the gun ownership attribute had three levels: no gun ownership, owning a pistol, and owning an AR-15,
This study design is bad, and they should feel bad. If they’re going to claim that people are afraid of AR-15s, they should compare it apples-to-apples with other rifles, or just ask about rifles generally, like they did with pistols.
Furthermore, any study asking opinion questions for what should be data-driven decisions are misleading at best and harmful at worst. If your concern is safety in communities, you should study actual safety, not feelings. It appears they want to make people feel safe, while not necessarily increasing safety.