What can we learn from predictive modeling?

Skyler J. Cranmer, Bruce A. Desmarais, Jr.

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

The large majority of inferences drawn in empirical political research follow from model-based associations (e.g., regression). Here, we articulate the benefits of predictive modeling as a complement to this approach. Predictive models aim to specify a probabilistic model that provides a good fit to testing data that were not used to estimate the model's parameters. Our goals are threefold. First, we reviewthe central benefits of this under-utilized approach from a perspective uncommon in the existing literature:we focus on howpredictive modeling can be used to complement and augment standard associational analyses. Second, we advance the state of the literature by laying out a simple set of benchmark predictive criteria. Third, we illustrate our approach through a detailed application to the prediction of interstate conflict.

Original languageEnglish (US)
Pages (from-to)145-166
Number of pages22
JournalPolitical Analysis
Volume25
Issue number2
DOIs
StatePublished - Apr 1 2017

Fingerprint

predictive model
regression
literature

All Science Journal Classification (ASJC) codes

  • Sociology and Political Science
  • Political Science and International Relations

Cite this

Cranmer, Skyler J. ; Desmarais, Jr., Bruce A. / What can we learn from predictive modeling?. In: Political Analysis. 2017 ; Vol. 25, No. 2. pp. 145-166.
@article{5b74335751d74ce393a3d5e8b5e5519f,
title = "What can we learn from predictive modeling?",
abstract = "The large majority of inferences drawn in empirical political research follow from model-based associations (e.g., regression). Here, we articulate the benefits of predictive modeling as a complement to this approach. Predictive models aim to specify a probabilistic model that provides a good fit to testing data that were not used to estimate the model's parameters. Our goals are threefold. First, we reviewthe central benefits of this under-utilized approach from a perspective uncommon in the existing literature:we focus on howpredictive modeling can be used to complement and augment standard associational analyses. Second, we advance the state of the literature by laying out a simple set of benchmark predictive criteria. Third, we illustrate our approach through a detailed application to the prediction of interstate conflict.",
author = "Cranmer, {Skyler J.} and {Desmarais, Jr.}, {Bruce A.}",
year = "2017",
month = "4",
day = "1",
doi = "10.1017/pan.2017.3",
language = "English (US)",
volume = "25",
pages = "145--166",
journal = "Political Analysis",
issn = "1047-1987",
publisher = "Oxford University Press",
number = "2",

}

What can we learn from predictive modeling? / Cranmer, Skyler J.; Desmarais, Jr., Bruce A.

In: Political Analysis, Vol. 25, No. 2, 01.04.2017, p. 145-166.

Research output: Contribution to journalArticle

TY - JOUR

T1 - What can we learn from predictive modeling?

AU - Cranmer, Skyler J.

AU - Desmarais, Jr., Bruce A.

PY - 2017/4/1

Y1 - 2017/4/1

N2 - The large majority of inferences drawn in empirical political research follow from model-based associations (e.g., regression). Here, we articulate the benefits of predictive modeling as a complement to this approach. Predictive models aim to specify a probabilistic model that provides a good fit to testing data that were not used to estimate the model's parameters. Our goals are threefold. First, we reviewthe central benefits of this under-utilized approach from a perspective uncommon in the existing literature:we focus on howpredictive modeling can be used to complement and augment standard associational analyses. Second, we advance the state of the literature by laying out a simple set of benchmark predictive criteria. Third, we illustrate our approach through a detailed application to the prediction of interstate conflict.

AB - The large majority of inferences drawn in empirical political research follow from model-based associations (e.g., regression). Here, we articulate the benefits of predictive modeling as a complement to this approach. Predictive models aim to specify a probabilistic model that provides a good fit to testing data that were not used to estimate the model's parameters. Our goals are threefold. First, we reviewthe central benefits of this under-utilized approach from a perspective uncommon in the existing literature:we focus on howpredictive modeling can be used to complement and augment standard associational analyses. Second, we advance the state of the literature by laying out a simple set of benchmark predictive criteria. Third, we illustrate our approach through a detailed application to the prediction of interstate conflict.

UR - http://www.scopus.com/inward/record.url?scp=85020656720&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85020656720&partnerID=8YFLogxK

U2 - 10.1017/pan.2017.3

DO - 10.1017/pan.2017.3

M3 - Article

VL - 25

SP - 145

EP - 166

JO - Political Analysis

JF - Political Analysis

SN - 1047-1987

IS - 2

ER -