The Integration of Social Media to the Theory of Planned Behavior: A Case Study in Indonesia

Abstract

Leader and leadership are one of the important aspects in the life of a country. This study aims to predict the intention of young voters to vote for state leader elections by expanding the theory of planned behavior to the Indonesian context. Apart from the importance of the presidential election, research rarely uses the theory of planned behavior, and to the best of researchers' knowledge, there are no studies that have applied the theory of planned behavior to predict the intention to vote for the president. Therefore, this study is an attempt to fill that gap. Two hundred questionnaires were distributed using non-probability purposive sampling. Data analysis was carried out using the structural equation modeling (SEM) approach. The results showed that attitude and behavior control were positively related to voters' intention to elect presidential candidates. Furthermore, information from social media also has a positive relationship with the attitude of choosing presidential candidates. However, the results also show that subjective norms do not have a significant relationship with voters' intention. This study contributes knowledge to researchers, practitioners, and policymakers about the factors that influence youth intention to vote in Indonesia, namely, attitudes, perceived behavior control, and information from social media.

1. Introduction

Leaders and leadership are important in many aspects of social life. The stability of a country is closely related to the stability of business, politics, society, and others. The president is the highest leader of a country. In this regard, the presidential election is important because the position of the president is the head of state and head of government. Therefore, many things depend on the leadership of the president. The success or failure of the president has an impact on many social aspects such as economy, politics, culture, democracy, and others. Regarding democracy, Indonesia is often seen as a beacon of democracy and pluralism in Southeast Asia (Arifianto et al., 2019; Takahashi, 2019).

The presidential elections are important in many ways. Many say that the presidential election is a party of democracy in Indonesia (e.g., Dedi, 2019; Khodijah & Yusuf, 2019; Setyawan, 2019; Natsir & Ridha, 2018; Soebagio, 2008), but there are impacts of the presidential election. First, studies show that many presidential elections create polarization in society (Smidt, 2017; Jones, 2015; Abramowitz & Saunders, 2008). In the context of Indonesia, the year of the presidential election is always a “hot” year for the people in Indonesia. Especially since 2014, the presidential election resulted in heavy voter polarization until the presidential election in 2019. Tensions between family, friends and the community are examples of how the polarization of certain presidential voters arises in daily life. This polarization is believed to continue until the upcoming presidential election (De Jong, 2019; Warburton, 2019; Fata, 2018). Second, presidential elections affect many aspects such as economic and business performance (Harymawan et al., 2020; Evelyn & Basana, 2018; Obradovi & Tomic, 2017; Sagita, 2017; Shen et al., 2017; Chieo et al., 2014; Imelda et al., 2014), public health issues (DeJonckheere et al., 2018; Mette & Bertolini, 2018; William & Medlock, 2017), public security (Masters, 2018; Nwolise, 2007), work engagement and job performance (Beck & Shen, 2018; Petrillo, 2016), and others.

One of the main targets for political parties is young people or millennial (Wanasida et al, 2021), considering the significant number of young people. Specifically, the youth population (15–35 years old) in Indonesia accounts for 34.5 to 50 percent of the total population, which shows a significant size. Youth voters are usually perceived as being half way from enthusiasm and political apathy. Indonesian youth is often portrayed as pragmatic and less interested in politics (Irawanto, 2019; Hasyim, 2018; Morissan, 2016). It can be said that they seemed eager and curious about the election. However, this enthusiasm is not necessarily in line with the reality of political behavior. Young people are identified as voters who change their minds easily (swing voters). They are easy to find and receive information from the Internet.

2. Literature Review

This research applied one theory of consumer behavior, namely, the theory of planned behavior (TPB), in the context of the intention to elect a presidential candidate. This is because researchers in marketing and consumer behavior show that voters can be analyzed as consumers in political markets (Peng & Hackley, 2009; O’Cass & Pecotich, 2005; Newman, 1985). Not only that, but researchers have also used theories in consumer behavior to predict voting behavior since the 1970s (Newman, 1985; Luc, 2020; Wu et al., 2020). Specifically, planned behavior theory is one of the most parsimonious and widely used theories in predicting behavior (Hasbullah et al., 2014; Connee & Armitage, 1998). Furthermore, this theory is also open for the addition of other variables. Ajzen (1991, 1988) states that the expansion of the theory of planned behavior can be done in an effort to understand more deeply people’s behavior. Researchers have broadened the theory of planned behavior by adding variables that fit their research context such as habits (Xin et al., 2019; Soh et al., 2018; Foltz et al., 2016; Moons & De Pelsmacker, 2015), future behavior (Parkinson et al., 2018; Raut et al., 2018; Wang & Zhang, 2016; Patiro & Sihombing, 2014; Sommer, 2011), knowledge (Chiu et al., 2019; Setyawan et al., 2018; Maichum et al., 2017; Park et al., 2017), trust (Ha et al., 2019; Rameez & Kulathunga, 2019), emotions (Parkinson et al., 2018; Londono-Roldan et al., 2017; Moon & De Pelsmacker, 2015), perceived benefits (Ha et al., 2019; Gangwal & Bansal, 2016) environmental care (Chen & Tung, 2014), moral norms (Botetzagias et al., 2015), among others.

