Posts
2 min read

How gun control and gun rights groups frame their fight

My sociology master's thesis: topic models on 16,687 press statements to find each movement's frames, then which frames get likes, comments and shares on Facebook.

My master's thesis in computational sociology at the University of North Texas, July 2023. Committee: William Scarborough (chair), Gabe Ignatow and Donna Barnes. Download the full thesis (PDF).

The paradox

The US stands out on gun deaths among wealthy countries.

US gun deaths per person in 2016, as a multiple of each country's rate

  • United Kingdom
    35.3×
  • Australia
    10.6×
  • Sweden
    8.2×
  • Canada
    5×

Source: Global Burden of Disease 2016 Injury Collaborators (2018), as cited in the thesis.

Most Americans back basic gun measures, yet little passes. Kristin Goss calls this the gun control paradox. One reason she points to is framing: how each side names the problem, the fix, and why it's urgent.

0%+of Americans back basic measures like concealed-carry permits (Pew)

Two questions, one pipeline

What frames does each movement use? And which ones does the public actually engage with?

1 · What frames do they use?

Scrape press statements

9 organizations, early 1990s to 2022

16,687 statements

7,332 gun control · 9,355 gun rights

LDA topic model

One per movement, 2 to 20 topics tried

6 frames a side

Read and named by hand

2 · How does the public engage with them?

Scrape Facebook posts

Everytown 8,891 · NRA 1,659

Same topic model

Scores each post's mix of frames

OLS regression

Controls: images, video, emotion, sentiment

Likes, comments, shares

Which frames do better or worse

Six frames a side

I trained a topic model for each movement and tried 2 to 20 topics. Six is where the gains level off.

Topic coherence by number of topics
0.360.400.440.4826101418Number of topics6 topics: where the gains level off

Redrawn from the thesis's Figures 2 and 3; values read off the plots.

Each topic is a cluster of words that tend to show up together. I read the statements behind each one and named the frame. Tap a frame for a real quote.

Gun control

Everytown, Brady, Giffords, Coalition to Stop Gun Violence, Sandy Hook Promise, Violence Policy Center

Gun rights

NRA, Gun Owners of America, Second Amendment Foundation

What people engage with

Then I scored every Facebook post from Everytown and the NRA, the largest group on each side, by its mix of frames, and checked which frames predict more or less engagement.

Change in

Everytown (gun control)

  • Concealed Carry
    +11%
  • Violence & Victims
    -31%***
  • Elections & NRA
    -14%
  • Firearms Industry
    +21%
  • Background Checks
    +34%**
  • School Shootings
    +6%

NRA (gun rights)

  • Elections
    -15%
  • State Laws
    +105%***
  • Firearms
    -57%**
  • Gun Rights
    +32%*
  • Shootings
    -1%
  • Legislation
    -48%**

A post made entirely of one frame, compared with posts that don't use it, from separate OLS models with controls for images, video, emotion and sentiment. Solid bars are significant (* p<.05, ** p<.01, *** p<.001); outlined bars aren't. Engagement is logged, so the percentages are approximate.

  • Everytown: Background Checks earned more likes, and Firearms Industry far more comments and shares. Violence & Victims did worse on all three.
  • NRA: State Laws roughly doubled likes and shares, and Gun Rights lifted all three. Firearms, Legislation and Shootings did worse.

The pattern

The frames that worked on both sides did three things at once.

  1. 1 · Diagnose

    Name a specific problem.

    e.g. The gun show loophole

  2. 2 · Prescribe

    Offer a specific fix.

    e.g. Universal background checks

  3. 3 · Motivate

    Give a reason to act now.

    e.g. A vote or protest coming up

Did well

Everytown: Background Checks, Firearms Industry

NRA: State Laws, Gun Rights

Did worse

Diagnosis only: Violence & Victims, Firearms

Contested: the NRA's "good guy with a gun" Shootings posts

The difference is in the foundation. Everytown wins with concrete problems and fixes. The NRA wins by treating the Second Amendment as the thing to defend. Its top State Laws post was a single image: "25 states have enacted NRA Constitutional Carry."

0+likes on the NRA's top State Laws post

Caveats

  • The model reads Facebook posts with the vocabulary it learned from press statements, so new topics on Facebook go unseen.
  • A comment or share can be an argument, not support.
  • Framing is one piece. Money, organizing and polarization among politicians matter too.

The full thesis, with the methods, tables and references: Prasanna (2023), PDF, 63 pages.