How do definitions of political violence affect comparisons between far right and far left incidents?
Executive summary
Definitions and coding choices change whether recent data show a surge in left-wing incidents or confirm a long-standing dominance of far‑right violence: CSIS reports that left‑wing attacks outnumbered far‑right ones in the first half of 2025 (CSIS counted incidents from Jan. 1–July 4) [1]. Other researchers and datasets — and a now‑removed DOJ study — show that over longer stretches the far right has produced more incidents and far more fatalities [2] [3] [4].
1. Labels matter: what counts as “terrorism,” “attack” or “plot”
Research teams use different inclusion rules. CSIS analyzed “attacks and plots” and over a near‑seven‑month slice in 2025 found more left‑wing incidents than right‑wing ones, in part because it counted arson of police cars and smaller actions alongside larger plots [1] [5]. Other datasets focus on ideologically motivated homicides or events meeting stricter terrorism thresholds; those show far‑right actors have historically accounted for more lethal events and more fatalities [2] [3] [4].
2. Time windows and baselines reshape the story
Short snapshots can flip trends. CSIS notes that 2025’s first half is the first time in 30+ years left‑wing attacks outnumbered right‑wing ones — a statement anchored to a narrow period and contrasted with decades‑long patterns showing right‑wing predominance [1] [6]. Analysts warn that comparing a handful of incidents in six months to multi‑decade baselines can exaggerate apparent shifts [7] [8].
3. Severity versus frequency: one axis often omitted
Frequency of incidents and lethality diverge. Multiple sources report that right‑wing violence has historically been more deadly even when counts are closer — former DOJ and university research found far‑right attacks caused far more ideologically motivated homicides and fatalities [2] [4]. CSIS itself concedes left‑wing incidents “remain much lower than historical levels of violence carried out by right‑wing and jihadist attackers,” even as counts rose in 2025 [1].
4. Coding disputes: which acts are ideologically driven?
Researchers debate whether some events should be coded as political terrorism at all. Critics of the CSIS framing say the report’s left‑wing count relies on a small number of plots and some property‑damage incidents that other coders might omit, which risks inflating the appearance of a left‑wing surge [7]. Journalists and scholars note that classification judgments — motive, target, organization, and whether violence was intended to influence politics — produce divergent tallies [5] [7].
5. Institutional and political pressures affect what data are published
Removal of government reports changes public perception. The Justice Department quietly removed a study that concluded far‑right extremists had committed far more ideologically motivated homicides since 1990, a move observers linked to political dynamics as officials debate whether left or right poses the greater threat [2] [9]. That disappearance highlights how institutional choices — not just methodology — shape which narratives dominate.
6. Context: why both trends can be true simultaneously
Multiple sources argue a coexistence of trends: long‑term dominance of far‑right lethality, and a recent uptick in left‑wing incidents from a low base. CSIS and other analysts explicitly state left‑wing violence has risen from very low levels, while the far right’s historic lethality remains a critical baseline [1] [6]. Observers warn that a temporary decline in right‑wing incidents can make even modest increases on the left appear momentous [5] [8].
7. What readers should watch for in future comparisons
Look for transparency about definitions, consistent timeframes, and whether counts weigh severity or frequency. Independent replication across multiple datasets — coding the same incidents with uniform rules — is the best way to test claims. Several analysts and outlets emphasize that the headline claim (left overtaking right in 2025) rests on definitional and temporal choices that change the result [7] [6].
Limitations and final note: available sources do not mention a single universally accepted standard for categorizing domestic political violence; researchers and agencies continue to disagree on exact inclusion rules and thresholds (not found in current reporting). The debate is methodological and political: the same underlying events can support differing narratives depending on who defines the terms and which time slice they choose [1] [2] [7].