Where do social media and education meet? A closer look at understanding of democracy

ABSTRACT Social media presents a contradictory relationship with democracy. Once, it was regarded as a tool for democracy, providing alternative sources of information and coordinating social movements for democratization. Later it also became a tool for authoritarian regimes to control information and spread propaganda to stay in power. This mixed perception suggests that both democratic and authoritarian forces can use social media to influence public opinion. This presents a puzzle to the relationship between social media use and democratic understanding. Does social media promote or erode understanding of democracy? This study argues that the effect of social media use on understanding of democracy depends on higher education. The relationship also differs between democracies and non-democracies. Using the newest wave of the World Values Survey (wave 7, 2017–2020), this study analyses the influence of social media use on understanding of democracy in non-democracies and democracies. The findings suggest that social media use positively affects understanding of democracy in democratic countries. However, the democratic effect of social media is nullified in non-democracy unless it interacts with higher education. The findings offer implications for the relationship between social media, higher education, and understanding of democracy.


Introduction
A better understanding of democracy is crucial for promoting democratic values, support, and participation. 1 Despite widespread support for democracy, scholars warned us of a lack of informed understanding of democracy. 2People even have a misguided understanding of democracy due to propaganda from authoritarian regimes. 3Globally, we seem to stand on the border between understanding and misunderstanding of democracy.In this view, studying individuals' understanding of democracy and the factors that nurture the understanding is essential to promote a healthy democracy.
What nurtures citizens' understanding of democracy?Citizens are the core force of democracy, as articulated in Almond-Verba's civic culture model and Lipset's modernization theory.The civic culture model posits that democracy requires citizens who understand and value democracy. 4Citizens' support for democracy is also critical in the modernization theory, in which economic development produces an educated middle class with enhanced information access and democratic values 5 and obtains post-materialistic attitudes. 6Understanding of democracy precedes democratic values and support. 7Given the importance of citizens' democratic values and support to the functioning of democracy itself, 8 nurturing citizens who correctly understand democracy is the first step to a well-functioning democracy.
Then, can we promote an understanding of democracy both in democratic and non-democratic regimes?People understand democracy differently based on their cultures, countries, and, particularly, regime types. 9Authoritarian regimes often manipulate the meaning of democracy to maintain their power, which leads to a widespread misinformed understanding of democracy.Individuals in authoritarian regimes may incorrectly understand democracy by failing to recognize free and fair elections and civil liberties as the fundamental elements of democracy.Instead, individuals tend to consider their country as a democracy as long as it offers economic prosperity and social order. 10A study has shown that higher income has a positive influence on democratic understanding in democracies, while it has a negative influence in authoritarian countries because wealthier individuals are content with the current economic situation under the authoritarian regime. 11The differences in cultures, countries, and regime types seem to condition the understanding of democracy.
While the above studies have examined the factors influencing understanding of democracy, the role of other important factors, such as social media and education, and regime types are less studied.Especially only limited research has explored the impact of social media and higher education on democratic understanding in both non-democratic and democratic regimes.Hence, in this article, we explore an answer to the following question: Do social media and higher education foster an understanding of democracy in both democratic countries and authoritarian countries?Answering this question is essential because the power of social media increases as more political communication occurs in non-traditional media sources in both democracies and nondemocracies.Social media has become another vital communication outlet and information source globally.At the same time, we have not probed the role of education associated with social media use in the understanding of democracy empirically and theoretically.Education has been considered an important factor in democracy and democratization, but its theoretical mechanism in influencing information collection, processing, and understanding of democracy has rarely been studied.
In this article, we argue that social media use, higher education, and the interaction of these factors play a significant role in nurturing a correct understanding of democracy in both democratic and non-democratic regimes.Both social media use and higher education improve people's understanding of democracy, primarily when they work together.Social media allows users to acquire another source/type of information that may differ from what traditional mass media offers.Exposure to broader sources of information through social media can improve the users' understanding of democracy.
On the other hand, when the users are exposed to a specific type of manipulative or fake information, their understanding of democracy may descend.Here, education plays a critical role: particularly, higher education equips citizens to discern quality information from fake and manipulative information. 12Furthermore, university-educated citizens often self-select to access politics-related social media information. 13hus, when social media use interacts with higher education, the positive influence of both factors may increase citizens' correct understanding of democracy even further.
We test the above argument in both democratic and non-democratic regimes.For our empirical test, we draw data from World Value Survey (WVS) Wave 7 (2017-2022) across 56 countries, including 39 democracies and 17 non-democracies. 14Results show that in democracies, social media and higher education's influence on understanding of democracy is positive and significant.On the other hand, in non-democracies, the association of social media use with an understanding of democracy is conditional on higher education.These findings indicate that social media use and higher education are essential factors influencing citizens' understanding of democracy in both democracies and non-democracies.
To our knowledge, this article is the first attempt to investigate the role of social media and higher education as an interaction term in democracies and non-democracies.In this view, our article contributes to the literature on social media and democracy: we probe that the previously inconclusive influence of social media on understanding of democracy is because of its conditional effect on education.This study also contributes to the literature on informational autocracy 15 that the tool of manipulation of information through social media may only be effective for nonhigher educated people.
The rest of the article proceeds as follows.In the upcoming section, we review existing studies on understanding of democracy and how social media usage and higher education may affect the understanding of democracy.At the same time, we will outline our hypotheses.The third section introduces the research design, including descriptions of the data, measures of variables, and research methods.We will present our empirical findings in the fourth section.The conclusions and implications of this research will be discussed in the final section.

Understanding of democracy
The pivot of democracy is free and fair elections and civil liberties. 16However, individuals understand democracy in different ways across the globe and associate the concept of democracy with different characteristics. 17In general, people worldwide understand democracy in three distinctive ways: electoral procedures, civil liberty, and social benefits. 18A study focusing on European countries finds that some individuals include more attributes in their conception of democracy than others. 19Understanding of democracy varies across individuals, countries, and regime types. 20long with the diverse understanding of democracy, misunderstanding of democracy is also rampant.The existing studies find that misinformed understandings of democracy are widespread despite solid support for democracy globally. 21Shin finds that 61% of respondents of WVS across countries have misinformed understandings of democracy. 22This finding seems relevant to Kirsch and Welzel's research that authoritarian regimes refer to themselves as "democracy" to their citizens. 23Also, citizens under authoritarian regimes tend to define democracy as economic prosperity and social order instead of free and fair elections or liberty. 24Considering the possibility that the propaganda from authoritarian regimes can subvert fundamental procedural and liberal meanings of democracy, the danger we face in a misinformed understanding of democracy is even more severe.
At the same time, citizens in authoritarian regimes even seem to believe that they live in democracies.Yeung finds that, in authoritarian countries, citizens tend to perceive their country to be more democratic than it actually is because the controlled authoritarian media tells them that their country is a democracy. 25Although higher educated citizens are resistant to such authoritarian manipulation, 26 the overestimation of democracy by the non-higher educated concerns us that individuals around the world may not correctly connect democracy with free and fair elections as well as civil liberties.As Shin claims, authoritarian regimes may propagate the definition of democracy as the government for the people without elections instead of the government by the people. 27Globally, we face an imminent danger of a misinformed understanding of democracy.
Still, understanding of democracy fosters democratic values, support, and participation.For instance, some studies find that electoral and liberal understanding of democracy improves democratic support across countries around the globe. 28mong Latin American countries, individuals with a more complex understanding of democracy have greater democratic support. 29Understanding of democracy increases individuals' support for democracy in Russia. 30A greater understanding of democracy also increases both electoral and non-electoral political participation. 31nderstanding of democracy precedes democratic values, and democratic values precede a well-functioning democracy.The political science literature has long articulated the importance of democratic values among citizens to have democratic transitions and a well-functioning democracy. 32Emphasis has been on civic culture, 33 educated middle class with democratic values, 34 and post-material values. 35Since understanding of democracy precedes democratic values, 36 it naturally follows that a well-functioning democracy requires citizens who understand democracy correctly.
Then, important questions arise: What nurtures citizens' correct understanding of democracy?Can we avoid a misinformed understanding of democracy both in democratic regimes and non-democratic regimes?Compared to the abundant studies on the influence of democratic knowledge on political attitudes and participation, limited studies exist on factors influencing correct and well-informed understanding of democracy.A previous study finds that higher income positively influences democratic understanding in general but negatively influences authoritarian countries because privileged people tend to prefer the status quo. 37Regardless of regime type, rich people prefer the existing regime they live under and benefit from. 38ther studies find that exposures to democratic practice and information are critical sources of democratic understanding. 39Cho finds that citizens who live in democracies understand democracy better than those who live under authoritarian regimes because they are more exposed to democratic practice. 40Political interest also improves democratic understanding, but its effect is conditional on levels of democracy, meaning that political interest improves democratic understanding in democratic regimes but not in authoritarian regimes. 41Exposure to information from the internet has similar effects.The effect of internet use on democratic support has a conditional effect on regime types: under a highly controlling authoritarian government, the democratic effect of the internet is nullified. 42Higher education and exposure to mass media also positively influence democratic understanding. 43Kirsch and Welzel similarly find that education and greater exposure to information reduce misinformed understanding (i.e.authoritarian definition) of democracy. 44espite the above research, what remains unknown is the interactive effect of media information and education on understanding of democracy in both democracies and authoritarian regimes.Especially social media has become an important source of information around the world in the last decade, but the effect of social media use on understanding of democracy remains unclear in democratic and non-democratic regimes.At the same time, the role of education, which may maneuver individuals' information collection and understanding process in both democracies and nondemocracies in the era of social media, remains less studied.Thus, this study focuses on the role of social media, education, and the interaction between these two factors on a correct understanding of democracy.The subsequent sections examine social media and education as channels through which informational exposures occur and a correct understanding of democracy can be acquired.

Social media and understanding of democracy
The existing scholarship debates if social media is a tool for or against democracies.In authoritarian regimes, social media can be an alternative information source to foster opposition against the regime. 45In Russia, social media increased awareness of electoral fraud in the 2011 legislative election by exposing people to opposition blogs. 46he key to such attitudinal changes is exposure to political information on social media, specifically opposition voices. 47Social media also facilitates collective actions against authoritarian regimes. 48During the Arab Spring, social media facilitated the democratization protest mobilization by disseminating dissenting voices. 49In these respects, social media is a "liberation technology." 50In democracies, social media exposes individuals to cross-cutting information and diverse perspectives through networks outside one's immediate communities. 51Barberá suggests that the echo chambers of extreme views in social media are rather a numerical minority. 52n the other hand, some scholars argue that social media is a tool against democracies.For instance, authoritarian regimes can use social media for surveillance. 53uthoritarian regimes can also utilize social media to garner regime legitimacy by spreading propaganda and mobilizing regime supporters to post pro-regime content. 54Accordingly, social media is referred to as "repression technology." 55In democracies, social media can be a source of misinformation, including fake news and disinformation. 56Hunter finds that a greater prevalence of social media disinformation and polarization is associated with a reduced level of democracy. 57According to Hunter's study, a greater use of social media is generally beneficial for democracy, especially when it presents diverse perspectives and is used for electoral campaigns. 58et, Hunter shows that social media diminishes democracy scores when it is used to organize violent activities. 59Given the negative effect of social media disinformation on democracy, it is critical to identify a means to equip people against disinformation.Education may be one such factor.Allcott and Gentzkow find that people with higher levels of education were more likely to distinguish actual and fake news related to the 2016 election in the U.S. 60 Nonetheless, when it comes to democratic values and support at the individual level, much of the existing literature suggests that social media has positive effects.Chang finds that internet use increases democratic values in Taiwan. 61 Placek (2017)  finds that social media use enlarges democratic support in Central and Eastern European countries because social media increases exposure to political information, generating greater democratic engagement. 62Further, the networking function of social media facilitates exposure to diverse perspectives and information through wider social networks. 63Similarly, other studies find that social media use positively influences support for civil liberties in the U.S., 64 and internet use increases democratic support across Asian and African countries. 65The targeted use of the internet is critical in influencing political attitudes.In the case of Taiwan, the use of the internet for informational purposes increased democratic values, but not the internet use for entertainment and shopping. 66ue to the different online environments, it is important to examine democratic and authoritarian countries, carefully.Cho finds that social media use is more likely to promote democratic support when the country's level of democracy is higher because democratic information and voices are more abundant and uncensored in democratic countries. 67However, Hassan, Kendall, and Whitefield find that using social media as a political information source increases support for democracy even in non-democratic countries, such as Egypt. 68Therefore, extant literature indicates a link between social media use and democratic values/support in both non-democracies and democracies.
What about understanding of democracy?As discussed previously, understanding of democracy precedes democratic values and support.Then, is there a link between social media and understanding of democracy?Indeed, social media may also deepen people's understanding of democracy in democratic and even non-democratic regimes.Placek finds that internet use improves understanding of democracy in authoritarian countries. 69The internet can expose citizens to the information censored in traditional media outlets, such as contents posted by people in democratic countries. 70This exposure allows citizens in authoritarian regimes to gain better knowledge and understanding of democracy. 71his study builds on Placek's study, but we focus on social media use rather than internet use, which may include reading internet newspapers, watching internet TV news, and using social media altogether in the same category.Compared to simply reading internet newspapers or watching internet TV, using social media suggests more active behaviour by participating in an interpersonal format where individuals create networks and communicate their opinions or follow someone else's opinions.Our study specifically explores the role of social media in understanding of democracy in both democratic and authoritarian countries.Social media increases exposure to cross-cutting information and diverse perspectives. 72The exposure may nurture knowledge and understanding of democracy regardless of regime type.Further, previous studies suggest that exposure to opposition voices on social media increases voters' awareness of not-free elections. 73Therefore, this study examines the following two hypotheses: H1: Social media use improves understanding of democracy in democratic countries.H2: Social media use improves understanding of democracy in non-democratic countries.

