Explanation in Causal Inference: Methods of Mediation and Interaction is an introductory text on two widely used methods in statistical analysis: mediation and interaction. The book is both meant to serve as an introduction to these two topics, but also provides considerable mathematical detail in a lengthy appendix. Importantly, the treatment of these two topics is entirely grounded in a counterfactual framework. The counterfactual framework, often referred to as the potential outcomes framework, has been hailed as a revolution in how we think about causality and statistical analysis. I would agree with that sentiment, but the impact of the counterfactual framework is varied. On some topics, the insights have been less revolutionary, but in other areas this framework has I think completely revised how we think. The topics of mediation and interaction analysis are two that I would say have been seriously changed by the counterfactual framework. I think there is already a fairly widespread understanding of how mediation analysis has changed, and this book will only help further spread that awareness. On the topic of interaction analysis, I think there is less appreciation for how the counterfactual framework has changed thinking. This book serves as the remedy.
{"title":"Book Review of Explanation in Causal Inference: Methods of Mediation and Interaction (author: T.J. Vanderweele)","authors":"L. Keele","doi":"10.1353/obs.2016.0007","DOIUrl":"https://doi.org/10.1353/obs.2016.0007","url":null,"abstract":"Explanation in Causal Inference: Methods of Mediation and Interaction is an introductory text on two widely used methods in statistical analysis: mediation and interaction. The book is both meant to serve as an introduction to these two topics, but also provides considerable mathematical detail in a lengthy appendix. Importantly, the treatment of these two topics is entirely grounded in a counterfactual framework. The counterfactual framework, often referred to as the potential outcomes framework, has been hailed as a revolution in how we think about causality and statistical analysis. I would agree with that sentiment, but the impact of the counterfactual framework is varied. On some topics, the insights have been less revolutionary, but in other areas this framework has I think completely revised how we think. The topics of mediation and interaction analysis are two that I would say have been seriously changed by the counterfactual framework. I think there is already a fairly widespread understanding of how mediation analysis has changed, and this book will only help further spread that awareness. On the topic of interaction analysis, I think there is less appreciation for how the counterfactual framework has changed thinking. This book serves as the remedy.","PeriodicalId":74335,"journal":{"name":"Observational studies","volume":"2 1","pages":"1 - 3"},"PeriodicalIF":0.0,"publicationDate":"2021-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1353/obs.2016.0007","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"43857358","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Review of the book “Causal Inference for Statistics, Social, and Biomedical Sciences” by G.W. Imbens and D.B. Rubin","authors":"F. Mealli","doi":"10.1353/obs.2015.0006","DOIUrl":"https://doi.org/10.1353/obs.2015.0006","url":null,"abstract":"","PeriodicalId":74335,"journal":{"name":"Observational studies","volume":"1 1","pages":"291 - 293"},"PeriodicalIF":0.0,"publicationDate":"2021-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1353/obs.2015.0006","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"43872273","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract:This article presents a pre-analysis plan for analyzing the evidential value in a selection of policy research taken from scholarly journals and two research clearinghouses run by the federal government. The analysis will collect p-values from selected studies and estimate the evidential value that they represent using the newly introduced p-curve. This article outlines a precise data collection routine, a set of decision rules for including p-values in the analysis sample, and exact hypothesis tests to be used.
{"title":"Evidence of False Positives in Research Clearinghouses and Influential Journals: An Application of P-Curve to Policy Research","authors":"S. Tanner","doi":"10.1353/obs.2015.0001","DOIUrl":"https://doi.org/10.1353/obs.2015.0001","url":null,"abstract":"Abstract:This article presents a pre-analysis plan for analyzing the evidential value in a selection of policy research taken from scholarly journals and two research clearinghouses run by the federal government. The analysis will collect p-values from selected studies and estimate the evidential value that they represent using the newly introduced p-curve. This article outlines a precise data collection routine, a set of decision rules for including p-values in the analysis sample, and exact hypothesis tests to be used.","PeriodicalId":74335,"journal":{"name":"Observational studies","volume":"1 1","pages":"18 - 29"},"PeriodicalIF":0.0,"publicationDate":"2021-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1353/obs.2015.0001","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"46784855","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract:This article presents a pre-analysis plan for a comparison of methods for the statistical adjustment of observed confounders. In the planned analysis, we intend to replicate five existing studies that used customized form of matching and substantive input from subject matter experts. We will replicate the treatment effect estimates from these studies using machine learning methods that need little user input. In this article, we outline the five studies we will use for replication and discuss the methods we use for replication.
