-Timescale Continuous-Time Approach

Intensive longitudinal designs space observations unequally. Fixed schedules leave overnight gaps, signal-contingent protocols randomize prompt times, and measurement-burst designs separate dense bursts by weeks, yet standard discrete-time vector autoregressive analyses treat all intervals as equal. This article derives what such analyses estimate under design-generated interval mixtures. The probability limit implies design-specific attenuation of cross-lagged effects, and occasion-level measurement error combined with contemporaneous correlation produces spurious cross-lagged effects even under stationarity. Unmodeled slow change inflates false-positive rates further, to .31 in single series and .94 under two-stage pooling with burst sampling. A two-timescale continuous-time state-space model, separating fast latent dynamics, a correlated slow trend, and occasion noise, maintained nominal Type I error and drift coverage across equidistant, diurnal, random, and burst schemes in 32,000 simulated series. Reanalysis of a 239-day experience sampling study shows model choice reversing substantive conclusions. All code and results are openly available.

Authors

Publication Details

Journal
Figshare
Published
2026-11-19
DOI
https://doi.org/10.6084/m9.figshare.35068047
Primary Topic
Mental Health Research Topics
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Sampling Schemes and the Recovery of Within-Person Dynamics in Intensive Longitudinal Data: A Two-Timescale Continuous-Time Approach

Hsiu-Ting Yu
Figshare
Mental Health Research Topics
article

Sampling Schemes and the Recovery of Within-Person Dynamics in Intensive Longitudinal Data: A Two-Timescale Continuous-Time Approach

Hsiu-Ting Yu
article en

Abstract

Intensive longitudinal designs space observations unequally. Fixed schedules leave overnight gaps, signal-contingent protocols randomize prompt times, and measurement-burst designs separate dense bursts by weeks, yet standard discrete-time vector autoregressive analyses treat all intervals as equal. This article derives what such analyses estimate under design-generated interval mixtures. The probability limit implies design-specific attenuation of cross-lagged effects, and occasion-level measurement error combined with contemporaneous correlation produces spurious cross-lagged effects even under stationarity. Unmodeled slow change inflates false-positive rates further, to .31 in single series and .94 under two-stage pooling with burst sampling. A two-timescale continuous-time state-space model, separating fast latent dynamics, a correlated slow trend, and occasion noise, maintained nominal Type I error and drift coverage across equidistant, diurnal, random, and burst schemes in 32,000 simulated series. Reanalysis of a 239-day experience sampling study shows model choice reversing substantive conclusions. All code and results are openly available.

Figshare
Mental Health Research Topics
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon