The Severity of War
A Longitudinal Analysis
Power Law
Introduction
“Long Peace” (Steven Pinker’s thesis that war is in decline
This study evaluates inter-state conflict from 1823 to 2003 through a dual lens. First, it applies statistical power-law modeling to determine if the structural deadliness of war shifted after 1945. Second, it contextualizes these global parameters by mapping the geographic distribution of per-capita fatalities, revealing that spatial density and regional geopolitics deeply distort generalized global trends.
Background Literature
What to cover: Discuss the literature surrounding the changing nature of warfare post-WWII (1945). Mention international relations theories concerning the “Long Peace,” the rise of international organizations (like the UN), nuclear deterrence, and democratic peace theory.
Connection to your data: Frame your Chronological Trend Graph and Power-Law Analysis as a formal empirical test of this literature. The literature asks if the international system became more stable after 1945; your statistical models test if the underlying “scale” (\(\alpha\)) of war deadliness actually changed or if the apparent decline in severity is statistically insignificant.
This is where your bridge happens. A statistical aggregate like a power-law distribution can tell us what the global system looks like, but it cannot tell us how it got that way. To understand why global war deadliness behaves as a stable structural system, we have to look inside the distribution at its geographic components. ## Methods
The datasets used for this assignment were pulled from the Correlates of War Project (COW). It includes the Wars dataset that contains data on the number of war deaths per country per war from 1823 until 2003 and Cap which contains the total population of each country per year from 1816 until 2012. The total number of entries in the Wars dataset is 337 observations and 25 variables while the Cap dataset is much larger at 15,171 observations and 11 variables. Both datasets were combined to accurately measure the correlation between war deaths (BatDeaths) and total population (tpop) per country. These variables were then calculated to find the ratio of war deaths by population and finally the aggregation of how many war deaths per a population of 100,000 occurred for each country between 1823 and 2003. To be as historically accurate as possible with consideration of mapping our data to a choropleth with modern state borders, certain countries were combined. The data of such countries were recoded as the present country that they later dissolved into. For example, the state of Prussia was recoded as ‘Germany’.