Almost half of US internet users say they have been a victim of online harassment or abuse ranging from name-calling to stalking to physical threats, a survey showed.
Women, the under-30s, and people identifying as lesbian, gay, or bisexual were all more likely to experience such harassment -- and were also more likely to self-censor what they post online as a result, researchers found.
Forty-seven percent of people reported experiencing at least one form or online harassment or unwanted contact, according to the study by the Data & Society Research Institute and Center for Innovative Public Health Research.
More than a third (36 percent) had suffered "direct harassment," including being called offensive names, threatened physically or stalked.
Three in 10 said they had been victims of "invasion of privacy," including having sensitive data or photographs stolen or posted without their permission, or having their online activity tracked.
A large majority of people -- 72 percent -- also said they had witnessed such conduct online.
"This study shows that not everyone has the same kind of experiences online -- so if you don't personally see or experience much harassment, that doesn't mean that isn't the case for many other Americans," lead researcher Amanda Lenhart said.
"These findings show that the presence -- or threat -- of online harassment can have effects on the overall tone of online discourse, even beyond those who are directly targeted," lead researcher Amanda Lenhart said.
Most of those who experienced harassment had taken steps to protect themselves.
Some 43 percent changed their email address or phone number, or created a new social media profile under a different name; 33 percent asked for help from a friend, family member, law enforcement, or domestic violence support organization.
One in four either reported or flagged unwelcome content or disconnected from online networks or social media altogether, the researchers found.
Social networks have been taking steps to curb online abuse and bullying.
Most recently, Twitter began rolling out a tool that allows users to "mute" unwelcome comments from so-called "trolls."