Browsing by Subject "structural equation modeling"

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    The Effects of Caregiving Resources on Perceived Health among Caregivers
    (Oxford University Press, 2016-08) Hong, Michin; Harrington, Donna; Labor Studies, School of Social Work
    This study examined how various types of resources influence perceived health of caregivers. Guided by the conservation of resources theory, a caregiver health model was built and tested using structural equation modeling. The caregiver health model consisted of caregiving situations (functional limitations and cognitive impairments of older adults and caregiving time), resources (financial resources, mastery, social support, family harmony, and service utilization), caregiver burden, and perceived health of caregivers. The sample included 1,837 unpaid informal caregivers drawn from the 2004 National Long-Term Caregiver Survey. The model fit indices indicated that the first structural model did not fit well; however, the revised model yielded an excellent model fit. More stressful caregiving situations were associated with fewer resources and higher burden, whereas greater resources were associated with lower burden and better perceived health of caregivers. The results suggest explicit implications for social work research and practice on how to protect the health of caregivers.
  • Thumbnail Image
    Item
    Granger mediation analysis of multiple time series with an application to functional magnetic resonance imaging
    (Wiley, 2019-09) Zhao, Yi; Luo, Xi; Biostatistics, School of Public Health
    This paper presents Granger mediation analysis, a new framework for causal mediation analysis of multiple time series. This framework is motivated by a functional magnetic resonance imaging (fMRI) experiment where we are interested in estimating the mediation effects between a randomized stimulus time series and brain activity time series from two brain regions. The independent observation assumption is thus unrealistic for this type of time‐series data. To address this challenge, our framework integrates two types of models: causal mediation analysis across the mediation variables, and vector autoregressive (VAR) models across the temporal observations. We use “Granger” to refer to VAR correlations modeled in this paper. We further extend this framework to handle multilevel data, in order to model individual variability and correlated errors between the mediator and the outcome variables. Using Rubin's potential outcome framework, we show that the causal mediation effects are identifiable under our time‐series model. We further develop computationally efficient algorithms to maximize our likelihood‐based estimation criteria. Simulation studies show that our method reduces the estimation bias and improves statistical power, compared with existing approaches. On a real fMRI data set, our approach quantifies the causal effects through a brain pathway, while capturing the dynamic dependence between two brain regions.
  • Thumbnail Image
    Item
    Longitudinal relationships between fatigue and depression in cancer patients with depression and/or pain
    (American Psychological Association, 2013-12) Brown, Linda F.; Rand, Kevin L.; Bigatti, Silvia M.; Stewart, Jesse C.; Theobald, Dale E.; Wu, Jingwei; Kroenke, Kurt; Social and Behavioral Sciences, Richard M. Fairbanks School of Public Health
    OBJECTIVE: Fatigue is one of the most common and debilitating symptoms reported by cancer patients, yet relatively little is understood about its etiology. Recently, as researchers have begun to focus attention on cancer-related fatigue (CRF), depression has emerged as its strongest correlate. Few longitudinal studies, however, have examined directionality of the relationship between the two symptoms. Our aim was to evaluate the directionality of the association between depression and CRF. METHOD: The study used a single-group cohort design of longitudinal data (N = 329) from a randomized controlled trial of an intervention for pain and depression in a heterogeneous sample of cancer patients. Participants met criteria for clinically significant pain and/or depression. Our hypothesis that depression would predict change in fatigue over 3 months was tested using latent variable cross-lagged panel analysis. RESULTS: Depressive symptoms and fatigue were strongly correlated in the sample (baseline correlation of latent variables = 0.71). Although the model showed good fit to the data, χ(2) (66, N = 329) = 88.16, p = .04, SRMR = 0.030, RMSEA = 0.032, and CFI = 1.00, neither structural path linking depression and fatigue was significant, suggesting neither symptom preceded and predicted the other. CONCLUSIONS: Our findings did not support hypotheses regarding the directionality of the relationship between depressive symptoms and fatigue. The clinical implication is that depression-specific treatments may not be sufficient to treat CRF and that instead, interventions specifically targeting fatigue are needed.