Replication data for: Theory and Evidence in International Conflict: A Response to de Marchi, Gelpi, and Grynaviski (doi:10.7910/DVN/S7JLEL)

View:

Part 1: Document Description
Part 2: Study Description
Part 3: Data Files Description
Part 4: Variable Description
Part 5: Other Study-Related Materials
Entire Codebook

Replication data for: Theory and Evidence in International Conflict: A Response to de Marchi, Gelpi, and Grynaviski

doi:10.7910/DVN/S7JLEL

Harvard Dataverse

2007-11-28

4

Beck, Nathaniel; King, Gary; Zeng, Langche, 2007, "Replication data for: Theory and Evidence in International Conflict: A Response to de Marchi, Gelpi, and Grynaviski", https://doi.org/10.7910/DVN/S7JLEL, Harvard Dataverse, V4, UNF:3:N0bEAswAlPPVXCxPOZYyqw== [fileUNF]

Replication data for: Theory and Evidence in International Conflict: A Response to de Marchi, Gelpi, and Grynaviski

doi:10.7910/DVN/S7JLEL

Beck, Nathaniel (University of Rochester)

King, Gary (Harvard University)

Zeng, Langche (UC San Diego)

2004

Harvard Dataverse

Harvard Dataverse

2006

2004

https://doi.org/10.7910/DVN/S7JLEL

Social Sciences

We thank Scott de Marchi, Christopher Gelpi, and Jeffrey Grynaviski (2003; hereinafter dGG) for their careful attention to our work (Beck, King, and Zeng, 2000; hereinafter BKZ) and for raising some important methodological issues that we agree deserve readers' attention. We are pleased that dGG's analyses are consistent with the theoretical conjecture about international conflict put forward in BKZ --- "The causes of conflict, theorized to be important but often found to be small or ephemeral, are indeed tiny for the vast majority of dyads, but they are large stable and replicable whenever the ex ante probability of conflict is large" (BKZ, p.21) --- and that dGG agree with our main methodological point that out-of-sample forecasting performance should always be one of the standards used to judge s tudies of international conflict, and indeed most other areas of political science. <br /> <br /> However, dGG frequently err when they draw methodological conclusions. Their central claim involves the superiority of logit over neural network models for international conflict data, as judged by forecasting performance and other properties such as ease of use and interpretation ("neural networks hold few unambiguous advantages... and carry significant costs" relative to logit; dGG, p.14). We show here that this claim, which would be regarded as stunning in any of the diverse f ields in which both methods are more commonly used, is false. We also show that dGG's methodological errors and the restrictive model they favor cause them to miss and mischaracterize crucial patterns in the causes of international conflict. <br /> <br /> We begin in the next section by summarizing the growing support for our conjecture about international conflict. The second section discusses the theoretical reasons why neural networks dominate logistic regression, correcting a number of methodological errors. The third section then demonstrates empirically, in the same data as used in BKZ and dGG, that neural networks substantially outperform dGG's logit model. We show that neural networks improve on the forecasts from logit as much as logit improves on a model with no theoretical variables. We also show how dGG's logit analysis assumed, rather than estimated, the answer to the central question about the literature's most important finding, the effect of democracy on war. Since this and other substantive assumptions underlying their logit model are wrong, their substantive conclusion about the democratic peace is also wrong. The neural network models we used in BKZ not only avoid these difficulties, but they, or one of the other methods available that do not make highly restrictive assumptions about the exact functional form, are just what is called for to study the observable implications of our conjecture. <br /> <br /> This paper is a response to a comment on Beck, Nathaniel; King, Gary; and Zeng, Langche, 2000, "Improving Quantitative Studies of International Conflict: A Conjecture," American Political Science Review, Vol. 94, No. 1, 21-36. (Article: <a href= "http://gking.harvard.edu/files/improv.pdf" target="_blank">PDF</a> | Abstract: <a href= "http://gking.harvard.edu/files/abs/improv-abs.shtml" target="_blank">HTML</a>) <br /><br /> See also: <a href="http://gking.harvard.edu/categor y/research-interests/applications/international-conflict" target="_blank">International Conflict </a>

This dataset is made available without information on how it can be used. You should communicate with the Contact(s) specified before use.

Beck, Nathaniel, Gary King, and Langche Zeng. 2004. Theory and Evidence in International Conflict: A Response to de Marchi, Gelpi, and Grynaviski 98: 379-389: <a href= "http://j.mp/jC8Yt2" target="_blank">Link to article</a>

Beck, Nathaniel, Gary King, and Langche Zeng. 2004. Theory and Evidence in International Conflict: A Response to de Marchi, Gelpi, and Grynaviski 98: 379-389: <a href= "http://j.mp/jC8Yt2" target="_blank">Link to article</a>