Education and understanding of democracy
The role of education in democracy has been noticed for a long time, including Lipset's modernization theory and Almond-Verba's civic culture model.Nevertheless, in this study, we attempt to clarify a mechanism of education in developing democratic understanding.Education plays a significant role in building a substructure of individuals' awareness of citizenship, social skills, and civic engagement, which assist in nurturing an understanding of democracy.
Firstly, education shapes the awareness of citizenship through socialization. 74This educational socialization happens in the early years of schooling.Experiences in a formal and informal education setting nurture civic and political engagement in citizens of all ages. 75According to Mann, schools become "institutions and fortunes" of the state because "the business of schoolrooms … become identical with the interest of society." 76Education and schoolrooms promote a strong sense of national identity. 77Through education, citizens learn what they should expect from their government and what their government expects from them.This learning is a gradual process of political culture formation 78 through socialization which deepens the awareness of citizenship.
At the same time, education favours democracy relative to dictatorship by sharing values and ideals 79 which are essential to the functioning of a democratic society. 80The shared values and interests through education foster interactions between citizens, and frequent interactions with diverse groups promote democratic values. 81Particularly, education promotes interactions in heterogeneous groups and encourages participation in common actions, which cultivates deliberative democracy. 82Every individual contributes to governing in a democracy, and a liberal education is essential in making the contributions significant by moulding free humans. 83In short, in democracies, schooling cultivates a democratic experience for citizens from a young age 84 by sharing values and ideals.
Education also fosters informed democratic decision-making 85 and strongly influences political participation, skills, knowledge, voting, and so on. 86Schools are not instruments of organized social control but a means to attain literacy and knowledge that people can use to defend themselves against a powerful government. 87Particularly, higher education, such as university education, provides and trains skills for students to discuss pressing issues related to social justice and democracy.They learn to effectively accommodate myriad issues and concerns during discussions. 88Students learn to consider multiple perspectives and assess the strength of different arguments to reach an agreement.They also learn how to resolve conflict fairly when an agreement fails through discussions. 89A university is an academic democracy: the open and critical environment moulds better political citizens. 90Therefore, the democratic nature of attaining higher education can improve the understanding of democracy.
However, knowledge learned through schooling and what students understand as their political system may differ depending on the political regimes in which they live.As discussed earlier, individuals in democracies and authoritarian regimes understand their political systems differently.Still, studies show that education raises political participation in both democratic and authoritarian regimes. 91Even though education in authoritarian regimes is likely to be focused on social engineering and indoctrination of regime values and attitudes, 92 universities are more likely to maintain a democratic environment.Huang shows that higher education in authoritarian regimes focuses on costly signalling to students that their regime is efficient for social and political order rather than strictly indoctrinating regime values or attitudes. 93In higher education, a simple and strict indoctrination of political values seems inefficient even in non-democracies.
In most autocracies, university students are more likely to challenge the government and push for democratic change.In places with more educated people, there are higher levels of protest activities. 94Universities influence a greater motivation to challenge autocratic governments. 95In dictatorships, localities with more universities are likely to experience more protests and these protests are often related to the demand for democracy and human rights.As we observed in practice, most of the democratization movements in South Korea in the late 1980s and the pro-democratic protests in Hong Kong in recent years had been led by university students.Even in non-democracies, college students are exposed to an environment where different perspectives are encouraged, thus mirroring democratic institutions.Therefore, the democratic environment in higher education promotes understanding of democracy, serving as a foundation for demanding democracy.
H3: Higher education improves understanding of democracy in democratic countries.H4: Higher education improves understanding of democracy in non-democratic countries.

Interaction effect between education and social media
The discussion above suggests a positive influence of social media and higher education on understanding of democracy.Social media opens alternative and diverse information sources for citizens. 96Political participation may be restricted in autocracies in general, but social media offers an alternative opportunity. 97Higher education cultivates critical thinking skills with which individuals process diverse information and make proper decisions. 98On social media platforms and at universities, the meaning of democracy is transmitted through participation and discussion of information. 99Both social media and higher education help citizens to learn how to voice their opinions freely and fairly.Participation in discussions about political information and freely and fairly sharing individuals' political opinions can be one of the core mechanisms of boosting understanding of democracy.Hence, through social media use and acquiring higher education, individuals can improve their understanding of democracy.
Individuals' understanding of democracy can particularly be enhanced when university-educated citizens self-select to access politics-related social media information. 100Having understood the ideals of democracy through higher education, university graduates are likely to be drawn to the social media environment that emulates their university environment.They visit a social media website that provides information from diverse sources and even visit multiple sites to collect related information and to further interact with other people about the information. 101At the same time university-educated citizens are more likely to distinguish between democratic and non-democratic characteristics of the government. 102They are less likely to be vulnerable to bigotry and state propaganda on social media in both democracies and autocracies. 103Education can be an inoculation against fake news 104 and build resistance to disinformation and false claims even inside of the cyber information world. 105Hence, higher education will allow individuals to differentiate correct information from misinformation about democracy on social media platforms.
Some may be concerned self-selected social media users influence an echo chamber effect even among the tertiary educated. 106However, the highly educated are also more likely to be exposed to diverse social media sites than the less educated. 107Hence, educated social media users can detect censorship or manipulation on social media platforms and manoeuvre the media environment better than less educated social media users. 108Because of the possible positive synergy effect of social media and higher education, when university-educated citizens self-select to access politicsrelated information through social media platforms, their correct understanding of democracy can be even more increased.Therefore, we propose additional hypotheses: H5: The positive effect of social media use is stronger among higher educated people in democratic countries.
H6: The positive effect of social media use is stronger among higher educated people in nondemocratic countries.

Data and method
To test our hypotheses, we utilize World Value Survey (WVS) Wave 7. 109 Wave 7 was conducted between 2017 and 2022, and our study covers 56 countries: 39 democratic countries and 17 non-democratic countries (see Table 1S in Appendix for a country list).We retrieved regime type data, regtype, from Polity V (2018) 110 and recoded it as a binary variable, 0 indicates non-democracies, and 1 indicates democracies.Hence, in our dataset, "autocracy," "closed anocracy," and "open anocracy" are considered non-democracies ("0"), while "democracy" and "full democracy" are considered as democracies ("1").
The dependent variable is understanding of democracy, indicating the degree of correct understanding of the vital procedural and liberal aspects of democracyfree election and civil rights.The key independent variable is social media usage for political purposes, higher education attainment, and the interaction between social media usage and higher education attainment.As control variables, we consider both individualand country-level factors.
To construct a dataset for this study, we first excluded individuals who did not answer the survey questions, answered "Don't know," or chose "Not asked" responses.We excluded "Don't know" answers along with missing answers because our interest in this study is to know how correctly individuals understand democracy and how positively or negatively social media and higher education influence individuals' correct understanding of democracy.Hence, the conceptual opposition to the understanding of democracy is the misunderstanding of democracy rather than the lack of understanding.In this sense, dropping "Don't know" answers would not cause systemic problems. 111We also had to exclude Egypt because the survey question (Q217), the usage of social media for political purposes, was not asked to Egyptian respondents. 112These exclusions leave us with 23,049 individual observations from non-democracies and 50,985 observations from democracies.
In our model specification, we separately specify our models by regime type.This separated model specification is critical to control potential endogeneity problems between the measures of education and understanding of democracy by regime type, which will be explained in the following section titled Estimation Strategy.

