Sanctioning hate: en experiment on counterspeech
Counter-speech is a direct response to hate speech that seeks to undermine it. It is a form of punishment: it enforces the norm against hateful discourses. Counterspeech is currently advocated by many as the best strategy to counteract online hate speech as it does not carry the negative externalities of other interventions like censoring content. For this strategy to work, however, counterspeech has to occur sufficiently often. In a series of experiments, we investigate how often observers punish online hate speech, test interventions to increase the frequency of punishment, and explore the micro-mechanisms that explain the effects.
We first investigate whether and how frequently counterspeech is produced as a spontaneous response when participants observe hate speech online on an online platform. We find the baseline rate of spontaneous counterspeech to be small. Participants with a positive attitude towards the target of hate were more likely to use counterspeech, however, no other specific variables related to the participants, such as their gender or level of education, were important in determining whether or not they used counterspeech.
In a second experiment, we test different interventions to increase the willingness to punish online hate speech. First, we test whether participants are more likely to respond to hateful messages with counterspeech when they can observe previous participants have done so. Second, we test whether informing participants of the inappropriateness of using hate prior to participating in an online platform, thus increasing the salience of the norm, would increase their willingness to sanction such behavior. We find that exposure to counterspeech by anonymous others strongly encourages participants to act against hate. However, increasing the salience of the norm does not have an effect on the number of sanctions and, if anything, participants are less willing to sanction. Because retaliation is ruled out by the experimental design, we attribute this result to participants considering that sanctioning individuals that deviated from the norm in the high saliency norm condition was pointless.
In a third experiment, we look at how the treatments affected the normative and empirical expectations of the participants regarding the use of hate speech and its punishment. We thus measure effects on beliefs at the normative and meta-normative levels to explore the mechanisms that drive the findings in experiment 2.
Our results show that sanctioning behavior is contagious: when participants could observe the previous counterspeech they were more likely to follow suit. However, this increase in punishment is linked to an increased perceived inappropriateness of hate speech while the acceptability of punishing does not change.
We first investigate whether and how frequently counterspeech is produced as a spontaneous response when participants observe hate speech online on an online platform. We find the baseline rate of spontaneous counterspeech to be small. Participants with a positive attitude towards the target of hate were more likely to use counterspeech, however, no other specific variables related to the participants, such as their gender or level of education, were important in determining whether or not they used counterspeech.
In a second experiment, we test different interventions to increase the willingness to punish online hate speech. First, we test whether participants are more likely to respond to hateful messages with counterspeech when they can observe previous participants have done so. Second, we test whether informing participants of the inappropriateness of using hate prior to participating in an online platform, thus increasing the salience of the norm, would increase their willingness to sanction such behavior. We find that exposure to counterspeech by anonymous others strongly encourages participants to act against hate. However, increasing the salience of the norm does not have an effect on the number of sanctions and, if anything, participants are less willing to sanction. Because retaliation is ruled out by the experimental design, we attribute this result to participants considering that sanctioning individuals that deviated from the norm in the high saliency norm condition was pointless.
In a third experiment, we look at how the treatments affected the normative and empirical expectations of the participants regarding the use of hate speech and its punishment. We thus measure effects on beliefs at the normative and meta-normative levels to explore the mechanisms that drive the findings in experiment 2.
Our results show that sanctioning behavior is contagious: when participants could observe the previous counterspeech they were more likely to follow suit. However, this increase in punishment is linked to an increased perceived inappropriateness of hate speech while the acceptability of punishing does not change.