One of the key tenets of life in a democracy is the ability to freely express one’s ideas without fear of government retribution (see e.g., Habermas, 1989; Hollander, 1975; see also, e.g., the Bill of Rights in the United States, Canada’s Charter of Rights and Freedoms, and France’s Declaration of the Rights of Man and of the Citizen). This protection from government retribution does not, however, mean that people living in a democracy always express their ideas openly, and there are a number of reasons why someone might elect to remain silent.
Indeed, fear of negative social consequences often outranks fear of government retribution when people are asked why they do not express their ideas and opinions (see Hyde & Ruth, 2002; Rosenberg, 1954-1955; Wyatt, Katz, Levinsohn, & Al-Haj, 1996).
Because the failure to express views openly can have negative, and at times disastrous, consequences for organizations, communities, or broader society, Hayes, Glynn, and Shanahan (2005) conceptualized a willingness to self-censor as an individual difference variable and developed a scale to measure it. Self-censorship is defined as: “The withholding of one’s true opinion from an audience perceived to disagree with that opinion” (p. 299).
This definition focuses on the estimated (in)congruence between the beliefs of an audience and those of a would-be speaker. If a person does not assume or possess knowledge about the beliefs of audience, then the failure to express their opinion would not be an act of self-censorship.
Defining self-censorship in this way helps maintain consistency between the specific construct of self-censorship and censorship more broadly. This definition conceptually distances the construct of self-censorship from related constructs such as shyness or communication apprehension. It does not, however, consider the motivations for self-censorship (besides the perception that the audience disagrees and may be hostile).
This definition also distinguishes self-censorship from conformity. The latter constitutes the expression of an opinion that is consistent with the audience, but inconsistent with one’s own opinion. Conformity can therefore be considered a form of self-censorship – the person in question is not expressing their true opinion as a result of social pressure. However, conformity is also distinct from self-censorship in that an opinion is expressed (albeit one more aligned with the perceived consensus than one’s own true beliefs). Both self-censorship and conformity can be considered examples of how one’s knowledge or perception of the beliefs and opinions of others can inhibit one’s own expression.
The construct of willingness to self-censor emerged from the literature on spiral of silence theory (Noelle-Neumann, 1974), which explains public opinion as a social process, where minority viewpoints gain or lose support as a function of how willing people are to express them. Spiral of silence research, as well as research in other areas (see e.g., Asch, 1951; Crutchfield, 1955), indicates that not everyone will silence themselves when they do not think their opinions are widely shared: there is variance in willingness to self-censor. The puzzle is to figure out who is most likely to self-censor and under which circumstances.
Method and Results
Hayes et al. developed the willingness to self-censor scale in two initial student samples, one from a private university (Dartmouth College) and one from a public university (The Ohio State University). These two samples produced an 8-item measure that was then administered to 4 new samples. Two of these samples consisted of students, from Cornell University and The Ohio State University. The other samples consisted of a random sample of adults from Ohio and a random sample of U.S. residents from across the country.
Dartmouth Sample (click to expand)
The Dartmouth sample consisted of 196 students enrolled in an Introductory Psychology course. Students responded via a survey link to a large pool of items, which included 21 initial items for the willingness to self-censor scale. Responses were measured on a 5-point agree-disagree scale. A total of 14 questions were identified as weak and removed. Questions were removed if:
- They showed little variability in responses.
- Only a small percentage of respondents made responses in a particular direction.
- An item possessed small initial communalities in a principal axis factor analysis.
- The inclusion of the item substantially lowered Cronbach’s Alpha.
Ohio State Sample (click to expand)
The Ohio State sample consisted of 323 students enrolled in an introductory-level communication course. Students responded to a large pool of items, which included the 7 items that remained from the Dartmouth sample and 4 new items. The same criteria for item removal used in the Dartmouth sample were then applied, resulting in the removal of 3 items (2 new items, and 1 item retained from the Dartmouth sample). A principal axis factor analysis yielded a single factor solution.
Crossvalidation Samples and Test-Retest Reliability (click to expand)
These 4 samples were also used to establish predictive, convergent, and discriminant validity. Hayes et al. asked respondents to complete subsets of the following measures:
- Concerns about negative interpersonal evaluations, measured by the public self-consciousness index (Fenigstein, Scheier, & Buss, 1975) and the short form of the fear of negative evaluation index (Leary, 1983).
- Shyness and social anxiety, measured by the shyness subscale of the shyness and sociability index (Cheek & Buss, 1981), the personal report of communication apprehension (McCroskey, 1970), and the social anxiety subscale of the self-consciousness scale.
- Argumentativeness, measured by the argumentativeness scale (Infante & Rancer, 1982).
- Self-esteem, measured by the Rosenberg self-esteem index (Robison et al., 1991).
As predicted by Hayes et al., relative to low scorers, those who scored higher on the measure were shyer, more apprehensive about communication, more socially anxious, more publicly self-conscious, feared negative evaluations from others to a greater extent, less argumentative, and had lower self-esteem. Yet confirmatory factor analyses revealed that willingness to self-censor is statistically distinguishable from the aforementioned constructs.
Finally, an additional sample of 66 students enrolled in communication courses at an unspecified Midwestern university completed the willingness to self-censor scale twice, with a four-week interval between administrations. Test-retest reliability was established.
The construct of willingness to self-censor offers scholars of public opinion a way to describe individual differences in opinion expression in an environment perceived as hostile and unwelcoming. Prior to the development of this scale in 2005, public opinion researchers effectively made a tacit assumption that most, if not all, people respond in the same way to the recognition that their beliefs are in the minority and not widely shared. Thus, by not considering individual differences in willingness to self-censor, large swaths of public opinion research may be confounded with a response bias. It is also conceivable that someone may score relatively high on willingness to self-censor, yet still express their views in a hostile setting because the issue is of great importance to them.
Willingness to self-censor may also be a determinant of political participation – as most, if not all forms, of political participation involve some form of public expression. Indeed, a willingness to self-censor could potentially impact a person’s willingness to discuss politics, donate to causes and candidates, and choice to vote or not.
Finally, willingness to self-censor could be conceptualized as an outcome variable, instead of a moderator. Treating the variable in this way allows public opinion researchers to investigate how such a style of communication develops into an enduring and important individual difference. How does a person’s socialization and upbringing impact their willingness to self-censor? Are there media effects on promoting or inhibiting a willingness to self-censor?
Why These Findings are Important
The willingness to self-censor scale is a brief 8-item measure that allows researchers to assess an important individual difference characteristic. This brevity is an advantage, as the scale can be administered quickly within the context of a larger survey. For instance, administering the willingness to self-censor scale within the context of Heterodox Academy’s Campus Expression Survey would provide scholars with a useful measure of a construct that is likely to predict how comfortable or reluctant students are to discuss controversial issues related to gender, race, politics, religion, or sexuality.
Full Reference: Hayes, A.F., Glynn, C.J., & Shanahan, J. (2005). Willingness to self-censor: A construct and measurement tool for public opinion research. International Journal of Public Opinion Research, 17, 298-323.