Measuring understanding of democracy (Dependent variable)
Given the widespread misunderstandings of democracy across the world, 113 this study focuses on the most fundamental definition of democracy: procedural democracy (i.e. the presence of free and fair election) and liberal democracy (i.e. the existence of civil rights and civil liberties).By doing so, we attempt to cover the most minimalist (i.e.free and fair election) to the more in-depth meaning of democracy (i.e.liberal democracy). 114ur dependent variable, understanding of democracy, measuring the degree of correct understanding of procedural and liberal democracy, is an ordinal variable utilizing Q243 and Q246 from WVS. Q243 specifically asks about citizens' understanding of the essential procedural characteristic of democracy, "people choose their leaders in free elections," while Q246 questions the liberal democracy aspect, "civil rights protect people from state oppression." 115The survey allows the scale "where 1 means 'not at all an essential characteristic of democracy' and 10 means it definitely is 'an essential characteristic of democracy.'" 116For our study, if a respondent answers 1, the respondent understand democracy very poorly, while answering 10, understands democracy very well.
In the dataset, the respondents' average understanding of procedural democracy is 7.93 in non-democracies and 8 in democracies.At the same time, the respondents' mean understanding of liberal democracy is 7.55 in non-democracies and 7.37 in democracies (Table 1).The higher mean scores of procedural democracies indicate that respondents from both regime types perceive procedural democracy as a more important aspect of democracy than liberal democracy. 117egarding the average understanding of democracy in each country, the variation in the average understanding of democracy is larger in democratic regimes than in nondemocratic regimes (Figure 1).Among democratic countries, Malaysia scores the lowest average of 4.7 in understanding of procedural democracy and 4.8 in liberal democracy.In contrast, Germany scores the highest average of 9.5 in understanding of procedural democracy and 8.7 in liberal democracy.Among non-democracies, the lowest average is 6.7 in Thailand on the understanding of procedural democracy and 5.1 in liberal democracy, while the highest average is 9.1 in Bangladesh on the understanding of procedural democracy and 8.6 in liberal democracy.The average scores indicate that the understanding of democracy varies more widely in democratic regimes.

Measuring social media users and higher education (Independent variable)
One of the key independent variables is Social Media Use for Political Information.To investigate a relationship between social media use and understanding of democracy, we measured respondents' usage of social media for political information.The WVS question Q217 asks whether respondents use "internet and social media tools like Facebook, Twitter, etc." for "searching information about policies and political events." 118Adopting the survey question, we coded social media use for political information as an ordered variable, 0, 1, and 2, indicating "would never do," "might do," and "have done," respectively.In the dataset, the means of social media use for political information in non-democracies (0.68) and democracies (0.88) differ slightly, meaning social media usage for political information in non-democracies is lower than in democracies.The difference is not immense (Table 1).
Another key independent variable is Higher Education, especially post-secondary education, including tertiary education, bachelor, master, and doctoral degree.We retrieved WVS Q275 asking about respondents' highest education level and created a dummy variable Higher Education with 1 indicating respondents educated in postsecondary tertiary education and above and 0 designating otherwise.In our dataset, 31% of respondents in non-democracies and 33% of respondents in democracies had higher education at the time when the WVS Wave 7 was conducted (Table 1).

Control variables
Other individual-level and country-level variables are controlled to increase the accuracy of our models.To control respondents' socioeconomic aspects, respondents' age, sex, income, and urban/rural (i.e.whether the respondent is from an urban or rural area) are included.These socioeconomic variables need to be controlled because they may influence our dependent variable, understanding of democracy, or main independent variables, social media usage or education level.These individual socioeconomic variables are retrieved from WVS Wave 7. Comparing control variables between non-democracies and democracies, the mean values of age, income, and urban/rural in democracies are slightly higher than in non-democracies (Table 1).
In addition to the individual-level control variables, we include each country's GDP per capita to control for the country-level socioeconomic aspect.Considering GDP per capita is important because respondents' perception of democracy may be influenced 1.56 Note: Minimum value of "0" is available for the understanding of democracy measures, since WVS allowed the "spontaneous" value "0" as "It is against democracy" in respondents' answers, although the survey questionnaires do not have the option in printed.However, the "0" only consists of less than 1% of total responses.
by economic prosperity. 119The understanding of democracy may also be associated with internet availability or social media use. 120Hence, we additionally control Internet Censorship in each country, which may influence the availability of individuals' social media use, especially in non-democracies.We retrieved the internet censorship data from V-Dem (v2mecenefi), which was surveyed by the Digital Society Project (DSP). 121We adopted V-Dem's zero rebased measurement interval model that varies between −4 and 4, from "The government successfully blocks Internet access except to sites that are pro-government or devoid of political content" to "The government allows Internet access that is unrestricted, with the exceptions mentioned above." 122The mean scores of Internet Censorship are −1.19 in non-democracies and 0.74 in democracies, which indicates possible differences in using social media for political information between the two regime types (Table 1).

