The Perception of Public Opinion between Likes, Comments and Media

Table of Contents

1. The Individual Perception Matters

2. What Determines the Individual Perception of Public Opinion? - An Overview
2.1 The Individual’s Own Opinion
2.1.1 Social Projection
2.1.2 Possible Intervening Variables
2.1.3 Implications in Regard to Filter Bubbles
2.2 The Spiral of Silence Applied: The Influence of Comments and Likes
2.2.1 The Spiral of Silence in the Age of the Web 2.0
2.2.2 Possible Opinion Cues From Likes and Approval Ratings
2.2.3 Reading Comments Online - Substantive and Substantial Influence or a Waste of Time?
2.3 The Persuasive Press Inference and its Implicationsp

3. Hypotheses

4. Bibliography

1. The Individual Perception Matters

The formation of public opinion is one of the an important activity of civil society. Especially for democratic political systems, public opinion plays a major role in the political process where citizens and organised groups voice their demands and hope that representatives will convert them into policies. It is so important that public opinion gets carried into the political system that democracies have institutions for this purpose for example elections or citizen participation platforms.

There are different hypotheses about public opinion but most of them acknowledge one basic relationship: public opinion influences the actions of political actors (e.g. Dvir-Gvirsman, Garrett, & Tsfati 2015). The implications for the individual (Glynn and Huge 2014, Noelle-Neumann 1974 and 1977), groups and whole political systems (e.g. Page and Shapiro 1983) have been examined in numerous works. Many scholars made attempts to explain how public opinion is formed, i.e. what factors influence it and which actors of the political system shape it (cf. for example Hoffman et al. 2007). Others, such as Page and Shapiro (1983), to name just one influential work, conducted research about the causal effects of aggregated public opinion on the macro-level. To sum it up, the bigger share of the works focused on aggregated effects. At first glance, this may seem plausible because public opinion is a macro-concept itself. But keeping Noelle-Neumann's spiral of silence (Noelle-Neumann 1974) and reactions as for example the spiral of empowerment (Lee and Chun 2016) in mind that showed how micro-processes and actions can in sum change the whole climate of opinion, one might have a closer look at individual perception. How do individuals asses what millions of people they have never met in their lives think about single issues?

One useful paper devoted to this question is the one of Gunther (1998). He starts where several scholars, among others Noelle-Neumann herself (Noelle-Neumann 1974 and 1993), have stopped: by noticing that mass media plays is a major information source for individuals trying to assess public opinion (Gunther 1998). After recognizing this, he virtually equates assessing the climate of opinion with getting information about it from mass media. Given the year of publication, this is no surprise. Gunther's findings about the persuasive press inference are still important, but interpersonal communication and mass media (Noelle-Neumann 1993) can no longer be seen as the principal sources of people’s public opinion perceptions. Nowadays, scholars also have to pay attention to a third factor: social network sites (SNS). They can be seen as a hybrid between mass media and interpersonal and group communication (e.g. Neubaum and Krämer 2016, Debatin 2008) because their users consume messages on topics of public interest together with reactions of other users in two forms: aggregated representations of other users for example as Likes and user-generated comments expressing opinions of other users (Walther and Jang 2012). Also in contrast to mass media there is no gate-keeping, so basically everyone with access to the internet can publish content for a potentially global audience (Zerback and Fawzi 2017). Because of this combination, it is highly likely that individuals make inferences about public opinion based on what they experience on SNS (Neubaum and Krämer 2016, Walther and Jang 2012). In recent years SNS have become increasingly important for politics. They pose a new mean of campaign activities and are targets of new laws as for example the Netzwerkdurchsetzungsgesetz in Germany, but have also been used for deception (cf. Hameleers and Schmuck 2017, Esser, Stepinska and Hopmann 2016). These developments and the lack of up to date evidence in the literature led me to the following research question:

What determines the individual perception of public opinion?

The remainder of the paper is structured as follows. After some remarks concerning the definition of the term public opinion I will mention single influential works for field in general, the literature of the past two decades examining the specific micro level relationship. As the paper aims to discuss implications for the daily life nowadays, a special focus lies on the effects of aggregated user opinion representations and comments on SNS embedded in the bigger frameworks of the spiral of silence (Noelle-Neumann 1974, 1977, 1993), social projection (Fields and Schuman 1976) and the persuasive press inference (Gunther 1998).

