Thread by @AnnaMeierPS

Welcome to a late-night thread about data, terrorism data, and what we think we’re measuring when we’re measuring terrorism. 1/

This thread was sparked by Benjamin Allison’s new article on coding inconsistencies in the CSIS and New America datasets on terrorist attacks in the U.S. As a frequent critic of the CSIS data in particular, I’m sympathetic to the goal here. BUT… universiteitleiden.nl/binaries/conte… 2/

…there are additional questions I want to raise. The first: are the CSIS/New America datasets unique in their somewhat sloppy coding of what is and isn’t terrorism and their judgments about motives in contexts of uncertainty? Not at all. 3/

I cannot stress this enough: every. single. dataset on something called “terrorism” includes cases that will challenge your intuition on what is or isn’t terrorism, depending on where you sit. Think hate crimes are terrorism? You’ll be upset. Think they’re not? So will you. 4/

You can pick almost any attack out of any dataset & argue for why it should or shouldn’t be there. There was such controversy over the Las Vegas shooting at the GTD that poor Erin Miller wrote a whole post explaining that its inclusion in the dataset. 5/ start.umd.edu/news/global-te…

Full disclosure: I used to work at the GTD, & its coding rules have profoundly influenced how I think about terrorism—namely, that it’s what we want it to be, which is a conclusion not many would take from the GTD or that GTD staff would support, I suspect! Let me explain. 6/

I remember my intuition initially chafing at including attacks by Fulani herders, skirmishes b/n Philippine troops & the NPA, & literally everything IS ever did in the database (even w/ variables for recording degree of doubt). That wasn’t how I’d understood terrorism before. 7/

But 2 examples stand out: the Ukrainian & Syrian conflicts. In Ukraine, rebels fired rockets at villages every day for a while. Was that terrorism, or was that war? What’s your intuition? How did you arrive at that? What’s the logic that affects how you answer that question? 8/

Syria was a different problem, bc we knew things were happening. but the quality of info on the ground was so poor & our rules for inclusion were demanding, so for a while we *knew* we were undercounting attacks. (The GTD took steps to address this after I left.) 9/

So here we have a few parallel problems: what is legible to you as terrorism, what fits the rules and what happens when they conflict with legibility, and how the hell do you know anything about anything for certain? 10/

Allison raises numerous issues w/ vague, contradictory, or generally incomplete info in his article. I won’t dispute individual codings; I agree w/ many of his analyses & disagree w/ others. My point is that this isn’t a CSIS/New America problem. It’s a terrorism problem. 11/

And it’s a problem because there are political interests driving what we do & don’t consider terrorism. The term is not objective; there is no discrete violent tactic that is “terrorism.” We can apply a definition, but that definition doesn’t exist in a void absent politics… 12/

…nor does that definition help us delineate a category that *actually helps us understand political violence* vs. one that advances a particular narrative about what violence is legitimate, what is illegitimate, and who gets to decide. 13/

Allison says it is important to code events “properly,” which for him I think means in accordance with available info about motives. There’s a reason the GTD codes motives conservatively and doesn’t code perp ideology: this is so, so complicated. 14/

What codes do you use? Broad ones? Nuanced? Can 1 event have multiple codes? Motivations are not monocausal. If the killer of 6 Asian women in Atlanta does not explicitly voice neo-Nazi sympathies, is his targeting of a racial group unrelated to structural white supremacy? 15/

I don’t have good answers for those who want to build terrorism datasets. My personal preference is to jettison the category of “terrorism” & focus on what we actually care about. Racist attacks? Go for it. Nonstate targeting of civilians? I’m all ears. 16/

But the “terrorist” category is a political project & a subjective one, & it’s measuring a category constructed by dominant groups & 1 we become complicit in perpetuating when we try to lift it from its context & reclaim it as something objective. It’s not. It will never be. /fin

P.S. To be clear, I commend @BenVAllison for applying such careful scrutiny to the CSIS/New America data, which got considerable media coverage. “Why this case and not that” is a crucial question. So let me extend it: why “terrorism” and not something else.