Home > Research . Search . Country . Browse . Small Grants

PSC In The News

RSS Feed icon

Kusunoki, Hall, and Barber find obese teen girls less likely to use birth control

Prescott finds reported sex offenses lower in neighborhoods with resident sex offenders

Geronimus says poor Detroiters face greater health risks given adverse social conditions


Bob Willis awarded 2015 Jacob Mincer Award for Lifetime Contributions to the Field of Labor Economics

David Lam is new director of Institute for Social Research

Elizabeth Bruch wins Robert Merton Prize for paper in analytic sociology

Elizabeth Bruch wins ASA award for paper in mathematical sociology

Next Brown Bag

PSC Brown Bags will be back fall 2015

Michael R. Elliott photo

Causal Methods for Mediation and Interaction

a PSC Research Project [ARCHIVE DISPLAY]

Investigator:   Michael R. Elliott

To answer causal hypotheses arising from several different randomized behavioral intervention trials for outcomes such as depression and suicide ideation, causal methods are proposed for two goals: 1) to aid the explanation of the mechanism of action of these interventions through intermediate factors such as process measures (i.e., mediation with possible interactions); and 2) to help identify sub-groups of patients based on post-randomization factors across which baseline intervention effects vary. For both aims, we relax a strong assumption of current mediation/interaction methods that assumes no unmeasured confounding for the mediating or interaction factors (sequential ignorability). We make other assumptions involving more parametric models with sensitivity analyses based on different modeling approaches.

For the mechanism of action goal, we propose extensions of structural mean models (SMM) to estimating prescriptive and natural direct effects and natural indirect effects. Natural effects are appropriate for the effectiveness research represented by the studies of interest as they provide a theoretical basis for indirect effects and accommodate interactions between baseline interventions and post-baseline behavioral and process factors. The proposed SMM’s are estimated with extensions of the weighted G-estimation approach. We also will focus on developing optimally efficient weights that improve precision without assuming sequential ignorability. For the post-randomization stratification goal, latent principal strata or sub-groups are identified under the principal stratification (PS) approach, using intermediate factor and randomized baseline intervention information. The baseline intervention effect in strata corresponding to constant intermediate factor levels regardless of randomization arm represent prescribed direct effects. In addition to optimal weights for the SMM approach, we extend both the SMM and PS approaches to multiple nested intermediate factors (e.g., physician prescription and patient medication behavior), longitudinal outcomes, binary outcomes, and cases where intermediate outcomes influence intermediate adherence behavior. The sensitivity of the SMM, PS, and standard mediation / interaction approaches to their respective assumptions will be investigated. The above methods and standard mediation/interaction procedures will be evaluated and compared with simulations and analyses of data from the studies of interest.

Funding Period: 09/10/2007 to 05/31/2012

This PSC Archive record is displayed for historical reference.

Search . Browse