{"title":"Pre-analysis Plan for a Comparison of Matching and Black Box-based Covariate Adjustment","authors":"L. Keele, Dylan S. Small","doi":"10.1353/obs.2018.0017","DOIUrl":"https://doi.org/10.1353/obs.2018.0017","url":null,"abstract":"Abstract:This article presents a pre-analysis plan for a comparison of methods for the statistical adjustment of observed confounders. In the planned analysis, we intend to replicate five existing studies that used customized form of matching and substantive input from subject matter experts. We will replicate the treatment effect estimates from these studies using machine learning methods that need little user input. In this article, we outline the five studies we will use for replication and discuss the methods we use for replication.","PeriodicalId":74335,"journal":{"name":"Observational studies","volume":"4 1","pages":"110 - 97"},"PeriodicalIF":0.0,"publicationDate":"2021-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1353/obs.2018.0017","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"45824703","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract:In the seminal paper from 1960, Thistlethwaite and Campbell (1960) introduce the key ideas underlying regression discontinuity (RD) designs, which, even if initially almost completely ignored, have then acted as a fuse of a blowing number of studies applying and extending RD designs starting from the late nineties. Building on the original idea by Thistlethwaite and Campbell (1960), RD designs have been often described as designs that lead to locally randomized experiments for units with a realized value of a so-called forcing variable falling around a pre-fixed threshold. We embrace this perspective, and in this discussion we offer our view on how the original proposal by Thistlethwaite and Campbell (1960) should be formalized. We introduce an explicit local overlap assumption for a subpopulation around the threshold, for which we re-formulate the Stable Unit Treatment Value Assumption (SUTVA), and provide a formal definition of the hypothetical experiment underlying RD designs, by invoking a local randomization assumption. A distinguishing feature of this approach is that it embeds RD designs in a framework that is fully consistent with the potential outcome approach to causal inference. We discuss how to select suitable subpopulation(s) around the threshold with adjustment for multiple comparisons, and how to draw inference for the causal estimands of interest in this framework. We illustrate our approach in a study concerning the effects of University grants on students’ dropout.
{"title":"Regression Discontinuity Designs as Local Randomized Experiments","authors":"Alessandra Mattei, F. Mealli","doi":"10.1353/obs.2017.0004","DOIUrl":"https://doi.org/10.1353/obs.2017.0004","url":null,"abstract":"Abstract:In the seminal paper from 1960, Thistlethwaite and Campbell (1960) introduce the key ideas underlying regression discontinuity (RD) designs, which, even if initially almost completely ignored, have then acted as a fuse of a blowing number of studies applying and extending RD designs starting from the late nineties. Building on the original idea by Thistlethwaite and Campbell (1960), RD designs have been often described as designs that lead to locally randomized experiments for units with a realized value of a so-called forcing variable falling around a pre-fixed threshold. We embrace this perspective, and in this discussion we offer our view on how the original proposal by Thistlethwaite and Campbell (1960) should be formalized. We introduce an explicit local overlap assumption for a subpopulation around the threshold, for which we re-formulate the Stable Unit Treatment Value Assumption (SUTVA), and provide a formal definition of the hypothetical experiment underlying RD designs, by invoking a local randomization assumption. A distinguishing feature of this approach is that it embeds RD designs in a framework that is fully consistent with the potential outcome approach to causal inference. We discuss how to select suitable subpopulation(s) around the threshold with adjustment for multiple comparisons, and how to draw inference for the causal estimands of interest in this framework. We illustrate our approach in a study concerning the effects of University grants on students’ dropout.","PeriodicalId":74335,"journal":{"name":"Observational studies","volume":"3 1","pages":"156 - 173"},"PeriodicalIF":0.0,"publicationDate":"2021-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1353/obs.2017.0004","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"66461026","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Mariia Samoilenko, L. Blais, B. Cossette, A. Forget, Geneviève Lefebvre
Abstract:PurposeInhaled corticosteroids (ICS) are the first-line controller therapy for asthma. The objective was to assess the impact of different ICS doses during pregnancy on birth weight (BW) using generalized propensity scores (GPS).MethodsA cohort of 7374 pregnancies from 6197 asthmatic women giving birth in Quebec (Canada) in 1998-2008 was constructed. The average treatment effects (ATE) of ICS daily doses (0, >0-125, >125-250, >250 μg/day) during pregnancy on BW were estimated using multilevel GPS and a conventional multivariable approach. Additional analyses were done to evaluate the robustness of the results.ResultsUsing GPS, we found no significant associations between ICS doses and BW (ATE for >0-125 vs 0 μg/day: 27.62 g, 95% confidence interval (CI): -8.68, 64.10; ATE for >125-250 vs 0 μg/day: 17.07 g, 95% CI: -55.85, 92.16; ATE for >250 vs 0 μg/day: -37.83 g, 95% CI: -117.74, 41.53). Similar results were obtained using the multivariable approach.ConclusionsWhile, in our primary analyses, no significant differences were found between the BW of babies exposed to the higher ICS doses, as opposed to no use of ICS, our sensitivity analyses, which adjusted for gestational age in the models, suggest the possibility of a small detrimental effect of the higher ICS doses on BW.
