Researchers have explored variables that may increase or decrease helpful bystander actions in different kinds of scenarios and identified some circumstances in which the likelihood of intervention is higher. For example, in some studies found that the bystander effect is attenuated when the scenario is perceived as dangerous (as compared to a non-dangerous one), when the perpetrators were present (versus non-present), and when the costs of intervening was physical (compared to non-physical).
From research, we know that bystander intervention can fail because potential responders do no:
Research suggests that therefore the most likely barriers to intervention are distractions, ignorance and ambiguity, failure to take responsibility, not being equipped to intervene, and "audience inhibition" such as norms against intervening or not wanting to appear intrusive.
Many of these barriers to intervention can be addressed through education such as workshops on bystander intervention that teach intervention skills and clarify ambiguity, and by addressing influencing factors in the social context, for example through community work.