Estimation strategy
We utilize multilevel modelling for two different subgroups of regime type: nondemocracies and democracies.The subgroup analysis is appropriate for our study because our main interest is to find the role of social media and education within each regime group rather than a direct comparison between the regime groups.The separated subgroup model specification is crucial to account for potential endogeneity problems between our independent and dependent variables based on regime type.For instance, one may consider the positive effect of education on the understanding of democracy in democratic countries based on a modernization effect.In the same vein, another may concern a negative effect of education in non-democratic countries due to an indoctrination effect of education.This possible endogeneity problem by regime type can be eliminated by using a subgroup analysis, specifying non-democracy and democracy models separately.The separated subgroup analysis enables us to control for unobserved or omitted regime-type variables that may influence the relationship between our measurements of social media, higher education, and understanding of democracy.
In our subgroup model specification, we utilize a series of hierarchical linear models for non-democracies and democracies separately.This multilevel modelling (mixedeffect modelling) strategy is adopted to incorporate two different levels of influence on the understanding of democracy: the first for the individual-level linear model and the second for the country-level linear model.The individual-level model is specified as follows: where the i indicates an individual and j indicates a country.The b 0j is the average intercept and some of deviation depending on the country-level: In this modelling framework, we estimate main fixed coefficients but letting intercept vary according to countries (i.e.random effect).By allowing this random intercept effect, we incorporate country-to-country variability in individuals' understanding of democracy in addition to the individual themselves' variability concerning their understanding of democracy.
With this framework, the coefficients we are interested in are b 1j, b 2j, and b 3j in both non-democracies and democracies.A positive b 1j would indicate that social media use for political information is associated with a more correct understanding of democracy.A positive b 2j would represent that university education is related to a better understanding of democracy.Hence, positive and statistically significant b 1j and b 2j in both non-democracies and democracies would support H1 and H2 as well as H3 and H4, respectively.If b 3j also has a positive value, it would indicate that higher educated social media users are especially associated with a more correct understanding of democracy.Statistically significant b 3j would favours H5 and H6, indicating that the influence of social media use differs by education level in both non-democracies and democracies.

Main variable result
Table 2 summarizes our main results.The understanding of procedural democracy (Model 1 and Model 2) and liberal democracy (Model 3 and Model 4) have similar outcomes.Our first key variable, Social Media Use for political information, presents some interesting results.Unlike our expectations, the association of social media with understanding of democracy differs by regime type.Specifically, using social media is positively and significantly associated with citizens' understanding of democracy in democratic countries (Model 2 and Model 4), while its influence is lost in non-democratic countries (Model 1 and Model 3).In other words, social media use for political information may increase citizens' understanding of procedural and liberal democracy in democratic countries.However, its influence is statistically not significant in nondemocracies.This unexpected outcome will be explored more in the Robustness Check section.
Another important variable, Higher Education, is positively and significantly associated with understanding of democracy in both non-democracies and democracies.However, the significance level of Higher Education has weakened in the case of non-democracies: Higher education is associated with a better understanding of democracy in non-democracies at p < 0.1 level, while in democracies at p < 0.01 level (Table 2).This statistical pattern applies to both understandings of procedural and liberal democracy.Having a tertiary and above education increases citizens' correct understanding of democracy.So far, the results of statistical tests provide empirical support for H1, H3, and H4.However, H2, the influence of social media use for political information on a better understanding of democracy in non-democracies, is rejected.
The statistical significance of higher education in this finding is important, considering the previous studies that find education as an indoctrinating process in non- democracies. 123As Yeung argues, the previous studies' indoctrinating education effect was mainly based on the primary or secondary level education, not the tertiary/higher education we focus on in this study. 124Furthermore, as Huang suggests, higher education in non-democracies works as a signal of the government's high efficiency in social control and political order rather than the indoctrination of pro-regime values and attitudes. 125Hence, our finding, the positive influence of higher education on a greater understanding of democracy in both regime types, is compatible with the previous studies and confirms the democratic effect of higher education even in nondemocracies.