{"title":"Assessing the dose-response relationship between maternal use of inhaled corticosteroids therapy and birth weight: a generalized propensity score approach","authors":"Mariia Samoilenko, L. Blais, B. Cossette, A. Forget, Geneviève Lefebvre","doi":"10.1353/obs.2016.0000","DOIUrl":"https://doi.org/10.1353/obs.2016.0000","url":null,"abstract":"Abstract:PurposeInhaled corticosteroids (ICS) are the first-line controller therapy for asthma. The objective was to assess the impact of different ICS doses during pregnancy on birth weight (BW) using generalized propensity scores (GPS).MethodsA cohort of 7374 pregnancies from 6197 asthmatic women giving birth in Quebec (Canada) in 1998-2008 was constructed. The average treatment effects (ATE) of ICS daily doses (0, >0-125, >125-250, >250 μg/day) during pregnancy on BW were estimated using multilevel GPS and a conventional multivariable approach. Additional analyses were done to evaluate the robustness of the results.ResultsUsing GPS, we found no significant associations between ICS doses and BW (ATE for >0-125 vs 0 μg/day: 27.62 g, 95% confidence interval (CI): -8.68, 64.10; ATE for >125-250 vs 0 μg/day: 17.07 g, 95% CI: -55.85, 92.16; ATE for >250 vs 0 μg/day: -37.83 g, 95% CI: -117.74, 41.53). Similar results were obtained using the multivariable approach.ConclusionsWhile, in our primary analyses, no significant differences were found between the BW of babies exposed to the higher ICS doses, as opposed to no use of ICS, our sensitivity analyses, which adjusted for gestational age in the models, suggest the possibility of a small detrimental effect of the higher ICS doses on BW.","PeriodicalId":74335,"journal":{"name":"Observational studies","volume":"2 1","pages":"181 - 90"},"PeriodicalIF":0.0,"publicationDate":"2021-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1353/obs.2016.0000","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"45170747","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Irwin Bross’s article, “Statistical Criticism,” gives advice that is surprisingly current, given that it appeared in the journal Cancer nearly sixty years ago. Indeed, the only obviously dated aspects of this paper are the use of the generic male pronoun and the sense that it was still an open question whether cigarette smoking caused lung cancer. In his article, Bross acts a critic of criticism, expressing support for the general form but recommending that critics go beyond hit-and-run, dogmatism, speculation, and tunnel vision. This all seems reasonable to me, but I think criticisms can also be taken at face value. If I publish a paper and someone replies with a flawed criticism, I still should be able to respond to its specifics. Indeed, there have been times when my own work has been much improved by criticism that was itself blinkered but which still revealed important and fixable flaws in my published work. I would go further and argue that nearly all criticism has value. Again, I’ll place myself in the position of the researcher whose work is being slammed. Consider the following sorts of statistical criticism, aligned in roughly decreasing order of quality:
{"title":"Learning from and Responding to Statistical Criticism","authors":"A. Gelman","doi":"10.1353/obs.2018.0003","DOIUrl":"https://doi.org/10.1353/obs.2018.0003","url":null,"abstract":"Irwin Bross’s article, “Statistical Criticism,” gives advice that is surprisingly current, given that it appeared in the journal Cancer nearly sixty years ago. Indeed, the only obviously dated aspects of this paper are the use of the generic male pronoun and the sense that it was still an open question whether cigarette smoking caused lung cancer. In his article, Bross acts a critic of criticism, expressing support for the general form but recommending that critics go beyond hit-and-run, dogmatism, speculation, and tunnel vision. This all seems reasonable to me, but I think criticisms can also be taken at face value. If I publish a paper and someone replies with a flawed criticism, I still should be able to respond to its specifics. Indeed, there have been times when my own work has been much improved by criticism that was itself blinkered but which still revealed important and fixable flaws in my published work. I would go further and argue that nearly all criticism has value. Again, I’ll place myself in the position of the researcher whose work is being slammed. Consider the following sorts of statistical criticism, aligned in roughly decreasing order of quality:","PeriodicalId":74335,"journal":{"name":"Observational studies","volume":"4 1","pages":"32 - 33"},"PeriodicalIF":0.0,"publicationDate":"2021-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1353/obs.2018.0003","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"42157120","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract:In an admirable essay, Bross makes many useful observations. The goal, however, should be to take a step beyond statistical criticism, arriving at an objective statement about what the (research design + data) say and fail to say. Often this entails saying a bit less than one might like in exchange for saying something definite and objective.
