2001
DOI: 10.1093/oxfordjournals.pan.a004868
|View full text |Cite
|
Sign up to set email alerts
|

Logistic Regression in Rare Events Data

Abstract: We study rare events data, binary dependent variables with dozens to thousands of times fewer ones (events, such as wars, vetoes, cases of political activism, or epidemiological infections) than zeros ("nonevents"). In many literatures, these variables have proven difficult to explain and predict, a problem that seems to have at least two sources. First, popular statistical procedures, such as logistic regression, can sharply underestimate the probability of rare events. We recommend corrections that outperfor… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

10
799
1
11

Year Published

2011
2011
2024
2024

Publication Types

Select...
10

Relationship

0
10

Authors

Journals

citations
Cited by 3,060 publications
(931 citation statements)
references
References 45 publications
10
799
1
11
Order By: Relevance
“…Since membership in a political organization could be considered a rare event (with a rate as low as 7 % in our sample), we tested the robustness of the findings by following the suggestions described in King and Zeng (2001) and using King's software for treating rare events in logistic regression, implemented in Stata 12. This procedure yielded exactly the same pattern of significant (and nonsignificant) statistical relationships that was described above.…”
Section: Resultsmentioning
confidence: 99%
“…Since membership in a political organization could be considered a rare event (with a rate as low as 7 % in our sample), we tested the robustness of the findings by following the suggestions described in King and Zeng (2001) and using King's software for treating rare events in logistic regression, implemented in Stata 12. This procedure yielded exactly the same pattern of significant (and nonsignificant) statistical relationships that was described above.…”
Section: Resultsmentioning
confidence: 99%
“…Therefore, I tested corrections of potential biases using the 'rare events logistic regression' model (ReLogit by King & Zeng (2001a, 2001b). However, running the models with 'ReLogit' changed only slightly the coefficients that were obtained (only at the 3 rd decimal).Thus, I decided to not use ReLogit given the rather small differences in the outcomes and the constraints imposed by this procedure on calculating predicted probabilities.…”
Section: Endnotesmentioning
confidence: 99%
“…extremely rare (King and Zeng 2001;Stuart and Sorenson 2008). With such data, regression parameters and probability estimates are biased.…”
mentioning
confidence: 99%