2. What Determines the Individual Perception of Public Opinion? - An Overview

At least since George Gallup, the concept of public opinion has attracted a lot of attention from many different (political) communication scholars (Keller 1999, Oberschall 2008). Already in 1937, the year after Gallups first major public success in predicting election outcomes (Keller 1999), Allport noticed in the first edition of the The Public Opinion Quarterly that the term public opinion was used by scholars, journalists and other writers meaning many different conceptions preventing a scientific approach to the concept (Allport 1937). His own definition should make public opinion measurable and recordable. Although other scholars in later works shared the idea of a manifest quantifiable concept, defining the concept remains a challenge (Childs 1939, Herbst 1991, Noelle-Neumann 1993, Donsbach and Traugott 2008). Some researchers take a more constructivist approach and argue that public opinion was not just aggregated preferences of individuals but an on-going product of conversation, embedded in the social relationships it is constructed in; so there was no single public opinion but various publics whose opinions can differ (e.g. Anstead and O'Loughlin 2015). Nevertheless Kwon et al. (2016) have identified a basic consensus: “First, public opinions are composed of responses from citizens as opposed to government organizations; second, they are publicly expressed, collective opinions as opposed to private conversations; and third, public opinions are relevant to public affairs, with potential influence on political decision-making (Herbst, 1991; Peterson & Gist, 1951; Price, 1988; Speir, 1950)” (Kwon et al. 2016, 202).

The literature of the last two decades concerning the topic mainly offers three independent variables influencing the individual perception of public opinion: the opinion of the subject, the content of the perceived media coverage and the group attachment of the subject.

2.1 The Individual’s Own Opinion

2.1.1 Social Projection

There is a great deal of empirical support (e.g. Fields and Schuman 1976, Marks and Miller 1987, Fabrigar and Krosnick 1995, Gunther and Christen 2002, Neubaum and Krämer 2016, Steinfeld, Samuel-Azran, Lev-On 2016) for the social projection hypothesis (Krueger 1998). The phenomenon is also described with varying other names such as looking-glass perception (Fields and Schuman 1976), egocentric attribution (Granberg and Brent 1983 via Joslyn and Haider-Markel 2002), or false-consensus effect (Ross, Greene, and House 1977 via Joslyn and Haider-Markel 2002). They all mean the tendency of individuals to perceive their own opinion in others.

There are several theoretical explanations for this. For example in the paper about their field study Fields and Schuman concluded that “it is likely that the tendency to see agreement with oneself in the general public is a basic way of perceiving the world in the absence of other information” (Fields and Schuman 1976, 442). They found even very small minorities perceiving of having at least half of the population on their side. They also mention the possibility that respondents simply regard their own opinions as so sensible that they are convinced that this exact opinion must be held by all other reasonable people (Fields and Schuman 1976). Marks and Miller offer another explanation for social projection in the context of public opinion. They equate agreement on public issues with social support, meaning that the individual has a motivation to perceive this exaggerated perceptions of social support because it enhances its self-esteem (Marks and Miller 1987). Joslyn and Haider-Markel attribute the effect to the cognitive accessibility bias after Tversky and Kahneman (1973), meaning that individuals show a tendency to use the most accessible information or the information that is available with the lowest cost. Based on this information, individuals make estimations about the frequency of classes or the probability of events, for example public opinion. Because availability is not only correlated with ecological frequency, but also affected by other factors, this reliance on the availability heuristic leads to systematic biases (Tversky and Kahneman 1973). As one’s own opinion is easily accessible, Joslyn and Haider-Markel argue that it is likely to influence perceptions of public opinion (Joslyn and Haider-Markel 2002).

2.1.2 Possible Intervening Variables

Several intervening factors have been discussed to possibly moderate the effect of one’s own opinion on the perception of public opinion, some also relating to social media use. For example the multivariate models of Wojcieszak and Price indicated that the positive correlation between the individual’s own opinion and false consensus gets attenuated if the individual is exposed to disagreement, may it be online or offline (Wojcieszak and Price 2009). This finding was supported in a later survey-based study later (Wojcieszak and Rojas 2011). Interestingly, the same paper could not report a similar moderating effect for discussion heterogeneity. Further a negative relationship between heterogeneous discussion networks and exposure to ideologically dissimilar news media and the phenomenon of social projection was suggested. Informational use of the internet was also associated with reduced projection (Wojcieszak and Rojas 2011). So, exposure to cross-cutting media and discussions with ideologically heterogeneous networks online and offline seem to be important moderators of social project when it comes to the perception of public opinion.

But neither the use of mass media nor the use of social networks sites does necessarily mean that one is exposed to cross-cutting media and discussions with ideologically heterogeneous networks (cf. overview by Flaxman, Goel, Rao 2016). As there is a vast amount of literature about selective exposure regarding mass media (cf. e.g. Hart et al. 2009), I will focus only on SNS, meaning the most important implications of the concept of filter bubbles (Pariser 2011 via Flaxman, Goel, Rao 2016), or echo chambers (Sunstein 2008).

