Appendix B — Glossary

Autoregressive effect
The effect of a variable at the previous occasion on the same variable now; an index of inertia or carry-over.
Between-person effect
An association between people’s average levels of two variables.
Compliance
The proportion of scheduled prompts that were answered; also called response rate.
Contextual effect
The difference between the between-person and within-person effects of the same predictor.
Cross-level interaction
A person-level variable moderating a within-person effect.
Centering within cluster
Subtracting each person’s own mean from their scores, leaving only within-person variation. Also called person-mean centering.
DSEM
Dynamic structural equation modelling: Bayesian multilevel time-series models with latent centering, random dynamic parameters and latent variables.
Ergodicity
The condition under which group-level (between-person) and individual-level (within-person) structures are equivalent.
Event-contingent sampling
Reports initiated by the participant when a predefined event occurs.
ICC
Intraclass correlation: the proportion of total variance due to stable differences between people.
Innovation
The unpredictable part of a variable at each occasion in a time-series model; its variance indexes moment-to-moment unpredictability.
JITAI
Just-in-time adaptive intervention: support delivered at moments identified from momentary data.
Latent centering
Centering on an estimated person mean that accounts for sampling error, rather than the observed mean.
Measurement burst
A design with repeated intensive periods separated by longer gaps.
Micro-randomised trial
A design that randomises intervention delivery at many decision points within each person.
MSSD
Mean squared successive difference: average squared change between consecutive observations; an index of instability.
Nickell’s bias
Downward bias in autoregressive effects when the lagged outcome is centred on the observed person mean.
Random slope
A person-specific deviation from the average effect of a level-1 predictor.
Signal-contingent sampling
Time-based sampling in which the device prompts the participant, usually at random or semi-random times.
Stratified random sampling
One prompt at a random time within each of several equal blocks of the day.
VAR
Vector autoregression: a time-series model in which each variable is predicted by lagged values of all variables.
Within-person effect
The association between a person’s deviations from their own averages on two variables.