2015
DOI: 10.1016/j.econedurev.2014.11.001
|View full text |Cite
|
Sign up to set email alerts
|

Horseshoes, hand grenades, and treatment effects? Reassessing whether nonexperimental estimators are biased

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

2
40
0

Year Published

2015
2015
2022
2022

Publication Types

Select...
6
1

Relationship

0
7

Authors

Journals

citations
Cited by 32 publications
(42 citation statements)
references
References 21 publications
2
40
0
Order By: Relevance
“…Fewer studies measure mean differences in bias between quasi-experimental and experimental estimates, and compare this to a threshold considered meaningful by policymakers [23,35,78]. Fewer still attempt to measure the statistical significance of differences in effect sizes between designs [10,29,30].…”
Section: Metric For Concordancementioning
confidence: 99%
See 1 more Smart Citation
“…Fewer studies measure mean differences in bias between quasi-experimental and experimental estimates, and compare this to a threshold considered meaningful by policymakers [23,35,78]. Fewer still attempt to measure the statistical significance of differences in effect sizes between designs [10,29,30].…”
Section: Metric For Concordancementioning
confidence: 99%
“…In order to determine whether differences between two designs are statistically significant, we use a bootstrapping method used in some of the more rigorous WSCs to date [10,29,30,69]…”
Section: Computing Standard Errorsmentioning
confidence: 99%
“…A potential limitation of traditional regression is that even if all relevant confounders are controlled for, the model may still yield biased estimates if the relationship between these factors and the outcome measure is not appropriately modelled. This issue may become a great concern in situations where the case and control groups differ essentially in the distribution of background characteristics (Fortson et al 2015). To estimate a population effect while also addressing this issue, we extend the analysis by using Propensity Score Matching (PSM) (Abadie and Imbens 2006).…”
Section: Methodsmentioning
confidence: 99%
“…However, similarly as in the regression approach, PSM cannot account for unmeasured confounders. Both methods, therefore, relied on the assumption that there are no additional factors correlated with both the T1DM and school grades (Fortson et al 2015). 4 The matched case-control design of the data in this study implies not only that T1DM is overrepresented in the sample compared to in the general population but also that the information about the distribution of the matching variables is lost.…”
Section: Methodsmentioning
confidence: 99%
“…PSM approach has become popular in the context of program evaluation, which is extensively used by statisticians and economists until recently. Many notable examples include (Heckman et al, 1997(Heckman et al, , 1998Harding, 2003;List et al, 2003;Mensah et al, 2010;Ponzo, 2013;Fortson et al, 2015).…”
Section: Propensity Score Matching Approachmentioning
confidence: 99%