Interaction between social media and education
As shown in the main results section, the association of social media with citizens' understanding of democracy was not straightforward: in non-democracies, the relationship between social media and understanding of democracy lost its statistical significance, while its significance firmly holds in democracies.Does this mean that social media use for political information offers any benefit in non-democracies?To answer this question, we further explore the role of social media and higher education together (interaction term between social media and higher education) in this section.The interaction term, Social Media * Higher Education, is positively related to the understanding of both procedural and liberal democracy and is statistically significant in both non-democracies and democracies.This positive and significant interaction term indicates that the effect of social media use is conditional on higher education: the impact of social media use on the understanding of democracy differs by whether the social media user acquires higher education or not.Hence, the role of social media can still be important in non-democracies if social media users have higher education.Figure 2 illuminates the conditional effect of social media on education in both non-democracies and democracies.
Figure 2 presents important findings.First, as stated above, the impact of social media usage on understanding of democracy clearly differs by whether the respondents have higher education or not.The gap between non-higher-educated and higher-educated citizens' understanding of democracy increases with the frequency of their social media use in both non-democracies and democracies (Figure 2 and Table 3).
In non-democracies, social media usage increases the predicted understanding score of procedural democracy by 0.5% for individuals without higher education but 5% for citizens with higher education.In democracies, social media usage increases the predicted understanding score by 5% for citizens without higher education and 8% for individuals with higher education.In other words, social media's positive impact on understanding procedural democracy is larger for individuals with higher education.
The above statistical pattern continues in the understanding of liberal democracy.In non-democracies, using social media improves the predicted understanding score of liberal democracy by 1% for individuals with no higher education and 4% for individuals with higher education.In democracies, social media increases the predicted understanding score of liberal democracy by 6% for the respondents without higher education and 11% for those with higher education.Social media's positive influence on understanding liberal democracy is larger for citizens with higher education.For both procedural democracy and liberal democracy, in both non-democracies and democracies, the impact of social media usage on understanding of democracy is greatly influenced by education levels.
Second, although the general statistical pattern is similar in non-democracies and democracies, the predicted interaction effect differs by the democracy concepts and regime types.In non-democracies, when social media and higher education interact, the predicted understanding score of procedural democracy is 1.5 percent points higher than that of liberal democracy (calculated from Table 3).On the other hand, in democracies, the predicted understanding score of liberal democracy is 2.7 percent points higher than that of procedural democracy (calculated from Table 3).Although the difference may not be immense, the interaction between social media and higher education adds nuance to the understanding of democracy, depending on the regime type.
Above findings provide empirical support for H5 and H6: the effect of social media users on the understanding of democracy is higher among citizens with higher education in both non-democracies and democracies.
Besides our main variables, some of the control variables present interesting patterns.Individual-level control variable, Age, exhibits positive and statistically significant association with understanding of both procedural and liberal democracy in both regime types.Male, respondents' gender, also shows a positive and significant association with the understanding of democracy except for the procedural concept in non-democracies.Urban is negatively and significantly related to the understanding of democracy for the democratic regime types, but the significance disappears for non-democracies.The country-level control variables are statistically insignificant in most cases.
In sum, the above discussion provides empirical support for H1, H3, H4, H5, and H6.One of our hypotheses, H2, the role of social media in non-democracies, is rejected.In the next section, we conduct additional analyses to check the robustness of our findings and discuss more about the unsupported role of social media on the understanding of democracy in non-democratic regimes.

Robustness check
Our main concern is the measurement of the dependent variable, understanding of democracy.Although we adopted both procedural and liberal democracy concepts, the perception of democracy may still differ by people or countries.Hence, to check the robustness of our findings, we conducted an additional statistical analysis with different measurements of understanding of democracy, the so-called misunderstanding of democracy.
The misunderstanding of democracy is an ordinal variable collected from WVS utilizing questions Q242 and Q245. 126The survey question, Q242, states that "religious authorities ultimately interpret the laws," while Q245 comments, "the army takes over when government is incompetent." 127Respondents can choose an answer from the scale "where 1 means 'not at all an essential characteristic of democracy' and 10 means it definitely is 'an essential characteristic of democracy.'" 128As a respondent choose a higher value as an answer, he/she incorrectly understands or misunderstands democracy.Since these questions measure how correctly respondents differentiate authoritarian aspects from democracy, they are appropriate alternative measures of our dependent variable, understanding of democracy.With this measure, we observe the role of social media and higher education in reducing the misunderstanding of democracy.
If our main finding is robust, additional models with this alternative measure, misunderstanding of democracy, should show the same results as our main models, except the direction of main coefficients.Since the questionnaires state authoritarian tendencies, not democratic characteristics, the main coefficients (i.e.social media use, higher education, and the interaction term between social media use and higher education) from the additional models are expected to be negative and statistically significant.
Table 4 presents the statistical result.The result supports the general statistical pattern of our main model.Social media, higher education, and the interaction term show a negative and statistically significant relationship with misunderstanding of democracy.A few important points arise, however.Unlike our main models, the association of social media with misunderstanding of democracy is statistically significant even in non-democracies.More exposure to social media significantly reduces the risk of respondents' misunderstanding the politically overpowered religion or military forces as an essential characteristic of democracy in both regime types.
On the other hand, higher education rightly has a negative association with the misunderstanding of democracy.However, its statistical significance disappears in correctly differentiating overpowered nondemocratic religious authorities from true democracy in non-democratic regimes (Model 5).We suspect this outcome may result from some authoritarian countries with strong religious elites who are also political elites with tertiary religious education.However, further study is necessary to examine the role of higher education in clerical states or states where political elites receive tertiary religious education.Lastly, but more importantly, the interaction term between social media and higher education continues to exhibit a statistically significant and negative association with the misunderstanding of democracy in both non-democratic and democratic regimes.
In sum, the additional models support most of our main findings and seem to offer an insight that the role of social media may be more powerful in increasing awareness of what is not democracy than what is democracy.Future investigation can probe this insight.

