A5015679618- Advanced Causal Inference Techniques
- Crime Patterns and Interventions
- Statistical Methods and Inference
- Urban, Neighborhood, and Segregation Studies
- Qualitative Comparative Analysis Research
- Policing Practices and Perceptions
- Homelessness and Social Issues
- Advanced Statistical Methods and Models
- Insurance, Mortality, Demography, Risk Management
- Youth Education and Societal Dynamics
- Experimental Behavioral Economics Studies
- Retirement, Disability, and Employment
- Statistical Methods and Bayesian Inference
- Contemporary Sociological Theory and Practice
- Philosophy and History of Science
- Complex Systems and Decision Making
- Economic Theory and Institutions
- Scientific Computing and Data Management
- Education Systems and Policy
- Bayesian Modeling and Causal Inference
- demographic modeling and climate adaptation
- Social and Cultural Dynamics
- American Constitutional Law and Politics
- Employment and Welfare Studies
- Religion and Society Interactions
Harvard University Press
2010-2023
Harvard University
2004-2023
National University of Singapore
2017
Arnold Arboretum
2016
John F. Kennedy University
2014
New York University Press
2014
William James College
1999-2011
State Street (United States)
2010
University of Michigan
2010
Brookings Institution
2010
Most major population surveys used by social scientists are based on complex sampling designs where units have different probabilities of being selected. Although weights must generally be to derive unbiased estimates univariate characteristics, the decision about their use in regression analysis is more complicated. Where solely a function independent variables included model, unweighted OLS preferred because they unbiased, consistent, and smaller standard errors than weighted estimates....
Endogenous selection bias is a central problem for causal inference. Recognizing the problem, however, can be difficult in practice. This article introduces purely graphical way of characterizing endogenous and understanding its consequences (Hernán et al. 2004). We use graphs (direct acyclic graphs, or DAGs) to highlight that stems from conditioning (e.g., controlling, stratifying, selecting) on so-called collider variable, i.e., variable itself caused by two other variables, one (or...
▪ Abstract When experimental designs are infeasible, researchers must resort to the use of observational data from surveys, censuses, and administrative records. Because assignment independent variables is usually nonrandom, challenge estimating causal effects with can be formidable. In this chapter, we review large literature produced primarily by statisticians econometricians in past two decades on estimation data. We first now widely accepted counterfactual framework for modeling effects....
Most discussions of ordinal variables in the sociological literature debate suitability linear regression and structural equation methods when some are ordinal. Largely ignored these for that natural extensions probit logit models dichotomous variables. If discrete realizations unmeasured continuous variables, allow one to include dependent independent into a way (I) explicitly recognizes their ordinality, (2) avoids arbitrary assumptions about scale, (3) allows analysis continuous,...
When observations in social research are selected so that they not independent of the outcome variables a study, sample selection leads to biased inferences about processes. Nonrandom is both source bias empirical and fundamental aspect many This chapter reviews models attempt take account their applications on labor markets, schooling, legal processes, mobility, networks. Variants these apply censored or truncated—whether explicitly incidentally—and include tobit model, standard for...
"We propose to change the default P-value threshold forstatistical significance for claims of new discoveries from 0.05 0.005."
This article proposes a solution to the long-standing methodological problem of incorporating discrete variables inoto causal models social phenomena. Only subset variety ways in which data arise empirical research can be satisfactorily modeled by conventional log-linear or logit approaches. Drawing on insights several literatures, this exposits general approach some all are discretely measured and shows that path analytic methods available permit quantification relationships among with same...
New criteria for indexes of residential segregation are developed. It is argued that a pattern random rather than complete desegregation should be used as baseline measuring segregation. shown any index whose relationship to one independent the proportion black in city, necessarily has dependent with respect other baseline. The dissimilarity adjusted serve measure deviation from Eta-square, which was by Duncan and depend on black, when measured situation its effects concern, but it causes...
This article reviews the causal turn in social sciences and accompanying efforts by criminologists to make policy claims more credible.Although there has been much progress techniques for estimation of effects, we find that link between evidence valid implications remains elusive.Drawing on criminological theory research insights from disciplines such as sociology, economics, statistics, assess principles strategies informing a causally uncertain world.We identify three distinct domains...
The collection of large-scale administrative records in electronic form by many cities provides a new opportunity for the measurement and longitudinal tracking neighborhood characteristics, but one that will require novel methodologies convert such data into research-relevant measures. authors illustrate these challenges developing measures “broken windows” from Boston’s constituent relationship management (CRM) system (aka 311 hotline). A 16-month archive CRM database contains more than...
This article offers a new approach to the identification of age–period–cohort (APC) models that builds on Pearl's work nonparametric causal models, in particular his front-door criterion for effects. The goal is specify mechanisms through which age, period, and cohort variables affect outcome doing so identify model. allows broader set strategies than has typically been considered literature and, many circumstances, goodness fit tests are possible. authors illustrate utility by developing an...
The work on this chapter was done while Joseph Schwartz a graduate student at Harvard University (with support from National Science Foundation Grant SOC76-24394, Harrison White, principal investigator) and Christopher Winship research associate the Institute for Research Poverty. This supported in part by funds granted to Poverty of Wisconsin-Madison Department Health, Education, Welfare pursuant provisions Economic Opportunity Act 1964. authors would like thank Anthony Atkinson,...
A large body of empirical research exists that attempts to determine whether or not police discriminate on the basis race. We investigate methods used typically produce valid inferences. find they often most likely do and results may diverge from reality in either direction, indicating discrimination when it is present alternatively a lack fact present. The reason for this tests make assumptions about behavior are implausible. Because this, simplest forms benchmark outcome should be used,...