An Appraisal of Project Mars and theDivided ArmiesArgument

Abstract

Project Mars released what it claims to be a completely new dataset of 250 wars fought since 1800 and claims these data do not suffer from Western biases of other data projects (e.g., the Correlates of War [CoW]) that apparently overlook non-Western conflicts. These data featured prominently in a recent argument in Divided Armies (2020) about the negative relationship between military inequality and battlefield performance. Our appraisal of both the data and the argument advanced in Divided Armies suggests some caution with these claims. Project Mars does not amount to a completely new dataset on conventional wars. Instead, Project Mars only evaluated CoW's interstate war data, missed that the bulk of its wars are available elsewhere in CoW's data repository (i.e., as intrastate or extrastate wars), and may have missed important observations in CoW's typology (prominently intrastate wars) that could double the size of Project Mars. Combined with additional misgivings about the quality of Project Mars’ coding and how the military inequality data were not constructed as Project Mars implies, these have important implications for the core argument advanced in Divided Armies. Scholars of war should not pool wars in their statistical models as Project Mars does.

{peacesciencer}: An R package for quantitative peace science research

“…4 All of these functions use raw or pre-processed data included in the package. For example, uses a dyadic dispute-year version of the Militarized Interstate Dispute (MID) data offered by Gibler et al (2016: v. 2.2.1) and merges in information about whether there was an ongoing MID or a MID onset in a dyad-year, leader-year, leader-dyad-year, or state-year. also has some data innovations included in these functions.…”

“…Indeed, the pipe has the effect of forming something analogous to a drop-down menu, in which the user can “select” additional data/commands that they may want simply by specifying the function in that does what they want. For example, a researcher can follow with another pipe and communicate they want information about ongoing conflicts and conflict onsets from the Gibler-Miller-Little (GML) dispute dataset (Gibler et al, 2016). Following with will add information about ongoing conflicts and onsets in a given dyad-year.…”

“… 11. The documentation for , along with the directory on Github, provides ample explanation for how interstate conflicts in the GML conflict data are connected to specific leaders. Since interstate dispute data can have missing days, connecting leaders to dispute onsets involves some important archival research that is only possible because of the information collected by Gibler et al (2016) over 10 years of scholarship. The documentation also considers the multiple ways in which users can more flexibly code the concept of “initiation” based on participant-level summaries.…”