Codebook
The rules used to read and code every post: one severity level, a shield flag (for posts that justify themselves by protecting children or women), and any number of narrative codes.
Codebook v3.2
Single coding standard for this analysis.
Change log
- 2026-11-04: v3 adopted unchanged.
- 2026-11-05: v3.1 (project decision). Outlet and tabloid crime coverage is no longer presumed neutral. Rule 5 was rewritten. A C2 bullet was added for motive displacement, and a C0 bullet was narrowed. Narrative MOTIVE_DISPLACEMENT added. Worked example: the NY Post headline of 2026-10-15 on the Aurora, Illinois Hindu temple shooting plot. The headline led with the suspect's trans identity and jail hormone treatment, while the suspect's self-declared neo-Nazi ideology, the documented motive, came second. The headline invites readers to link the crime to being trans. That is bad-faith framing: C2 SHOOTER_ATTRIBUTION MOTIVE_DISPLACEMENT GOV_SPENDING, with the correction naming the documented motive. Posts already coded before this change are recoded under it (crime-story candidates in the coded sample).
- 2026-11-05: v3.2 (from coder reports on the rule-5 recode). Rule 5 now covers bare links (the linked headline or slug is the frame), hooks on a suspect's partner or associates, and single-incident political violence. The rule-5 recode batches were coded under v3.1. One coder already applied these readings; any cases coded otherwise are left as coded and noted in the methods.
- 2026-11-09: clarifications (after the bridge check and the adjudication of 236 cases; no new code, v3.2 stands). Outlet headlines that foreground an anti-trans measure or grievance are at least C1 (rounds 1-5 had coded many C0 under v3; all 1,043 earlier C0 posts were rechecked and 128 changed after adjudication). A post singling out one identifiable trans athlete or student is C4. The shield flag needs the protection argument stated or plainly implied. A hook on a suspect's trans partner is C2 unless the partner appears only for a role in the proceedings. C5 under rule 9 needs the post to point an audience or the authorities at the person or institution. Rule 11 is for real ties, not for unreadable replies. Full wording: the project's analysis item 9; applied by coders from full-pass on.
Each post gets:
- one severity level: the highest that applies;
- a shield flag;
- any number of narrative codes, for C1-C5 posts only;
- a target, for C4 and C5 posts.
Severity:
- C0 Not anti-trans:
- news headlines from outlets that do not foreground trans identity (see rule 5; a headline is never presumed neutral because an outlet posted it);
- support for trans people, and counter-speech;
- unrelated uses of "trans" (trans-Atlantic, trans fats).
- C1 Anti-trans rhetoric without a false claim or dehumanizing language:
- grievance framings: sports; protecting children; youth medical care; schools and "parental rights"; bathrooms, changing rooms, prisons and shelters; "gender ideology";
- plain denial of trans identity ("trans women are men");
- personal accounts foregrounding a perpetrator's trans identity in a crime story. C1 is anti-trans rhetoric. There is no "policy position" category, and C1 is never treated as legitimate or neutral.
- C2 False or misleading claim, with the correction named. Includes:
- trans "violence epidemics";
- attributing an attack to a trans person before officials identified them;
- motive displacement: framing that presents a crime as explained by, or linked to, the suspect's trans identity or transition (hormones, pronouns, prison placement) when officials or the suspect's own statements document a different motive (for example neo-Nazi, white-supremacist, antisemitic or other extremist ideology), and that motive is placed second or left out. The correction names the documented motive, and that being trans is not a motive;
- claims that regret or detransition is common;
- claims that minors routinely get surgery, or that schools arrange it;
- claims that trans athletes are numerous, or are taking over women's sport;
- claims stated as fact that care is "experimental" or has no evidence base. The correction cites the positions of the major US medical associations;
- fabricated or distorted statistics.
- C3 Dehumanizing or degrading:
- slurs;
- "mentally ill", "delusional", "confused" or "disordered" applied to trans people in any context, including a policy argument ("your child's delusion" is C3);
- saying trans people or trans minors belong in a mental institution or psychiatric ward;
- "mutilation", "castration", "butchering" or "child abuse" applied to gender-affirming care;
- predator or "groomer" smears;
- mocking misgendering; "it"; "freaks", "mutants", "demons";
- mockery of appearance.
- C4 Targeting: singles out a specific person for ridicule, condemnation or collective attention. A minor who is singled out is at least C4.
- C5 Harassment or incitement:
- pointing an audience or the authorities at a person or an institution (tagging an employer, school, hospital, regulator, committee or law enforcement, or posting contact details);
- doxxing; threats; suicide-baiting; mocking a death;
- calling for trans people to be institutionalized, forcibly treated or forced to detransition (for example "send them to the detransition clinic").
Shield flag: true when the post justifies itself by protecting children, or by women's or girls' safety or fairness.
Narrative codes (any number, C1-C5 posts only):
- DENIAL: denies trans identity or existence; misgendering; scare quotes.
- POLITICAL_ENEMY: trans people or activists as "the left", antifa or the enemy.
- MEDICAL_HARM: care as harm ("mutilation", sterilization, regret).
- VIOLENCE_TERROR: trans people as violent or terrorists as a group.
