Results 11  20
of
180
Proceeding from observed correlation to causal inference: The use of natural experiments
 Perspectives on Psychological Science
, 2007
"... ABSTRACT—This article notes five reasons why a correlation between a risk (or protective) factor and some specified outcome might not reflect environmental causation. In keeping with numerous other writers, it is noted that a causal effect is usually composed of a constellation of components actin ..."
Abstract

Cited by 32 (2 self)
 Add to MetaCart
ABSTRACT—This article notes five reasons why a correlation between a risk (or protective) factor and some specified outcome might not reflect environmental causation. In keeping with numerous other writers, it is noted that a causal effect is usually composed of a constellation of components acting in concert. The study of causation, therefore, will necessarily be informative on only one or more subsets of such components. There is no such thing as a single basic necessary and sufficient cause. Attention is drawn to the need (albeit unobservable) to consider the counterfactual (i.e., what would have happened if the individual had not had the supposed risk experience). Fifteen possible types of natural experiments that may be used to test causal inferences with respect to naturally occurring prior causes (rather than planned interven
Estimation in the Regression Discontinuity Model
, 2003
"... The regression discontinuity model has recently become a commonly applied framework for empirical work in economics. Hahn, Todd, and Van der Klaauw (2001) provide a formal development of the identification of a treatment effect in this framework and also note the potential bias problems in its estim ..."
Abstract

Cited by 30 (0 self)
 Add to MetaCart
The regression discontinuity model has recently become a commonly applied framework for empirical work in economics. Hahn, Todd, and Van der Klaauw (2001) provide a formal development of the identification of a treatment effect in this framework and also note the potential bias problems in its estimation. This bias difficulty is the result of a particular feature of the regression discontinuity treatment effect estimation problem that distinguishes it from typical semiparametric estimation problems where smoothness is lacking. Here, the discontinuity is not simply an obstacle to overcome in estimation; instead, the size of discontinuity is itself the object of estimation interest. In this paper, I derive the optimal rate of convergence for estimation of the regression discontinuity treatment effect. The optimal rate suggests that with appropriate choice of estimator the bias difficulties are no worse than would be found in the usual nonparametric conditional mean estimation problem (at an interior point of the covariate support). Two estimators are proposed that attain the optimal rate under varying conditions. One new estimator is based on Robinson’s (1988) partially linear estimator. The other estimator uses local polynomial estimation and is optimal under a broader set of conditions.
Program Evaluation and Research Designs
 of Handbook of Labor Economics, Elsevier, chapter 5
, 2011
"... This chapter provides a selective review of some contemporary approaches to program evaluation. One motivation for our review is the recent emergence and increasing use of a particular kind of “program ” in applied microeconomic research, the socalled Regression Discontinuity (RD) Design of Thistle ..."
Abstract

Cited by 26 (0 self)
 Add to MetaCart
This chapter provides a selective review of some contemporary approaches to program evaluation. One motivation for our review is the recent emergence and increasing use of a particular kind of “program ” in applied microeconomic research, the socalled Regression Discontinuity (RD) Design of Thistlethwaite and Campbell (1960). We organize our discussion of these various research designs by how they secure internal validity: in this view, the RD design can been seen as a close “cousin ” of the randomized experiment. An important distinction which emerges from our discussion of “heterogeneous treatment effects ” is between ex post (descriptive) and ex ante (predictive) evaluations; these two types of evaluations have distinct, but complementary goals. A second important distinction we make is between statistical statements that are descriptions of our knowledge of the program assignment process and statistical statements that are structural assumptions about individual behavior. Using these distinctions, we examine some commonly employed evaluation strategies, and assess them with a common set of criteria for “internal validity”, the foremost goal of an ex post evaluation. In some cases, we also provide some concrete illustrations of how internally valid causal estimates can be supplemented with specific structural assumptions to address “external validity”: the estimate from an internally valid "experimental " estimate can be viewed as a “leading term ” in an extrapolation for a parameter of interest in an ex ante evaluation.
Nonlinear policy rules and the identification and estimation of causal effects in a generalized regression kink design. National Bureau of Economic Research, Working Paper No
, 2012
"... Uppsala, Wharton and Zürich. Andrea Weber gratefully acknowledges research funding from the Austrian ..."
Abstract

Cited by 25 (1 self)
 Add to MetaCart
Uppsala, Wharton and Zürich. Andrea Weber gratefully acknowledges research funding from the Austrian
Robust Nonparametric Confidence Intervals for RegressionDiscontinuity Designs
, 2013
"... In the regressiondiscontinuity (RD) design, units are assigned to treatment based on whether their value of an observed covariate exceeds a known cutoff. In this design, local polynomial estimators are now routinely employed to construct confidence intervals for treatment effects. The performance o ..."
Abstract

