Many statistical analyses aim at a causal explanation of the data. When discussing this topic it is important to specify the exact query we want to talk about. A typical causal question can be categorized in two main classes: questions on the causes of observed effects and questions on the effects of observed causes. In this dissertation we consider both EoC and CoE causal queries from a particular perspective that is Mediation. Mediation Analysis aims to disentangle the pathway between exposure and outcome on a direct effect and an indirect effect arising from the chain exposure-mediator-outcome. In the EoC framework, if the goal is to measure the causal relation between two variables when a third is involved and plays the role of mediator, it is essential to explicitly define several assumptions among variables. However if any of these assumptions is not met, estimates of mediating effects may be affected by bias. This phenomenon, known with the name of Birth Weight paradox, has been explained as a consequence of the presence of unmeasured confounding between the mediator and the outcome. In this thesis we discuss these apparent paradoxical results in a real dataset. In addition we suggest useful graphical sensitivity analysis techniques to explain the potential amount of bias capable of producing these paradoxical results. From a CoE perspective, given empirical evidence for the dependence of an outcome variable on an exposure variable, we can typically only provide bounds for the “probability of causation” in the case of an individual who has developed the outcome after being exposed. We show how these bounds can be adapted or improved if further information becomes available. In addition to reviewing existing work on this topic, we provide a new analysis for the case where a mediating variable can be observed. In particular we show how the probability of causation can be bounded in two different cases of partial and complete mediation.

Mediation analysis for different types of Causal questions: Effect of Cause and Cause of Effect

MURTAS, ROSSELLA
2016-03-17

Abstract

Many statistical analyses aim at a causal explanation of the data. When discussing this topic it is important to specify the exact query we want to talk about. A typical causal question can be categorized in two main classes: questions on the causes of observed effects and questions on the effects of observed causes. In this dissertation we consider both EoC and CoE causal queries from a particular perspective that is Mediation. Mediation Analysis aims to disentangle the pathway between exposure and outcome on a direct effect and an indirect effect arising from the chain exposure-mediator-outcome. In the EoC framework, if the goal is to measure the causal relation between two variables when a third is involved and plays the role of mediator, it is essential to explicitly define several assumptions among variables. However if any of these assumptions is not met, estimates of mediating effects may be affected by bias. This phenomenon, known with the name of Birth Weight paradox, has been explained as a consequence of the presence of unmeasured confounding between the mediator and the outcome. In this thesis we discuss these apparent paradoxical results in a real dataset. In addition we suggest useful graphical sensitivity analysis techniques to explain the potential amount of bias capable of producing these paradoxical results. From a CoE perspective, given empirical evidence for the dependence of an outcome variable on an exposure variable, we can typically only provide bounds for the “probability of causation” in the case of an individual who has developed the outcome after being exposed. We show how these bounds can be adapted or improved if further information becomes available. In addition to reviewing existing work on this topic, we provide a new analysis for the case where a mediating variable can be observed. In particular we show how the probability of causation can be bounded in two different cases of partial and complete mediation.
17-mar-2016
causal interference
cause of effect
effect of cause
mediation
probability of causation
File in questo prodotto:
File Dimensione Formato  
PhD_Thesis_MurtasR.pdf

accesso aperto

Tipologia: Tesi di dottorato
Dimensione 1.11 MB
Formato Adobe PDF
1.11 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11584/266647
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact