Predicting People’s Intention Towards Sharing Political Contents in Social Media: The Moderating Effect of Collective Opinion

The growing availability and affordability of high-speed Internet (especially using Smart phones) has increased the popularity and use of online social media and social networking sites (hereinafter, social media). Today, social media becomes an unprecedented way of providing information, comment, video, and image - content as a whole - to the community. One of the most attractive characteristics of social media is its support for user-generated content, transforming individuals from passive consumers of content to active producers [17, 26]. Social media has proved itself as strong enabler for political engagements demonstrated in recent political movements in the Middle East where political contents were consumed and shared very quickly, and by large numbers of people. However, the consequence of content sharing, especially political ones, are not always pleasant; in fact sometimes can result in ugly punishments. In social networking sites, some ‘criminals’ (read activists) may create a content that can be ‘followed’, ‘liked’ or shared by an innocent person who not necessarily supports the ideology of the content but still is vulnerable for trial. In December 2015, a man in Thailand has been arrested for ‘liking’ a doctored photo of the King and ‘sharing’ with 608 ‘friends’ on Facebook; he could be jailed for 32 years [3]. In November 2015, another woman in Thailand was charged with ‘sharing’ of a similar type of photo; she had committed suicide [21]. In Saudi Arabia the penalty is as worse as execution for them who ‘spread rumors’ (not necessarily produce it) about the government on social media and ‘cause confusion in societies’ [1]. In spite of the susceptibility of such consequences, “37 % of internet users have contributed to the creation of news, commented about it, or disseminated it via postings on social media sites like Facebook or Twitter”, at least 17 % of the same respondents have shared these news [25, p. 4].

Despite the phenomenon and consequence of sharing political news in social media is growing, knowledge about the associated factors is not well documented in literature. Hence, it is imperative that researchers explore content sharing process - one best way is examining the determinants of content sharing in social media by individuals. Moreover, prior studies claim that, especially under uncertain conditions, factors might be influenced by collective opinion. Hence, the primary objective of this study is to identify the factors that influence people’s intention to share political contents in social media; alongside, we try to understand how the roles of these antecedents are affected by collective opinion.

In order to answer the research questions, we developed a research model from literature. The model is validated by survey data using partial least square (PLS) modeling. We start this paper presenting the theoretical background and research framework, and then present data analysis techniques and discuss the results.

Looking at social media research in academia, reasonable efforts can be acknowledged that enriched the behavioral aspect of social media acceptance. However, relatively little is known about the mechanism of content sharing in social media. Yet, prior studies have identified some antecedents including perceived social attention/recognition [9, 15], means of socialization [15, 17] or information and knowledge sharing [9], entrainment gratification (perceived enjoyment/hedonic benefit) [22], and users’ prior experience [17]. In the context of political content sharing, we contend that some of the previously identified factors from generic content sharing in social media may still be relevant due to the inherent nature of participation. For example, seeking social status is relevant for political content sharing context too – a person might perceive that sharing political news in social media would support the belief that he is politically conscious, and sharing the news would enhance his social status than the people who do not. However, we recognize that some of the factors important for generic content sharing may not be applicable in our context. Factors such as perceived enjoyment, which is very important for generic content sharing might not be important for political content sharing [17]. Similarly, although prior studies documented that socializing is one of the strong drivers for using social media [22] and to participate in content sharing [17]; sharing political content, rather, may de-socialize a person with the people from opposite political mindset.

Our study incorporates both personal and content characteristics to provide an integrated model explaining people’s intention for political content sharing in social media on the following rationales: (1) altruistic motivation provides an understanding explaining people’s welfare tendency without consciously considering the benefits to own; (2) perceived social recognition is a fundamental dimension that explains why people share information with their reference groups; there are enormous political contents available over the Internet, but rational people would share the contents after (3) applying truthfulness judgment, and (4) evaluating the overall quality; and (5) people in a social media considers collective opinion as important to make a decision under uncertainty [27].

2.1 Altruistic Motivation

Altruism refers to one’s sharing behavior that promotes welfare of others without conscious regard for one’s own interest, and expectation of a return [7, 14]. Ma and Chan [20] proposed that altruism is performed voluntarily and is important for knowledge sharing particularly in social media environments where communities are formed based on common interests. They also suggest that, in online social environments, altruistic users are more likely to show their care for and offer help to others, intentionally. Other people contribute to communities because they enjoy helping others [10]. Similarly, other studies [e.g., 23] found altruism as a major incentive for sharing tourism information in online social media and claimed that, people show altruistic behavior from the mental obligation to repay the benefit they received earlier from the knowledge community. Also, Hsu and Lin [15] provided an account explaining the impact of altruism for sharing information in blogs. Proactive online activists come to know political contents from various sources and provide them to others as a one-stop source. They feel that they should share political content either to educate the community or to establish a statement. Therefore, we deduce the first hypothesis as:

H1. Altruistic motivation will have a positive influence on people’s intention for sharing information in online social media

2.2 Perceived Social Recognition

Theory of social exchange claims that, as opposed to altruism, individuals engage in social interaction and offer helps to other with the expectation to receive some forms of social rewards in exchange such as recognition, status, and respect [22]. Social recognition or reputation is a social variable that is evaluated and endorsed by other people in the society. In the current context, perceived social recognition (PSR) can be defined as the degree to which a person believes that active participation through sharing contents in social media platform would enhance his/her personal status among the other users [adapted from 15]. PSR is a highly acceptable variable in online knowledge sharing research illustrating that, people share information and knowledge because they want to be recognized (mostly informally) as an expert or aware individual than common individuals. Davenport and Prusak [7] emphasized that, PSR is an important factor in shaping behavior of the people in online information and knowledge sharing – people who share more knowledge receive higher social recognition. Thus, PSR significantly affect people’s attitude towards participation in social blogging [15]. In fact, it is the strongest factor that enhances people’s intention to share information within online investment communities [22]. Building on the above argument, we postulate that:

H2. Perceived social recognition (PSR) will have a positive influence on people’s intention for sharing information in online social media

2.3 Perceived Truthfulness of Content

Literature is very limited on perceived truthfulness (PT), particularly in social media. Generally speaking, audience of any message tries to comprehend the truthfulness of it [24] before deciding its appropriateness [8]. Bauer and Greyser [2] found that PT of advertising, in general, is one of the dominant perceptual dimensions that explains people’s reactions to it. Towards television advertising, Chan [5] claimed that PT is necessary to make people to like an ad and to attract their attention. The closest variable to perceived truthfulness is perceived credibility, which received reasonable attention in information sharing literature. For example, past studies suggest that when a credible source communicates a rumor, the believability and sharing of the rumor increase [18]. However, “credibility is related to truthfulness in that more credible information should be perceived as more truthful” [18, p. 279]. Social media contains millions of political contents and receive further hundreds every day; people are more likely to share the content, which they perceive as more truthful than be doubtable. Thus:

H3. Perceived truthfulness (PT) will have a positive influence on people’s intention for sharing information in online social media

2.4 Perceived Value of Content

Perceived value (PV) or perceived usefulness has been consistently identified as a significant variable explaining user behavior; however, relatively is less studied in online social media research. PV can be defined as people’s overall assessment about the quality and utility of an online content benefiting a given society. The utility theory stresses that – very often – customers do not buy a product or service for its own sake but they buy a ‘bundle of attributes’ which derive value [29]. For example, the value of an online content related to political issues is measured against few characteristics including informativeness, usefulness, derived value, helpfulness and so on [22]. Helkkula and Kelleher [11] claimed that the value perception process involves internal and external dialogue while it is created and co-created by customers to each other. Evaluating the value of a political content, people do not solely perceive the value to one but to include the society. In online social media, prior studies found that perceived value is a strong determinant of intention to share information [22]. Therefore, we propose that:

H4. Perceived value (PV) will have a positive influence on people’s intention for sharing information in online social media

2.5 Collective Opinion

People do not find trouble accepting a good music or rejecting a bad one; popularity of the pieces in between vary depending on whether people know the number of download the music had [27]. Prior studies strongly claim that others’ opinion influence ones’ in online environments [18]. In social media environment, people liked a same news/story more when it had many existing supporters than had only a few [26]; interestingly, people even switched their preferences when the assumed numbers are flipped. In our context, people would share a message when they are certain about the truthfulness of it; but, for a statement that is debatable or uncertain, people tend to consider collective opinion (CO) and would likely share if it already received a good number of share or ‘like’. Hence, a moderating effect of CO between perceived truthfulness and sharing intention of a message is plausible. Based on the prior work of Li and Sakamoto [18] who investigated the role of CO on perceived truthfulness and collective sharing likelihood of a statement in social media, our study investigates the moderating effect of CO on the antecedents to people’s intention for sharing political content, which is:

H5. The relationship of altruism, social recognition, perceived truthfulness, and perceived value to intention for sharing content is moderated by collective opinion; that is, the relationships are weaker under conditions of low collective opinion and stronger under conditions of higher collective opinion.

3.1 Data Collection and Validation

We approached to seven online social networking groups in Facebook and Twitter that entertain political debates and contents. Among them, three agreed to host a link inviting people to our questionnaire. A total of 188 responses were usable for data analyses. Standard demographic measures were analyzed in order to characterize the sample. The sample is well distributed between male (54 %) and female (46 %). Among them, 45 % belong to 18-25 age group followed by 21 %, 12 %, 14 %, and 8 % within 26–33, 34–41, 42–49, and 50+, respectively. Majority of the respondents (73 %) have been using Internet for at least 4 years; and 47 % use social media for 2–3 h per day followed by 38 % for one hour or less. Majority of the respondents (57 %) follow the group as a source of entertainment, 21 % consider as a consolidated source of information (21 %), 15 % are highly concerned on politics, and 7 % are active in politics. We validated the collected data in two ways: first, testing non-response bias using Kolmogorov-Smirnov method; and then checking the common method variance (CMV) using marker variable (MV) technique [19]. Considering the exploratory nature of the research partial least squares (PLS), particularly SmartPLS (version 3.2.3, www.smartpls.de), has been employed for data analyses.

3.2 Measures

This study consists six constructs, measured using multiple-items used in prior studies. All of the indicators were considered as reflective and were based on a seven point Likert–type scale, evaluated and examined from users’ perspective [11]. Specifically, altruism was measured by the instrument developed in [20, 28], perceived social recognition from [28], perceived truthfulness from [8], perceived value from [22, 29], and collective opinion from [26].

4.1 Assessment of Measurement Properties

The measurement properties of the research model were assessed with convergent and discriminant validity of the measures. The first test of discriminant validity proves that a construct is more strongly related to its own measures than with any other construct in the model. Referring to Table 1, all constructs met the acceptable criterion for AVE (>=0.5), establishing reliability of the latent variables [6]. Moreover, the square root of AVE to construct correlations proves that each construct is more highly related to its own measures than with other constructs. Then, internal consistency for each block of indicators was examined applying two measures: Cronbach’s α and composite reliability (CR) – Table 1 shows that all constructs met the 0.7 threshold value for both measures.

Chin [6] recommended that “not only each measure be strongly related to the construct it attempts to reflect, but it should not have a stronger connection with another construct” (p. 671). To address this, first, item loading of each item was calculated taking 0.6 as the minimum cut-off level [16]; following this rule, two items were discarded. Later, we compared the correlations of each item to all other constructs – Table 2 shows that no item loads higher value on other constructs than on the construct it represents, ensuring discriminant validity at item level. Then, convergent validity was examined. Although there is no set prescription, ‘the narrower the range and higher the lowest loading’ ensures convergent validity; our range of item loading (0.601-0.891) confirms no issue with convergent validity.

4.2 Assessment of the Structural Model

For assessing the structural model, the direction of path coefficient, magnitude of t-statistics, and explanatory power of the independent variables (R²) were checked. The results, summarized in Table 3, find that all four primary hypotheses are supported while our model ‘moderately’ explains 44.7 % of the variance of people’s intention to share political content in social media [13, p. 303].

4.3 Assessing the Moderating Effect

In order to examine the moderating effect of collective opinion, we used PLS product-indicator approach. In this process, for altruistic motivation for example, we developed a dummy variable consisting 16 indicators (product term between the indicators of the exogenous variable and moderator) [see 12 for detail]. To claim a moderation effect, the path coefficient between interaction term and the endogenous variable has to be significant. In our case the moderating effect of collective opinion is established except for perceived value.

Now a day, sharing political contents in online social media becomes a regular activity. It is not that people are becoming more politically concerned than before; rather they now have easy access to technologies (Web 2.0 and social networking applications) than in previous years. If the sacrifice is affordable, what one knows should be known by others. Moreover, they can now verify – to some extent – the credibility of the content. Hence, sharing political content online can be viewed as an agglomeration of the characteristics of the content itself and the sharer’s perceptions.

5.1 Effect of Personal Characteristics

The results established that, overall, people intent to share political content in social media when they perceive that the content is truthful and possesses value to a given society. Interestingly, people share contents without expecting any external rewards or returns (altruism) but, at the same time, feel a strong need for personal recognition and prestige – consistent with prior studies [e.g., 4].

Supporting the first hypothesis, our results find that altruistic motivation (AM) is one of the influencing factors that drive people to share political content in social media (β = 0.157; t = 2.275; p = 0.023). Also, the moderating effect of collective opinion between AM and Intention is supported (β = 0.561; t = 2.384; p = 0.017). Despite some researchers’ doubt, previous studies revealed that altruism ‘does exist’ and is a part of human nature and inner desire [24]. Helping others is a common tendency of most people [28]; such ‘other-regarding sentiments’ give them contentment by contributing to public goods but benefiting little to oneself. Specifically in online media, the sacrifice against fulfilling such desire is minimal; it requires relatively less effort, time, and costs than tradition media to share in. For political content in particular, a person feels mental obligation or social justice that other people in his group should know and be aware of. Generally, the contents related to corruption, incapability, conspiracy, and mockery of a government are most likely to be shared. Sharing such political content fulfill the need of forming collective outlook e.g., liking or detestation to a political entity. Such tendency is more prominent when the content receives substantial support from a collective group considering that it benefitted a numbers of people and thus would contribute to social justice, thus worthy of sharing.

