Utilizing The Correlates of War project

[Kevin23]

Over my summer break from school, I've become interested in The Correlates of War series of studies. I figure the data from this particular project would be helpful in future papers.

Here is a link to the COW homepage,

http://www.google.com/url?sa=t&sourc...eyhj68Eyr_Xx8w

However, I'm really at somewhat of loss in terms of correctly interpreting and potentially using the data as it sometimes appears disorganized to me.

If any fellow members of SWJ has experience with COW, please enlighten me in regards to how to use it.

[Cavguy]

Think of COW as a PoliSci/IR Wiki of conflicts. Its fields are coded to allow statistical analysis for patterns. It can be dense why it is organized the way it is until you get into "hard" IR theory studies.

That said, quality varies since it is user contributed, and many statistics are disputed between warring parties.

My verdict - as best as can be had for large-N studies of conflict, but beware use for small-N (<100) studies due to coding issues. That said, it is an accepted academic "standard" dataset in the IR community.

[arashinokage]

Cavguy hit the nail on the head. Some things that may be of greater relevance to your studies:

Issue Correlates of War (studied under Dr. Hensel):
www.paulhensel.org/icow.html

Data Visualizations by IBM
http://manyeyes.alphaworks.ibm.com/manyeyes/

Also recommend a search for "information aesthetics" and "visual complexity".

Let me know if this helps...

[Kevin23]

Think of COW as a PoliSci/IR Wiki of conflicts. Its fields are coded to allow statistical analysis for patterns. It can be dense why it is organized the way it is until you get into "hard" IR theory studies.

That said, quality varies since it is user contributed, and many statistics are disputed between warring parties.

My verdict - as best as can be had for large-N studies of conflict, but beware use for small-N (<100) studies due to coding issues. That said, it is an accepted academic "standard" dataset in the IR community.

From what you've told me Cavguy and fro what I've seen the studies look strong for the inter-state conflicts like WWII, WWI, and major conflicts like the American Civil War. However, they have very scarce and not very useful data for alot of extra-state conflicts, especially ones in recent military history like the wars in the Balkans and in Iraq and Afghanistan. Which is partially due to the fact again like you said, many of these conflicts have only just cooled down as in the case of the Balkans, or are still ongoing like Iraq and Afghanistan, and thus are subject to dispute and debate. However, despite this I figure by now that someone would have contributed or added data on these recent conflicts, even though it isn't entirely certain.

[Kevin23]

Cavguy hit the nail on the head. Some things that may be of greater relevance to your studies:

Issue Correlates of War (studied under Dr. Hensel):
www.paulhensel.org/icow.html

Data Visualizations by IBM
http://manyeyes.alphaworks.ibm.com/manyeyes/

Also recommend a search for "information aesthetics" and "visual complexity".

Let me know if this helps...

Thank you arashinokage,

Both the links you provided and the keywords you suggested are helping me better understand and visualize COW data.

[jmweinberg]

You might look up the economist Paul Collier for some good papers based of COW.

Here is the other big conflict data set used by academics...
http://www.prio.no/CSCW/Datasets/Armed-Conflict/

You need statistical analysis software like STATA or SPSS to properly use these data. Although you might be able to come up with some averages using excel, without doing regression analysis (controlling for variables while trying to find correlations) your results will be biased.

[Cavguy]

From what you've told me Cavguy and fro what I've seen the studies look strong for the inter-state conflicts like WWII, WWI, and major conflicts like the American Civil War. However, they have very scarce and not very useful data for alot of extra-state conflicts, especially ones in recent military history like the wars in the Balkans and in Iraq and Afghanistan. Which is partially due to the fact again like you said, many of these conflicts have only just cooled down as in the case of the Balkans, or are still ongoing like Iraq and Afghanistan, and thus are subject to dispute and debate. However, despite this I figure by now that someone would have contributed or added data on these recent conflicts, even though it isn't entirely certain.

The problem as I found in my graduate research is that much information is private in internal conflicts - meaning it is hidden or unknown, sometimes to the warring parties themselves.

Examples:

- Insurgent troop strength. We still can't tell you how many Taliban insurgents there are. Do you count just fighters? Auxillaries? Sympathizers? I doubt the Taliban knows either. Repeat times number of insurgent groups.

- External support. How much foreign aid was provided to an insurgent group or regieme? How much does the ISI contribute to the Taliban annually? How much from Saudi sources? Even years later it's opaque. Many countries aid is covert for the obvious reasons, and not published in official budgets.

- Outcomes. Was Northern Ireland a win, lose, or draw for the UK? Iraq for the US? El Salvador? Hard to define "win" in intra-state conflict, and depends on your time metric.


The bottom line is that while inter-state conflict is generally easy to count divisions, tanks, industrial output, budgets, etc., even decades later intrastate warfare data remains very very unreliable and subjective. This adds a lot of uncertainty in the traditional IR modeling/statistical model.