Despite the importance of presidential elections, research rarely applies TPB and there is no research in applying extended theories to predict intention to choose a president. Furthermore, understanding young voters is important. They are future leaders. They are at the forefront in changing the country. Their role is important in politics where they are not just voters, but active actors in politics themselves (Lestari & Arumsari, 2018; Skelton, 2010; McFarland & Thomas, 2006). Therefore, this study aims to predict the intentions of young voters in electing presidential candidates by expanding the TPB. This study applied the TPB by adding a research variable, namely, information from social media. In this study, TPB was added with information from social media as an antecedent variable of attitude. The addition of this variable is because social media is the main element in the daily lives of young people. And social media is also becoming the dominant channel in political marketing (Kardashian, 2019; Harris & Harrigan, 2015). Specifically, social media is also used as an important tool in political campaigns where they can be used for sending political campaign advertisements (Bright et al., 2019; Dasli, 2019; Weeks et al., 2017; Vonderschmitt, 2012). Further, social media is believed to shape individual choices especially in election context (Biswas et al., 2014). Several studies have shown the influence of social media in presidential and legislative elections (Sekarwulan et al., 2020; Adinugroho et al., 2019; Munzir & Zetra, 2019; Ratnamulyani & Maksudi, 2018).

3. Research Methods and Materials

This study applied a quantitative paradigm (see Figure 1). Specifically, descriptive research with the aim of testing hypotheses is the type of research carried out. Furthermore, the sampling method in this study is a nonprobability sampling method, that is, purposive sampling which aims to select respondents with certain criteria that are consistent with the study (Sekaran & Bougie, 2019). In this study, the sample used was students from private universities who had the right to vote in the 2019 Presidential Election. Young people were selected in this study given the important role of young people in nation building. Young people live with social media and are quickly influenced by information from social media. The sample size was 200 respondents. One guidelines of sample size is samples less than 100 are “small” samples, 100 to 200 are “medium,” and more than 200 are “large” (Kline, 2005). Other researchers stated that a range of sample size from 30 to 460 is appropriate (Wolf et al., 2013). Thus, the samples in this study have met the requirements of the number of samples.

Figure 1: Research Model

This study applied items from previous research to measure the research variables (Hsu et al., 2015; Ajze, 2006). To measure the answers of respondents, the questionnaire used the interval scale because this scale shows a clear distance between each scale point (scale point) that is in the respondent’s answer (Hair et al., 2007). The type of interval scale used in this study is a Likert scale. Likert scale is a method of measurement that will indicate the response of the respondents to their attitude towards an object. The measurement method using the Likert scale consists of a range of answers from “strongly agree” to “strongly disagree,” which will be answered by respondents (Hair et al., 2007). Likert scale is used because it is a method of measuring attitudes that is simple and convenient (Chyung et al., 2017).

Before testing hypotheses, reliability and validity tests are performed first to ensure the research indicators are reliable and valid (Sekaran & Bougie, 2019). The reliability test conducted in this study used the Cronbach’s alpha method. Furthermore, validity tests were conducted by using Composite Reliability and AVE to test convergent validity and Fornell-Larcker Criterion to test discriminant validity (Hair et al., 2010). Then, data were analyzed using structural equation modeling (SEM). This is because many important variables in social science cannot be observed directly (latent variables), for example, the variables of attitude, intention, motivation, and so others. These variables will then be measured by various indicators that may contain measurement errors. Structural equation modeling (SEM) has become the main tool for testing and understanding the relationship between latent variables (Deng et al., 2018; Byrne, 2010; Guo & Lee, 2007).

4. Results and Discussion

4.1. Results

A total of 200 questionnaires were returned, resulting in a response rate of 82.5%. However, two questionnaires could not be used because they were not filled out. Thus, 163 questionnaires could be used or a response rate of 81.5%. Two thirds of respondents are women and one third are men. All respondents are students from a private university.

The research instrument test was conducted by testing the reliability and validity of the indicators used in measuring each variable. Table 1 shows that the indicators used for the variables are reliable and valid. Specifically, the table shows that each construct demonstrated internal consistency with a Cronbach’s alpha value of α in the range of 0.671 to 0.855. These alpha values were described as reasonable (Taber, 2018).

Table 1: Reliability and Validity Results

Table 1 also shows that all composite reliability values were above 0.8 and AVE values above 0.6. Thus, it can be stated that convergent validity had been achieved in this study (Hair et al., 2010). Discriminant validity was assessed using the Fornell and Larcker criteria by comparing the square root of each AVE in the diagonal with the correlation coefficient (off-diagonal) for each construct in the relevant row and column. Table 2 shows the estimated correlations between constructs, and the square root of the extracted mean variance (AVE) on diagonal values. AVE is between 0.699 and 0.856 (Table 1), which is above the required value of 0.50 (Fornell & Larcker, 1981). The square root AVE, which is from 0.836 to 0.925, is higher than the value of the correlation, which is from -0.020 to 0.582. Thus, it can be stated that the discriminant validity of the research construct has been achieved.