2.1.3 Implications in Regard to Filter Bubbles

The concept of filter bubbles (Pariser 2011 via Flaxman, Goel, Rao 2016) describes algorithms used for SNS amplifying ideological segregation by automatically recommending content the individual is likely to agree with and suggesting to connect it with persons that have similar opinions (Pariser 2011 via Flaxman, Goel, Rao 2016). Because of this mechanism, many scholars conclude that it is unlikely to encounter cross-cutting opinions and ideologies in social media (e.g. Neubaum and Krämer 2016), but there are also arguments and even empirical evidence (e.g. Messing and Westwood 2012, Goel, Mason, and Watts 2010) for the exact opposite, namely that the use of SNS means more exposure to cross-cutting networks and diverse ideas (cf. overview by Flaxman, Goel, Rao 2016). In regard to this, more research is needed to gain greater confidence if the use of SNS rather decreases or increases the level of experienced disagreement. Also as filter bubbles and more in general the way individuals select information on SNS could amplify the projection bias (Schulz, Roessler 2012), this could be a starting point for further research. This seems especially important to me because I expect the importance of SNS for the perception of public opinion rather to ascend.

2.2 The Spiral of Silence Applied: The Influence of Comments and Likes

2.2.1 The Spiral of Silence in the Age of the Web 2.0

Another factor that has been found to influence the perception of public opinion is perceived media coverage. In her work about the spiral of silence Noelle-Neumann showed that not only public opinion itself but also the perception of it gets influenced by the media (Noelle-Neumann 1974). According to that, humans constantly observe their social environment with a quasi-statistical sense (Noelle-Neumann 1977), gauging the climate of opinion to avoid isolation. As the media offers suitable information for this process apart from the interpersonal communication which is very limited in the number of individuals taking part, people will apply their quasi-statistical sense (Noelle-Neumann 1977, 1993) searching for opinion cues (cf. Zerback, Koch and Krämer 2015, Zerback and Fawzi 2017).

2.2.2 Possible Opinion Cues from Likes and Approval Ratings

From this considerations among others Neubaum and Krämer concluded that individuals make inferences about public opinion also based on opinion-cues on social networks sites, for example on user-written comments or Likes (Walther and Jang 2012, Neubaum and Krämer 2016). They conducted an online experiment consisting of respondents receiving the stimulus of a fictitious Facebook news feed showing memes about two topics controversially discussed, which an either positive or negative main message, a high or low number of Likes and a selection of comments that were either positive, negative or balanced. One observation was that Likes had a much weaker effect on the perception of public opinion than user-written comments (Neubaum and Krämer 2016). An article by Lee and Jang supports these findings offering empirical evidence from a South Korean sample (Lee and Jang 2010). Here the stimulus were two news articles covering controversial topics, related fictitious comments of readers and approval ratings, e.g. “8 people disapproved while 48 approved” (Lee and Jang 2010, 835). The authors found only a very weak relationship between aggregated approval ratings and news readers’ assessment of news influence on others concerning the specific articles that was not present in all groups of subjects (Lee and Jang 2010). There is also empirical evidence that numerical representation of opinions on SNS such as Likes does not affect the recipients’ inferences about public opinion at all (Winter, Brückner and Krämer 2015). Winter, Brückner and Krämer report about an online experiment conducted in Germany where Facebook postings of a reputable news site and the corresponding article with systematically varying type of user comments and number of Likes were shown to respondents (Winter, Brückner and Krämer 2015). Contrary to the expectations of the authors, the number of likes was not found to influence the way how respondents evaluated the shown news story or its content (Winter, Brückner and Krämer 2015).

The explanation offered in their paper is based on a possible negativity bias: The argue that because Likes can only express agreement, they might fail to arouse attention of users (Winter, Brückner and Krämer 2015). Additionally they bring to attention that Likes may not be able to provide an overview on the percentage of proponents and opponents in the public that can be interpreted by users easily (Winter, Brückner and Krämer 2015). This is definitely a point that should be further researched in light of the introduction of more differentiated reactions on the well-known SNS Facebook (Facebook 2016) as well regarding other kinds of aggregated opinion presentations that differentiate different emotions such as Instagram. Another explanation mentioned in the paper by Winter, Brückner and Krämer is that comments representing statements by peers can be regarded as exemplars, whereas numbers of Likes constitute a less concrete statistics, leading to the phenomenon that comments can be interpreted easier; additionally, they also contain more potentially persuasive content to think about (Winter, Brückner and Krämer 2015). But they limit the explanatory power of their suggested causal relationship by mentioning the possibility that extraordinarily high numbers of Likes, for them an operationalization of extreme popularity or controversy about an initially negative impression of a source, might be an intervening variable moderating the effect (Winter, Brückner and Krämer 2015). The explanations Lee and Jang use in the discussion of their own findings heads in the same direction; they also think that the form how the opinion is represented, as exemplars or statistics, is responsible for the different effects together with the recipients’ cognitive propensity as well as the type of judgments (Lee and Jang 2010). These are interesting hypotheses definitely worth to examine further and complement with suitable empirical findings. That is why I will come back to this topic later.

[...]