Two-stage model for time-varying effects of discrete longitudinal covariates with applications in analysis of daily process data

Hanyu Yang, James A. Cranford, Runze Li, Anne Buu

Research output: Contribution to journalArticle

7 Citations (Scopus)

Abstract

This study proposes a generalized time-varying effect model that can be used to characterize a discrete longitudinal covariate process and its time-varying effect on a later outcome that may be discrete. The proposed method can be applied to examine two important research questions for daily process data: measurement reactivity and predictive validity. We demonstrate these applications using health risk behavior data collected from alcoholic couples through an interactive voice response system. The statistical analysis results show that the effect of measurement reactivity may only be evident in the first week of interactive voice response assessment. Moreover, the level of urge to drink before measurement reactivity takes effect may be more predictive of a later depression outcome. Our simulation study shows that the performance of the proposed method improves with larger sample sizes, more time points, and smaller proportions of zeros in the binary longitudinal covariate.

Original languageEnglish (US)
Pages (from-to)571-581
Number of pages11
JournalStatistics in Medicine
Volume34
Issue number4
DOIs
StatePublished - Feb 20 2015

Fingerprint

Two-stage Model
Covariates
Time-varying
Reactivity
Risk-Taking
Sample Size
Depression
Statistical Analysis
Health
Proportion
Simulation Study
Binary
Research
Zero
Demonstrate
Voice

All Science Journal Classification (ASJC) codes

  • Epidemiology
  • Statistics and Probability

Cite this

@article{38980fbd93f0415685939e5f68403a1f,
title = "Two-stage model for time-varying effects of discrete longitudinal covariates with applications in analysis of daily process data",
abstract = "This study proposes a generalized time-varying effect model that can be used to characterize a discrete longitudinal covariate process and its time-varying effect on a later outcome that may be discrete. The proposed method can be applied to examine two important research questions for daily process data: measurement reactivity and predictive validity. We demonstrate these applications using health risk behavior data collected from alcoholic couples through an interactive voice response system. The statistical analysis results show that the effect of measurement reactivity may only be evident in the first week of interactive voice response assessment. Moreover, the level of urge to drink before measurement reactivity takes effect may be more predictive of a later depression outcome. Our simulation study shows that the performance of the proposed method improves with larger sample sizes, more time points, and smaller proportions of zeros in the binary longitudinal covariate.",
author = "Hanyu Yang and Cranford, {James A.} and Runze Li and Anne Buu",
year = "2015",
month = "2",
day = "20",
doi = "10.1002/sim.6368",
language = "English (US)",
volume = "34",
pages = "571--581",
journal = "Statistics in Medicine",
issn = "0277-6715",
publisher = "John Wiley and Sons Ltd",
number = "4",

}

Two-stage model for time-varying effects of discrete longitudinal covariates with applications in analysis of daily process data. / Yang, Hanyu; Cranford, James A.; Li, Runze; Buu, Anne.

In: Statistics in Medicine, Vol. 34, No. 4, 20.02.2015, p. 571-581.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Two-stage model for time-varying effects of discrete longitudinal covariates with applications in analysis of daily process data

AU - Yang, Hanyu

AU - Cranford, James A.

AU - Li, Runze

AU - Buu, Anne

PY - 2015/2/20

Y1 - 2015/2/20

N2 - This study proposes a generalized time-varying effect model that can be used to characterize a discrete longitudinal covariate process and its time-varying effect on a later outcome that may be discrete. The proposed method can be applied to examine two important research questions for daily process data: measurement reactivity and predictive validity. We demonstrate these applications using health risk behavior data collected from alcoholic couples through an interactive voice response system. The statistical analysis results show that the effect of measurement reactivity may only be evident in the first week of interactive voice response assessment. Moreover, the level of urge to drink before measurement reactivity takes effect may be more predictive of a later depression outcome. Our simulation study shows that the performance of the proposed method improves with larger sample sizes, more time points, and smaller proportions of zeros in the binary longitudinal covariate.

AB - This study proposes a generalized time-varying effect model that can be used to characterize a discrete longitudinal covariate process and its time-varying effect on a later outcome that may be discrete. The proposed method can be applied to examine two important research questions for daily process data: measurement reactivity and predictive validity. We demonstrate these applications using health risk behavior data collected from alcoholic couples through an interactive voice response system. The statistical analysis results show that the effect of measurement reactivity may only be evident in the first week of interactive voice response assessment. Moreover, the level of urge to drink before measurement reactivity takes effect may be more predictive of a later depression outcome. Our simulation study shows that the performance of the proposed method improves with larger sample sizes, more time points, and smaller proportions of zeros in the binary longitudinal covariate.

UR - http://www.scopus.com/inward/record.url?scp=84920398155&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84920398155&partnerID=8YFLogxK

U2 - 10.1002/sim.6368

DO - 10.1002/sim.6368

M3 - Article

C2 - 25395042

AN - SCOPUS:84920398155

VL - 34

SP - 571

EP - 581

JO - Statistics in Medicine

JF - Statistics in Medicine

SN - 0277-6715

IS - 4

ER -