• Number of cases: 23529

• No. of variables per record: 24

• Type of File: text/tab-separated-values

UNF:3:N0bEAswAlPPVXCxPOZYyqw==

The data with all variables used in dGG's and our analyses

• YEAR - CALENDAR YEAR
• DISP - DISPUTE ORIGINATED THIS YEAR
• ASYM - ASYMMET LAGGED T-1
• ASYMSQ - ASYMMETRY SQUARED
• CONTIG - CONTIG LAGGED T-1
• ALLY - ALLIANCE LAGGED T-1
• SQ - STATUS QUO LAGGED
• DEMA - DEMOCRACY SCORE FOR A - LAGGED T-1
• DEMB - DEMOCRACY SCORE FOR B - LAGGED T-1
• DISTANCE - DISTANCE FROM A TO B
• LNDIST - LN OF DISTANCE FROM A TO B
• MAJDYD - MAJOR POWER DYAD
• JNTDEM - DEM A(0-21) X DEMB(0-21)
• JNTDEMSQ - JOINT DEMOCRACY SQUARED
• PY - NEW PEACEYRS VARIABLE
• PEACYR1 - PY: (.,4)
• PEACYR2 - PY: (4,9)
• PEACYR3 - PY: (9,16)
• PEACYR4 - PY: (16,26)
• PEACYR5 - PY: (26,.)
• R - UNIFORM RANDOM VARIABLE FOR DATSET ASSIGNMENT
• DATASET - 1=TRAINING SET; 2=POST-85; 3=RANDOM DRAW
• STATEA - RANDOMLY GENERATED STATE A
• STATEB - RANDOMLY GENERATED STATE B

f101684 Location:

Variable Format: numeric

Notes: UNF:3:QdeKiGu90BKm2egRRP8J7Q==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:3oXkATjrKuA1C2D2G6vmIw==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:yx5stl+X5KkFJBxv/oQRIA==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:fU3iqEmhundKbpl4nsqjqA==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:8INQ3PdHAx1/wQhmd9+OHw==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:Tccbc7kc9J4YpjI+t36RnA==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:UYM3ezlVLfBsAptImiP+lw==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:kRkgvK18N6x/P4z9CQh1ig==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:WnwHVviz17LxL3410RzW0A==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:CzKQ8YVLCKsDXSyDBo2qpw==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:19nmRYWqziAwG/+ttDMZxg==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:4KK9bxRN0+q+u+0ewi+r2g==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:SA7Lxy+9PJpfFek8jqR61w==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:oKZAw1GUBMZxtvrnOPaXjg==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:Z1xBCo8hpghvQBRX4SUrnw==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:Ei0WK1U3plmblWzxMTindw==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:xgJcBawedCYX/FOV5OjhnA==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:h716yd/Tg6ZasXP+cffC0A==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:BWiRUd0jWs+diZ6VwxP96A==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:kP8QhjFOHsKhW4DfjysehQ==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:4WKJO9DelX4VPiucInOb9g==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:1LRhLlJ/Un+vXfkO8ac3eA==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:92jBBtFu9jXNzQfc6DH6mw==

f101684 Location:

Variable Format: numeric

Notes: UNF:3:A2JwbPPkSz7clKotnII7Ig==

Label:

duke.dta

Text:

The data with all variables used in dGG's and our analyses, Stata format

application/x-stata

Label:

main.zip

Text:

Collects together: duke.dta, nn.x, nn.y, nn.spec, w.1, w.2, w.3, nnroc_post85.out

application/zip

Label:

mc.zip

Text:

Collects together the parms*.dat files, and the final weights for the 150 nn's, zip file format

application/zip

Label:

nn.spec

Text:

An example bb5 (BigBack version 5) specification file for our final model

text/plain; charset=US-ASCII

Label:

nn.x

Text:

Contains the (normalized) input data for neural networks

text/plain; charset=US-ASCII

Label:

nn.y

Text:

Contains the dependent variable data

text/plain; charset=US-ASCII

Label:

nnroc_post85.out

Text:

roc data files

text/plain; charset=US-ASCII

Label:

nnroc_random.out

Text:

roc data files

text/plain; charset=US-ASCII

Label:

parms1.dat

Text:

MC experiments data, exact configuration for nn's, contains 50 nn's, with random seeds for starting values generated from a uniform draw using a seed number, in ASCII format

text/plain; charset=US-ASCII

Label:

parms3.dat

Text:

MC experiments data, exact configuration for nn's, contains 50 nn's, with random seeds for starting values generated from a uniform draw using a seed number, in ASCII format

text/plain; charset=US-ASCII

Label:

parms9.dat

Text:

MC experiments data, exact configuration for nn's, contains 50 nn's, with random seeds for starting values generated from a uniform draw using a seed number, in ASCII format

text/plain; charset=US-ASCII

Label:

readme.txt

Text:

Detailed information on data and documentation files

text/plain; charset=US-ASCII

Label:

rndsmpl.g

Text:

MC experiments supplementary document: random sample with replacement

text/plain; charset=US-ASCII

Label:

TheoryEvidenceArticle.pdf

Text:

Article related to this study: Theory and Evidence in International Conflict: A Response to de Marchi, Gelpi, and Grynaviski

application/pdf

Label:

toe-resp-repl.zip

Text:

Data in zipped file format

application/zip

Label:

w.1

Text:

Files of weights at convergence for the members of the committee that can be checked against replication run results

text/plain; charset=US-ASCII

Label:

w.2

Text:

Files of weights at convergence for the members of the committee that can be checked against replication run results

text/plain; charset=US-ASCII

Label:

w.3

Text:

Files of weights at convergence for the members of the committee that can be checked against replication run results

text/plain; charset=US-ASCII