A spate of recent publications have focused on the construct of self-censorship (e.g., here, here, and here), extending and expanding the work of Andrew Hayes and his colleagues (e.g., here, here, and here). New research by Yariv Tsfati and Shira Dvir-Gvirsman adopts an alternative approach. Instead of focusing on the construct of self-censorship, they defined and conceptualized a belief in the importance of silencing others (BISO) as an individual difference variable.

Two studies are presented that validate a measure that assesses BISO. Study 1, conducted in two waves, established the reliability and stability of the BISO scale. It also differentiated BISO from other relevant and related constructs, such as political intolerance and authoritarianism. Study 2 found that those scoring higher on the BISO scale were more supportive of writing to an editor to request the removal of comments that they disagreed with from an online forum and they also were more favorable of writing directly to the authors of those comments to make it clear that the views they expressed were unacceptable.

Tsfati & Dvir-Gvirsman begin with a brief review of spiral of silence theory (Noelle-Neumann, 1974). Noelle-Neumann suggested that people tend to remain silent when their views are in opposition to the majority because they fear reprisal and social isolation as a consequence for expressing their opposing views. The spiral of silence occurs over time, as those in the minority become more reluctant to express their views, those in the majority, on the other hand, become more encouraged to speak out. It is then noted that “scholars have generally overlooked the active role individuals play in silencing others by creating the impression that certain opinions are unpopular and that expressing them entails social risk” (p. 392).

Tsfati & Dvir-Gvirsman propose an individual difference variable, the belief in the importance of silencing others (BISO), that captures the tendency for people to attempt to actively silence others. This active silencing is considered legitimate, important, and necessary because it targets beliefs that are deemed unacceptable, and even harmful.

The active silencing of others, in the political domain, is an expressive and unconventional form of political activity. The silencing of others is expressive because asking people not to express themselves involves communication with others, and it is unconventional because it departs from what is typically considered political activity in civics courses (e.g. attempts to persuade legislators to support specific policies, attempts to persuade voters to support specific candidates, donations to candidates and political action committees).

Examples of actively silencing others include: asking a neighbor to remove a lawn sign, asking someone to stop expressing their opinion during a conversation, threatening people attending a protest, and speaker disinvitation campaigns. A distinction is also made, by Tsfati & Dvir-Gvirsman between attempts to silence others and requests that others speak (more) politely.

Several potential theoretical underpinnings of BISO are then reviewed. Specifically, cognitive dissonance theory (e.g., Festinger, 1957), theories of group cohesion and group dissent (e.g., Jetten & Hornsey, 2014), groupthink pressures (e.g., Janis, 1972), Durkheim’s theories about taboos in premodern and modern societies (Durkheim, 1957), ideas about “corrective action” (Rojas, 2010), and the phenomenon of the third-person effect in communication (Davison, 1983).

Finally, Tsfati & Dvir-Gvirsman suggest their conceptualization BISO differs from a number of relevant and potentially related constructs, specifically terrorism, support for government censorship, authoritarianism, political tolerance, pluralism, and political participation. Aspects of Studies 1 and 2 allowed Tsfati & Dvir-Gvirsman to assess if BISO was empirically distinguishable from these constructs.

Study 1 (Click to expand)

Method

Study 1 occurred in two stages and was conducted in the summer of 2014. The first wave of data collection occurred in Israel, roughly six weeks after conflict broke out in Gaza between Israeli and Palestinian forces. A total of 687 respondents provided valid responses and were thus retained for analysis. A total of 485 respondents from the first wave also participated in wave 2. During wave 1 and wave 2, all respondents completed the BISO scale, as well as measures of expressive political participation, support for democratic rights, authoritarianism, and political intolerance.

Results

Upon collection of data during wave 1, an exploratory factor analysis identified 4 items with loadings below .50. Thus, a scale with 14 items was retained. A confirmatory factor analysis was performed on the 14 remaining items, in both wave 1 and wave 2. Three additional items were identified as having low loadings and were removed from the scale. The common feature that cut across all the retained items was that they all included support for taking some kind of action against the expression of negative, unacceptable views.

A second confirmatory factor analysis was then performed on the remaining 11 items. In both waves, these measurement models showed good fit with data.

Authoritarianism and political intolerance were positively correlated with BISO during wave 1 and wave 2, but analyses revealed that BISO was empirically distinguishable from both constructs. Items assessing support for government censorship were only asked in wave 2 and also positively correlated with BISO. Yet, as with authoritarianism and political intolerance, BISO was empirically distinguishable from support for government censorship. Support for democratic norms was negatively correlated with BISO during wave 1 and wave 2, but, as with other constructs assessed, BISO was empirically distinguishable. Lastly, expressive political participation was positively correlated with BISO during wave 1, but not during wave 2, and BISO was empirically distinguishable.

Finally, BISO was positively correlated with religiosity, right-wing political views, and political extremism (extreme left or right views). It was negatively correlated with education, and not significantly correlated with age or income. Males, on average, scored higher on the BISO scale than females.