Table 2: Discriminant Validity (Fornell-Larcker Criterion & Heterotrait-Monotrait Criterion)

Square Roots of AVE (Diagonally in Bold).

Hypothesis testing is conducted after the goodness of the data is examined by using the reliability and validity tests. Structural equation modeling is applied to test the structural relationships between variables. This research applied T-statistics and p-value to test significance coefficient hypothesis. The results of the structural equation modeling analysis are presented in Table 3.

Table 3: Result of Hypotheses Testing

4.2. Discussion

The research aims to examine the extended Theory of Planned Behavior by adding variables information from social media to predict the intention to choose a presidential candidate. This research confirms the first hypothesis that shows the relationship between information from social media and attitude toward choosing a president candidate. This result is consistent with attitude theory that information is the basis for attitude formation. Specifically, a person processes information about an object before he arrives at the object’s evaluation and forms his attitude towards the object (Jacoby et al., 2002; Lutz, 1978). This result of the relationship between information and attitude is consistent with previous findings (e.g., Utami & Rahyuda, 2019; Sihombing, 2017; Elliot & Speck, 2005).

The results of this study confirm the second hypothesis that shows a positive relationship between attitude and intention to behave. Attitude theories such as Theory of Reasoned Action, Theory of Planned Behavior, and Theory of Acceptance Model show that attitude is a predictor of intention to behave (Ijzen & Fisbein, 2005; Davis, 1989). The results of this study are also consistent with previous studies that show a positive relationship between attitude and intention to behave (e.g., Cho & Son, 2019; Nguyen et al., 2019; Natalia & Sihombing, 2018; Hussein, 2017; Patiro & Sihombing, 2016; Gunadi & Sihombing, 2015; Sihombing, 2011; Hansen & Jensen, 2007).

The results show that the third hypothesis is not supported. The unsupported hypothesis is related to the relationship between subjective norms and the intention to choose a presidential candidate. Subjective norms refer to the influence of people who are considered important (e.g., family, siblings, and friends) that will affect one’s behavior. The results of this study indicate that subjective norms do not have a significant relationship with the intention to choose a presidential candidate. This result can be explained through the current political situation in Indonesia. It can be said that everyone who has the right to vote already has made their “choice”. In other words, families of fathers and mothers and their children can have different choices. Each choice is often considered a result of political polarization. Political polarization between supporters of two presidential candidates Jokowi and Prabowo was sharper in the 2019 presidential election. This was because the battle between the two was a continuation of the previous period in 2014. The battle continued in the 2019 presidential election.

The battle between the two camps (i.e., Jokowi vs. Prabowo) created acute fanaticism, which divided the community into two camps in the 2019 presidential election. Jokowi won the presidential election in 2014 ago with a narrow margin. The difference in votes between Jokowi and Prabowo in 2014 was the thinnest among presidential elections since 1998, marking intense competition between the two camps. Prabowo’s reluctance to admit his defeat in the 2014 presidential election was considered a trigger for the continuation of political segregation during the Jokowi administration in 2014–2019 and created polarization in society (Triwibowo, 2019). It has been reported in many newspapers or other forms of media that the president’s choice is different between husband and wife or parents with children or with close friends has made family relations or friendships can heat up during the election.

The results of this study indicate that there is a positive relationship between perceived behavioral control and intention to behave. It is related to the context of the general election (president, regional head, and others) in Indonesia, which adheres to the “LUBER” principle, which stands for “Direct (L), General (U), Free (B) and Confidential (R)” (Fatayati, 2017). Specifically, “Free” means that voters are required to vote without coercion from any party. Free also reflects control from within the individual. The results of this study are also consistent with past studies that show a positive relationship between perceived behavioral control and behavioral intention (e.g., Saputra & Sihombing, 2018; Patiro & Sihombing, 2016; Sihombing, 2011; Liao et al., 2010; Wang, 2010).

5. Conclusion

This study aims to predict the voting intention in a presidential election using an extended planned behavior theory. The results show that only one hypothesis is not supported, namely, the relationship between subjective norms and the intention to choose a presidential candidate. Supported hypotheses show that attitude and behavioral control have a positive relationship with the intention to choose a presidential candidate. Information from social media also has a positive relationship with attitudes to choose presidential candidates.

There are two main limitations of the study. First, this study uses non-probability sampling by selecting student respondents at a private university. Therefore, the results do not reflect the general voting student population in Indonesia. Thus, future studies can replicate research by using young voter respondents with different characteristics, for example, young voters who are already working. Thus, the results of research using different respondents can give the same or different results, and are useful to provide an understanding of the intention to vote. Second, data for this study were collected in only one period (cross-sectional study). Therefore, this study only describes the phenomenon at one time. Future research can try to develop causality.