As proposed in hypothesis H2, people perceive that their social recognition will be increased if they share political contents on social media; in fact, it came as the strongest antecedent (β = 0.366; t = 5.469; p = 0.000). Standing at the core of social capital literature, since the past to present, people feel a desire to be recognized. People who expect social recognition develop interpersonal relationships and are more likely to share political content with others in the communities, consistent with prior studies [e.g., 22]. Also, sharing a content would enhance prestige when the content’s merit is already established by collective opinion, assuming that the content would be further shared a number of times and would contribute to one’s prestige who share than who do not (β = 0.679; t = 2.319; p = 0.021). The outcome of this is that people do not usually share a political content, which failed to receive attention by prior readers (indicated by share/like with respect to elapsed time since the incident occurred).

5.2 Effect of Content Attributes

Every individual shares one of the two opposite opinions (e.g., for or against, true or false, believable or not believable) and share the opinion by interacting with its neighbors in a physical or virtual communication field. Assuming him/her as a rational social being, the tendency to let other people know about the content depends on the quality of the message such as credibility of it. Our third hypothesis exhibited a significant relationship between perceived truthfulness (PT) and intention to share (β = 0.142; t = 3.053; p = 0.002). People’s intention to share political content should develop if the truthfulness is appropriately managed. As a means to enhance truthfulness, the content should contain some specific details including source and authenticity of the source, if verified and verification method, and so on [8]. Our data analysis established that collective opinion (CO) would have significant moderating influence on the relationship between PT and intention (β = 0.364; t = 2.092; p = 0.037). That means, the relation between PT and intention is more prominent under the influence of CO; i.e., when CO is low PT will have less effect on intention compared to when CO is high. Higher CO enhances people’s confidence on the truthfulness of a political content trusting that less truthful contents are rejected by most people and are not disseminated.

Lastly, the fourth hypothesis posited that perceived value (PV) of the content would have a positive influence on people’s intention towards sharing political content in social media. The study revealed a significant result (β = 0.213; t = 2.806; p = 0.005), and thus accepting the H4. It is evident that political content becomes viral when they attract readership and are shared by people, and thus in turn contributes to the popularity of the content generator. This finding implies that, content developers should pay close attention on enhancing value proposition. Also, PV is a relative term and is “uniquely and phenomenologically” determined by the user [30, p. 7] based on his/her expertise or knowledge about the subject matter; therefore the moderating role of collective opinion should have least effect – it is supported by our data analysis (β = 0.279; t = 1.45; p = 0.148). Therefore, targeting the audience of political contents based on people’s socio-academic background is important for these to be shared.

5.3 Theoretical Contributions

The primary contribution of this study is investigating people’s attitude in the context of political content sharing in social media. To the best of our knowledge, past research in online social media was silent if truthfulness judgment of a message affects its sharing – our study provided support in favor of it. As the second contribution, we contemplated in a single model that, people’s attitude to share political contents not only is reflected by their inherent psychological factors (i.e., altruism motivation and social recognition) but the perceived merit of the message too (i.e., perceived truthfulness and value). Moreover, although previous studies claimed that one’s decision is influenced by others’ opinion but the nature of such impact was unclear in literature: do the opinions of others actually contribute to form one’s perception of the story? Hence, the third theoretical contribution of the current study is to investigate the effect of collective opinion in content sharing. Now it is clear that collective opinion increases truthfulness of a political content but does not change the perception on its value, also a story liked/shared by many others is expected to contribute to the welfare of others, and improve people’s reputation if further shared.

5.4 Limitations

Despite contributing new and valuable insights to social media literature, this study has some limitations. First, people’s political orientation could be a strong predictor or moderator of political content sharing in social media – it is beyond the scope of this current study and would be worthwhile to study in future. Also, the use of political discussion groups only in the survey limits the perceptions of people who share political contents in common groups, and worthy to be included. Second, few prior studies claim that perceived value is a multi-dimensional construct which may include hedonic and social value in addition to utilitarian value [11], we considered it as a first-order construct. Future study could examine the contribution of the multi-dimensional constructs related to perceived value in a hierarchical model. Third, we used self-reported survey that might result self-selection bias. Although the CMB tests did not expose any concern on the data quality, still it was not possible to claim strongly that data are free from such bias. Future research could use actual data on political content sharing and thus would be more confident to claim the effect of the variables we discussed. Fourth, future research is urged to collect more data and replicate the study in other settings such as in cultures, nature of government (democracy or monarchy), etc. Finally, Chan [5] claims that older people trust less to media content; the investigation on the effect of control variables such as age, education, and previous experience using social media would enhance our further understanding.