Cited by 24 (2 self)
 Add to MetaCart
In the regressiondiscontinuity (RD) design, units are assigned to treatment based on whether their value of an observed covariate exceeds a known cutoff. In this design, local polynomial estimators are now routinely employed to construct confidence intervals for treatment effects. The performance of these confidence intervals in applications, however, may be seriously hampered by their sensitivity to the specific bandwidth employed. Available bandwidth selectors typically yield a “large” bandwidth, leading to datadriven confidence intervals that may be severely biased, with empirical coverage well below their nominal target. We propose new, more robust, theorybased con…dence interval estimators for average treatment e¤ects in sharp RD, kink RD, fuzzy RD and fuzzy kink RD designs. Our proposed confidence intervals rely on a recentered RD estimator together with a novel standarderror estimator. For practical implementation, we propose a consistent standarderror estimator that does not require an additional bandwidth choice, as well as valid bandwidth choices compatible with our underlying largesample theory. In a simulation study, we find that our novel datadriven confidence intervals exhibit closetocorrect empirical coverage and good empirical interval length on average, remarkably improving upon the alternatives available in the literature. We illustrate the performance of our proposed methods with household data from Progresa/Oportunidades, a conditional cash transfer program in Mexico. All the results in this paper are readily available in STATA using our companion package (rdrobust) described in Calonico, Cattaneo, and Titiunik (2013).
Quantile Treatment Effects in the Regression Discontinuity Design: Process Results and Gini Coefficient
, 2010
"... ..."
Programme Evaluation with Multiple Treatments
 Journal of Economic Surveys
, 2004
"... Abstract. This paper reviews the main identification and estimation strategies for microeconometric policy evaluation. Particular emphasis is laid on evaluating policies consisting of multiple programmes, which is of high relevance in practice. For example, active labour market policies may consist ..."
Abstract

Cited by 20 (0 self)
 Add to MetaCart
(Show Context)
Abstract. This paper reviews the main identification and estimation strategies for microeconometric policy evaluation. Particular emphasis is laid on evaluating policies consisting of multiple programmes, which is of high relevance in practice. For example, active labour market policies may consist of different training programmes, employment programmes and wage subsidies. Similarly, sickness rehabilitation policies often offer different vocational as well as nonvocational rehabilitation measures. First, the main identification strategies (controlforconfoundingvariables, differenceindifference, instrumentalvariable, and regressiondiscontinuity identification) are discussed in the multipleprogramme setting. Thereafter, the different nonparametric matching and weighting estimators of the average treatment effects and their properties are examined.
Building program evaluation into the design of public research support programs, Technology Policy and Innovation
 Technology Policy for Energy and the Environment,” Papers from the NBER Meeting on Innovation Policy and the
, 2002
"... It is widely accepted that, in the absence of policy intervention, the social rate of return to R&D expenditure exceeds the private rate, leading to a socially suboptimal rate of investment in R&D (Guellec and van Pottelsberghe, 2000). Indeed, empirical evidence suggests that, even ..."
Abstract

Cited by 20 (0 self)
 Add to MetaCart
It is widely accepted that, in the absence of policy intervention, the social rate of return to R&D expenditure exceeds the private rate, leading to a socially suboptimal rate of investment in R&D (Guellec and van Pottelsberghe, 2000). Indeed, empirical evidence suggests that, even
Going to a Better School: Effects and Behavioral Responses: Dataset.” American Economic Review.
, 2013
"... This paper applies a regression discontinuity design to the Whether students would benefit from attending higherachievement schools is an important question in education. Clear evidence on this issue is scarce, in large part because students are not randomly allocated to schools. Nevertheless, as ..."
Abstract

Cited by 20 (0 self)
 Add to MetaCart
(Show Context)
This paper applies a regression discontinuity design to the Whether students would benefit from attending higherachievement schools is an important question in education. Clear evidence on this issue is scarce, in large part because students are not randomly allocated to schools. Nevertheless, as discussed below, several papers provide credible estimates of the effect of having access to a better school. Such estimates do not provide a complete road map for policy, however, as they may reflect but not reveal behavioral responses that amplify or reduce the impact of educational quality. For instance, parents might react to their children going to a better school by lowering their own effort. There might also be reactions on the part of students; for example, an individual who makes it into a better school might feel inferior or be stigmatized. 1 Importantly, these responses might change over time, and may thus influence estimates differently depending on when outcome data are collected. Additionally, some of these responseswhich we will refer to as equilibrium effectsmay only emerge once interventions are taken to scale and sustained for a period of time. 2 To illustrate, stratifying students by ability might lead to reactions in the school system itself, e.g., the emergence of norms that assign more qualified 1 Partially along these lines, Cullen, Jacob, and Levitt (2006) explore how school choice affects students' attitudes and behaviors. 2 See, for example, the discussions in Banerjee and Duflo
The consequence of high school exit examinations for lowperforming urban students: Evidence from Massachusetts. Educational Evaluation and Policy Analysis
, 2010
"... In specifying a minimum passing score on examinations that students must pass to obtain a high school diploma, states divide a continuous performance measure into dichotomous categories. Thus, students with scores near the cutoff either pass or fail despite having essentially equal skills. The autho ..."
Abstract

Cited by 18 (1 self)
 Add to MetaCart
In specifying a minimum passing score on examinations that students must pass to obtain a high school diploma, states divide a continuous performance measure into dichotomous categories. Thus, students with scores near the cutoff either pass or fail despite having essentially equal skills. The authors evaluate the causal effects of barely passing or failing a high school exit examination on the probability of graduation using a regression discontinuity design. For most Massachusetts students, barely failing their first 10th grade mathematics or English language arts (ELA) examination does not affect their probability of graduating. However, lowincome urban students who just fail the mathematics examination have an 8 percentage point lower graduation rate than observationally similar students who just pass. There is no analogous impact from just passing or failing the ELA exit examination. For these urban, lowincome students, barely failing the mathematics test does not affect the likelihood of ontime grade promotion, but it does cause students to be 4 percentage points more likely to drop out of school in the year following the test. Lowincome urban students are just as likely to retake the test as equally skilled suburban students, but they have less success on retest.