Study 2 (Click to expand)

Method

Study 2 was conducted a year later, and had three objectives:

1. Replicate the confirmatory factor analysis for the BISO.
2. Test convergent validity of the BISO scale against a different indicator, specifically how people would respond to online comments.
3. Test if the BISO scale was positively associated with attempts to silence online comments considered unacceptable.

A total of 177 respondents were recruited for participation in Study 2. Each respondent completed the BISO scale and then was randomly presented one of two online articles, followed by a series of audience comments. Levels of agreement with, and possible audience reactions to each comment were assessed. Twenty comments in total were presented, 10 perceived as left-wing and 10 perceived as right-wing. Specifically, respondents, following each comment, were asked their level of agreement with the following statements:

• Publishing such a comment is inappropriate.
• One should make it clear to people expressing such views that they should shut up.
• One should react to such a comment and explain that expressing such views is unacceptable.
• One should write to the editor of the online news outlet and request that the comment be deleted.

The final three statements represented possible reactions that would silence the expression of views that the respondent not only disagreed with, but found inappropriate and unacceptable. In Study 2, they served as the indicators to which individual scores on the BISO were compared.

Results

A replication of the confirmatory factor analysis from Study 1 occurred, as the BISO scale possessed good reliability and demonstrated a good fit with the data. For the statements that the respondents disagreed with, a correlational analysis revealed the BISO was positively correlated with all four statements. Thus, the results of Study 2 provide evidence for convergent validity.

General Discussion

In two studies, Tsfati & Dvir-Gvirsman established the reliability and stability of the BISO scale, and empirically distinguished it from a number of relevant and potentially related constructs. Convergent validity was also demonstrated in Study 2, as higher BISO scores were associated with support for making it clear to those expressing such views that their views are unacceptable and they should shut up, and with a willingness to write to the editor to ask that unacceptable comments be removed. Construct validity was also established, as BISO scores were higher during active conflict in Gaza, and lower once the intensity of the conflict declined.

Interestingly in both studies, average support for silencing others was high, even though support for democratic norms was also high. Additionally, support for government censorship (in wave 2 of Study 1) was below the midpoint and lower than the average BISO score. Tsfati & Dvir-Gvirsman suggested that these findings indicate that respondents in their samples understood that free speech rights in a democracy allow people to express unacceptable views, but also that it was their responsibility, and not necessarily the government’s, to silence those views and explain why others should refrain from expressing them.

Potential future applications of the BISO scale are also discussed. One possibility is a hypothesis generated by spiral of silence theory: That BISO scores will be positively correlated with a greater perception of majority opinion, and negatively correlated with fear of isolation, as people should fear the conflict that could result from asking others to refrain from expressing their views. Another interesting question to explore is whether the connection between BISO and actual attempts at silencing are stronger in online contexts, compared to face to face conversations.

Finally, these two studies possess their limitations. First, they were both conducted in Israel and thus primarily sampled Israeli respondents. Given this, the generalizability of the BISO scale outside of an Israeli political context cannot be assumed. Second, the samples themselves were not representative of Israeli society as a whole, since ultra-Orthodox Jews were under sampled in both studies. Third, the fit statistics that established convergent and discriminant validation did not always meet the strictest cutoff criteria, a finding that was the most severe when assessing if BISO is distinct from support for government censorship. Tsfati & Dvir-Gvirsman noted that of the additional constructs assessed, research on psychometric properties for measuring support for government censorship is scant. The failure to meet strict cutoff criteria may, therefore, have resulted from measurement error in assessing support for government censorship.

Why These Findings are Important

The literature on self-censorship has recently been expanding, yet this literature has almost exclusively focused on whether certain people are more likely to self-censor than others, what contexts are likely to increase the likelihood of self-censorship, and how people react to feeling like their ability to express their views is restricted (e.g., here). The present research however, proposes that the active censorship of others’ views and opinions is also an important individual difference variable.

Importantly, Tsfati & Dvir-Gvirsman found that a belief in the importance of silencing others can coexist with support for democratic norms, such as free speech, and is not necessarily related to support for government censorship. This suggests that citizens may recognize the importance of free speech rights, and support them in most contexts, but may also consider it their personal responsibility, not the government’s, to silence certain views — and explain why others should refrain from expressing them.

The model of sociopolitical tolerance developed by Sullivan, Piereson, and Marcus (1979; 1982) described elsewhere, can help us understand when and why political intolerance may occur. The present findings fit nicely with this model, and with other literature on political tolerance and support for civil liberties, such as free speech. People tend to support democratic norms, particularly in the abstract — but, when faced with groups and/or contexts that are threatening, they may become politically intolerant and seek to silence those they perceive as objectionable.

Full Reference: Tsfati, Y. & Dvir-Gvirsman, S. (2018). Silencing fellow citizens: Conceptualization, measurement, and validation of a scale for measuring the belief in the importance of actively silencing others. International Journal of Public Opinion Research, 30, 391-419.