- IDEOLOGY_CULT: "gender ideology" as cult, religion, indoctrination or agenda.
- CHILD_INDOCTRINATION: schools, books or Pride "pushing" trans identity on children; "parental rights".
- WOMEN_SPORTS: trans women and girls in female sport.
- SHOOTER_ATTRIBUTION: links an attack to a perpetrator's trans identity, or speculates that a shooter is trans.
- CRIME_CASE: a non-shooting crime story foregrounding trans identity.
- MOTIVE_DISPLACEMENT: a crime story that foregrounds the suspect's trans identity or transition over a documented motive or ideology (for example neo-Nazism), or omits that motive. Always at least C2 (see C2).
- WOMEN_SPACES: bathrooms, changing rooms, prisons, shelters.
- INSTITUTIONAL_CAPTURE: medicine, media, government or corporations "captured".
- MENTAL_ILLNESS: trans identity as mental illness or delusion (always at least C3).
- SPEECH_PERSECUTION: critics censored, fired or compelled (pronouns).
- CONTAGION: "social contagion", "rapid onset", a craze or a phase.
- FETISH_FRAUD: fetish or autogynephilia framings; transition as fraud.
- CHILD_PREDATOR: predator, grooming or pedophilia smears.
- VICTIM_DISMISSAL: denying or mocking discrimination, violence or hardship against trans people.
- APPEARANCE: mockery of looks, voices or bodies.
- GOV_SPENDING: taxpayer money for trans care, research or prisoners.
- COVERUP: authorities or media "hiding" trans perpetrators or data.
- CONSPIRACY: shadowy funders, pharma profit, depopulation plots.
- LGB_WEDGE: the trans movement framed as harming gay men and lesbians.
- SUICIDE_TAUNT: suicide or suicide statistics used as a taunt; suicide-baiting.
- RELIGIOUS_DEMONIZATION: "demonic", "satanic", trans people as evil.
- ISLAM_LINK: linking trans people to Islam, Hamas or pro-Palestinian politics.
- STATE_TAKES_CHILDREN: the state removing children from non-affirming parents; custody cases.
- PARENT_ABUSER: parents of trans children as abusers.
- TRANSVESTIGATION: claiming a public figure is secretly trans.
- MEDIA_PRONOUN_COMPLAINT: complaints that media or police call trans suspects "women".
- ACTIVIST_HARASSMENT: incident stories of activists harassing critics ("trantifa", "trans mob").
Decision rules:
- There is no legitimate-policy category. A grievance framing is at least C1. Code higher whenever anything else applies.
- Mental-illness or institutionalization language about trans people is C3 at minimum, whatever the tone or setting. It is C5 when it calls for confinement, forced treatment or forced detransition.
- "Mutilation"-type language about trans care is C3.
- C2 needs a correction you can name. The background facts you are given supply most of them.
- (v3.1) Crime and shooting coverage: outlets, tabloids, aggregators and personal accounts are coded by the same rules, and none is presumed to be "just reporting the news". Read the headline, the first sentence and any image as the frame.
- C0 only if the post does not foreground trans identity: it is absent from the headline and opening, or mentioned in passing and not as the hook.
- C1 CRIME_CASE or SHOOTER_ATTRIBUTION if the headline, opening or image foregrounds the suspect's trans identity or transition (hormones, pronouns, deadname, prison placement, "trans-identified male").
- C2 MOTIVE_DISPLACEMENT if it does so while a different documented motive or ideology is placed second or left out; or if it implies that being trans explains the crime; or if it asserts the suspect is trans before officials say so. Name the documented motive in c2_correction.
- Add GOV_SPENDING when the hook is taxpayer-funded care in custody, and COVERUP when it claims media or police are hiding trans identity.
- "Trans terror", "trans violence" as a series, "trans maniacs" and the like are C2 (a trans violence epidemic) and C3 if dehumanizing.
- Coverage of crimes against trans people, and counter-speech pointing out motive displacement, stay C0.
- (v3.2) A bare link, or a link with a short comment, adopts the linked headline. When the trans framing is in the link's headline or URL slug, code that as the frame.
- (v3.2) The rule covers anyone connected to the suspect: hooks on a partner's, relative's or associate's trans identity are coded like hooks on the suspect's.
- (v3.2) It covers single incidents of political violence or protest too ("trans flag attack", "trantifa"). Framing one incident as typical of trans people is C2 (trans violence epidemic) plus VIOLENCE_TERROR.
- People named in crime or court news are not targets just because they are named.
- A quote-post that endorses another post adopts it. Code the combination.
- A mockery account posting someone's words without comment is C4. A news aggregator doing the same is C0.
- C5 covers institutions (schools, hospitals, committees) as well as people, when the post points an audience or the authorities at them.
- An argumentative reply with a dehumanizing insult is C3. If it singles out a person for ridicule, it is C4.
- When torn between two levels, choose the higher one and mark confidence low.
- Code every author by the same rules.
Low-reliability codes (blind second-coder kappa under 0.6, full coding pass, codebook v3.2): MOTIVE_DISPLACEMENT.
Agreement was measured on 5,000 posts coded a second time, blind: severity kappa 0.81, anti-trans or not 0.88, shield flag 0.81.