{"title":"Beyond Statistical Criticism","authors":"P. Rosenbaum, Dylan S. Small","doi":"10.1353/obs.2018.0008","DOIUrl":"https://doi.org/10.1353/obs.2018.0008","url":null,"abstract":"Abstract:In an admirable essay, Bross makes many useful observations. The goal, however, should be to take a step beyond statistical criticism, arriving at an objective statement about what the (research design + data) say and fail to say. Often this entails saying a bit less than one might like in exchange for saying something definite and objective.","PeriodicalId":74335,"journal":{"name":"Observational studies","volume":"4 1","pages":"65 - 70"},"PeriodicalIF":0.0,"publicationDate":"2021-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1353/obs.2018.0008","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"45312970","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract:It has long been observed that defendants who are subject to pre-trial detention are more likely to be convicted than those who are free while they await trial. However, until recently, much of the literature in this area was only correlative and not causal. Using an instrumental variable that represents judge severity, we apply near-far matching–a statistical methodology designed to assess causal relationships using observational data–to a dataset of criminal cases that were handled by the New York Legal Aid Society in 2015. We find a strong causal relationship between bail–an obstacle that prevents many from pre-trial release–and case outcome. Specifically, we find setting bail results in a 34% increase in the likelihood of conviction for the cases in our analysis. To our knowledge, this marks the first time matching methodology from the observational studies tradition has been applied to understand the relationship between money bail and the likelihood of conviction.
摘要:长期以来，人们一直观察到，在候审期间被羁押的被告比被羁押的被告更容易被定罪。然而，直到最近，这一领域的许多文献都只是相关的，而不是因果的。我们使用代表法官严厉程度的工具变量，对纽约法律援助协会(New York Legal Aid Society) 2015年处理的刑事案件数据集应用了近距离匹配(near-far matching)——一种旨在利用观察数据评估因果关系的统计方法。我们发现保释(一个阻碍许多人审前释放的障碍)与案件结果之间存在很强的因果关系。具体来说，我们发现在我们的分析中，保释会使案件定罪的可能性增加34%。据我们所知，这标志着观察研究传统的匹配方法首次被应用于理解保释金与定罪可能性之间的关系。
{"title":"The causal impact of bail on case outcomes for indigent defendants in New York City","authors":"K. Lum, Erwin Ma, M. Baiocchi","doi":"10.1353/obs.2017.0007","DOIUrl":"https://doi.org/10.1353/obs.2017.0007","url":null,"abstract":"Abstract:It has long been observed that defendants who are subject to pre-trial detention are more likely to be convicted than those who are free while they await trial. However, until recently, much of the literature in this area was only correlative and not causal. Using an instrumental variable that represents judge severity, we apply near-far matching–a statistical methodology designed to assess causal relationships using observational data–to a dataset of criminal cases that were handled by the New York Legal Aid Society in 2015. We find a strong causal relationship between bail–an obstacle that prevents many from pre-trial release–and case outcome. Specifically, we find setting bail results in a 34% increase in the likelihood of conviction for the cases in our analysis. To our knowledge, this marks the first time matching methodology from the observational studies tradition has been applied to understand the relationship between money bail and the likelihood of conviction.","PeriodicalId":74335,"journal":{"name":"Observational studies","volume":"3 1","pages":"38 - 64"},"PeriodicalIF":0.0,"publicationDate":"2021-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1353/obs.2017.0007","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"43742046","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：