Full Citation
Title: Erratum for Glynn and Quinn (2011)
Citation Type: Journal Article
Publication Year: 2019
ISBN:
ISSN:
DOI: 10.1093/pan/mpr021
NSFID:
PMCID:
PMID:
Abstract: In the pages of this journal, Glynn and Quinn (2011a) suggests that the typical parametric and semi-parametric approaches to causal inference yield impossibly high estimates of election day registration (EDR), one of the most scrutinized forms of convenience voting. After consulting their replication materials, available in the journal’s public Dataverse collection (hdl: 1902:1/ 15920; Glynn and Quinn [2011b]), I identified an unfortunate oversight in their coding. Glynn and Quinn mistakenly code Michigan as an EDR state and Maine as a control state in their data set. The state variable in the Current Population Survey (CPS) data set is GESTFIPS (or alternatively, GESTCEN); the authors create an EDR variable and assign a value of 1 to the states with GESTFIPS codes of 16, 27, 26, 33, 55, and 56. Referencing the 2004 CPS codebook, these values correspond to Idaho, Minnesota, Michigan, New Hampshire, Wisconsin, and Wyoming. As evidenced by the authors’ commentary within their code, they actually intended to include Maine and omit Michigan to reflect the true collection of states that offered EDR in 2004. I reproduce the authors’ findings exactly when I retain the coding error (Table A1 of . . .
User Submitted?: No
Authors: Glynn, Adam N; Quinn, Kevin M
Periodical (Full): Political Analysis
Issue: 1
Volume: 24
Pages: e1-e2
Data Collections: IPUMS CPS
Topics: Other
Countries: