causal diagrams
Recently Published Documents


TOTAL DOCUMENTS

122
(FIVE YEARS 33)

H-INDEX

17
(FIVE YEARS 3)

Author(s):  
Dean Knox ◽  
Christopher Lucas ◽  
Wendy K. Tam Cho

Social scientists commonly use computational models to estimate proxies of unobserved concepts, then incorporate these proxies into subsequent tests of their theories. The consequences of this practice, which occurs in over two-thirds of recent computational work in political science, are underappreciated. Imperfect proxies can reflect noise and contamination from other concepts, producing biased point estimates and standard errors. We demonstrate how analysts can use causal diagrams to articulate theoretical concepts and their relationships to estimated proxies, then apply straightforward rules to assess which conclusions are rigorously supportable. We formalize and extend common heuristics for “signing the bias”—a technique for reasoning about unobserved confounding—to scenarios with imperfect proxies. Using these tools, we demonstrate how, in often-encountered research settings, proxy-based analyses allow for valid tests for the existence and direction of theorized effects. We conclude with best-practice recommendations for the rapidly growing literature using learned proxies to test causal theories. Expected final online publication date for the Annual Review of Political Science, Volume 25 is May 2022. Please see http://www.annualreviews.org/page/journal/pubdates for revised estimates.


2021 ◽  
Author(s):  
Fuzhong Xue ◽  
Xiaoru Sun ◽  
Hongkai Li ◽  
Yuanyuan Yu ◽  
Zhongshang Yuan ◽  
...  

Genome-wide association study (GWAS) is fundamentally designed to detect disease-causing genes. To reduce spurious associations or improve statistical power, about 80% of GWASs arbitrarily adjusted for demographic and clinical covariates. However, adjustment strategies in GWASs have not achieved consistent conclusions. Given the initial aim of GWAS that is to identify the causal association between a specific causal single-nucleotide polymorphism (SNP) and disease trait, we summarized all complex relationships of the target SNP, covariate and disease trait into 15 causal diagrams according to various roles of the covariate. Following each causal diagram, we conducted a series of theoretical justifications and statistical simulations. Our results demonstrate that it is unadvisable to adjust for any demographic or clinical covariates. We illustrate our point by applying GWASs for body mass index (BMI) and breast cancer, including adjusting and non-adjusting for age and smoking status. Genetic effects and P values might vary across different strategies. Instead, adjustments for SNPs (G') should be strongly recommended when G' are in linkage disequilibrium with the target SNP, and correlated with disease trait conditional on the target SNP. Specifically, adjustment for such G' can block all the confounding paths between the target SNP and disease trait, and avoid over-adjusting for colliders or intermediaries.


2021 ◽  
pp. 101-114
Author(s):  
Nick Huntington-Klein
Keyword(s):  

2021 ◽  
pp. 87-100
Author(s):  
Nick Huntington-Klein
Keyword(s):  

2021 ◽  
Vol 50 (Supplement_1) ◽  
Author(s):  
Jennifer Dunne ◽  
Gizachew Assefa Tessema ◽  
Milica Ognjenovic ◽  
Gavin Pereira

Abstract Background Establishing causal effects in reproductive and perinatal epidemiology is challenging due to the many selection and attrition processes from preconception to the postnatal period. Further challenging, is the potential for the misclassification of exposures, outcomes and confounders, contributing to measurement error. The application of simulation enables the illustration and quantification of the magnitude of various types of bias commonly found in observational studies. Methods A systematic search was conducted in PubMed, Medline, Embase, CINAHL and Scopus in August 2020. A gray literature search of Google and Google Scholar, followed by a search of the reference lists of included studies, was undertaken. Results Thirty-nine studies, covering information (n = 14), selection (n = 14), confounding (n = 9), protection (n = 1), and attenuation bias (n = 1) were identified. The methods of simulating data and reporting of results varied, with more recent studies including causal diagrams. Few studies included code for replication. Although there has been an increasing application of simulation in reproductive and perinatal epidemiology since 2015, overall this remains an underexplored area. Conclusions The studies demonstrated effectiveness in the quantification of multiple types of bias using simulation. The limited use implies that further effort is required to increase knowledge of the application of simulation, which will thereby improve causal interpretation in reproductive and perinatal studies. Key messages Practical guidance for researchers is required in the development, analysis and reporting of simulation methods for the quantification of bias.


2021 ◽  
Vol 50 (Supplement_1) ◽  
Author(s):  
Tim Watkins

Abstract Focus of Presentation Most researchers do not use causal diagrams, in this case meaning directed acyclic graphs (DAGs), despite being widely recommended in epidemiology. They can help to identify the biases that might lead to faulty conclusions or suggest variables for which data should be collected and included in a model. Seeking to understand this reluctance and develop alternative strategies that might increase the use of causal diagrams, we searched the cognitive science literature for potential reasons and suggestions. Findings Insights from cognitive psychology led to a better understanding of the barriers that might underlie the reluctance to use causal diagrams. This includes our built-in desire for cognitive ease and suggests that strategies which lower the effort required to create a diagram may help. We explain these findings using example projects from neuropsychiatry big data research and describe how an online resource we have created has helped. Conclusions/Implications A causal diagram website has been created that aims to lower the effort needed to create a diagram for a study. It contains tutorials and a terminology guide, as well as links to other tutorials; a guide to software and other resources that might be used; and a searchable database of example causal diagrams with links to published articles that include them. Key messages A website has been developed to help overcome barriers to the use of causal diagrams. With contributions welcome.


Author(s):  
Thibaut Pressat Laffouilhère ◽  
Julien Grosjean ◽  
Jacques Bénichou ◽  
Stefan J. Darmoni ◽  
Lina F. Soualmia

In the context of causal inference, biostatisticians use causal diagrams to select covariates in order to build multivariate models. These diagrams represent datasets variables and their relations but have some limitations (representing interactions, bidirectional causal relations). The MetBrAYN project aims at building an ontological-based process to tackle these issues. The knowledge acquired by the biostatistician during a methodological consultation for a research question will be represented in a general ontology. In order to aggregate various forms of knowledge the ontology will act as a wrapper. Ontology-based causal diagrams will be semi-automatically built. Founded on inference rules, the global system will help biostatisticians to curate it and to visualize recommended covariates for their research question.


Sign in / Sign up

Export Citation Format

Share Document