Discussion
The above findings suggest some positive but, at the same time, worrisome aspects of contemporary democracy and democratization.First, citizens' understanding of democracy may be hindered by the socioeconomic gapacquiring higher education or not.In many studies, education level has been considered a proxy of socio-economic status because the education level closely relates to jobs and income level.Hence, our finding that social media is more helpful for higher educated individuals expands the already existing gap between those with and without higher education, even to the political aspect, understanding of democracy.
Second, information exposure through social media may generate a vulnerable group of people who can easily be victims or targets of authoritarian repression tools (i.e.authoritarian propaganda or information manipulation).Without the democratic skill sets they could have learned in tertiary education, the non-higher educated social media users must face the challenge of differentiating misinformation or fake information from authentic information in social media.On the other hand, higher-educated social media users can harness social media as an information source.With these skill sets, they can differentiate misinformation or fake information from authentic information in social media.In short, this new information source, social media, may create another democratically vulnerable group of people in non-democracies.
Lastly, this study casts our further attention to the role of education in democracy and democratization in authoritarian regimes.As Yeung and Huang suggest, the role of tertiary or higher education may differ from primary and secondary education, which focuses on indoctrinating the regime values and attitudes. 129Our study results confirm that acquiring higher education increases the understanding of democracy, even in non-democracies.This outcome invites a further investigation of the Note: Misunderstanding of democracy comes from WVS Q242 (i.e.religious authorities ultimately interpret the laws) and Q245 (i.e. the army can take of power when government is incompetent); Income takes a ten-point scale from 0 (lowest) to 10 (highest); GDP per capita has rescaled to denominate the amount in billions of dollars.Turkey is excluded from Model 7 because the Q245 was not asked; *p < 0.10, **p < 0.05, ***p < 0.01 different roles of the various levels of education in democracy instead of accepting the previous claims about the indoctrinating role of education in authoritarian regimes without questions.

Conclusion
Individuals understand democracy in various ways. 130Some even have a misinformed understanding of democracy because authoritarian regimes tell their citizens that they are democratic. 131While individual citizens generally support democracy around the world, a misinformed understanding of democracy is widespread. 132This suggests the importance of examining individuals' understanding of democracy along with factors influencing informed understanding.This study examined the influence of social media and higher education on citizens' understanding of democracy across 56 countries around the globe.The multilevel regression analyses of the survey data found that higher education has a positive relationship with the understanding of democracy in both democracies and nondemocracies.Higher educated individuals are more likely to have an informed understanding of democracy regardless of regime type.This is because tertiary education fosters democratic values and understanding through exposure to diverse perspectives and information.Tertiary education also cultivates skills to discuss societal issues and voice opinions in a constructive manner.This practice is directly relevant to the understanding of democracy.
The influence of social media differs based on regime types.In democracies, citizens who use social media for political information are more likely to have an informed understanding of democracy.Social media use increases exposure to cross-cutting information and democratic engagement.The meaning of democracy is transmitted through participation and information.Social media use also has an interactive effect with tertiary education in democracies.The positive effect of social media is stronger among the higher educated.Higher-educated social media users may selfselect into the democratic environment on the internet, having obtained democratic values through tertiary education.They might also have a greater skillset to assess information online and resist misinformation critically.
In non-democracies, the effect of social media use is conditional on higher education.Social media use improves understanding of democracy only among higher educated citizens.It does not influence the understanding of democracy among citizens without higher education.In non-democracies, a greater skill set is required to manoeuvre the social media environment accompanied by censorship and propaganda.As discussed above, authoritarian regimes can even use social media to foster authoritarian conceptions of democracy to bolster regime legitimacy by posing as a democratic regime.Still, social media serves as an alternative information source since it is a much more open media environment than traditional media sources controlled by the government.Educated and informed social media users can detect censorship and propaganda and effectively access democratic information online.This increases their informed understanding of democracy.Unfortunately, this is not the case for individuals without higher levels of education.
The findings generally provide optimistic views of social media, yet with caution.Overall, greater exposure to information through education and social media helps individuals deepen their understanding of democracy.At the same time, social media seems to further a gap based on socio-economic status.In democracies, social media is more helpful for the higher educated to acquire democratic knowledge than for the non-higher educated.This exacerbates the existing gap between those with and without higher education in regard to democratic understanding.In nondemocracies, social media improves democratic understanding only among the higher educated and not the non-higher educated.
As discussed earlier, social media can be a liberation tool or an authoritarian tool depending on how it is used.The current social media environment, filled with a large amount of information, including misinformation and propaganda, presents a challenge for its users to filter and assess information effectively.In authoritarian regimes, educated critical social media users seem to be able to harness social media as a tool to deepen democratic understanding.However, for those without such a skill set, social media may be a tool of repression.More studies are needed to examine social media's specific usage and contents as it relates to democratic understanding.Future studies can also examine what skillsets are required for people to maximize the benefits of the ever-expanding social media environment.

Notes
does not aim to find the reason for this interesting difference, one may suspect that relatively weak protection of civil rights in non-democracies makes citizens perceive its importance more critically than citizens in democracies.Particularly, citizens in competitive authoritarian regimes with elections, coded as non-democracy in our study, may have realized that procedural democracy without protected civil rights does not offer them the true benefit of democracy.

Table 1 .
Summary statistics of variables.

Table 2 .
Understanding of democracy, social media, and higher education.
Note: Understanding of Democracy comes from WVS Q 243 (i.e.people choose their leaders in free elections) and Q246 (i.e.civil rights protect people from state oppression); Income takes a ten-point scale from 0 (lowest) to 10 (highest); GDP per capita has rescaled to denominate the amount in billions of dollars.; *p < 0.10, **p < 0.05, ***p < 0.01

Table 3 .
Predicted scores of understanding of democracy.

Table 4 .
Misunderstanding of democracy, social media, higher education.