From echo chambers to ‘idea chambers’: Concurrent online interactions with similar and dissimilar others

Abstract

Countering conventional theory, this study found that online homophily and heterophily—connectivity with both similar and dissimilar others—are not necessarily countervailing phenomena, among representative surveys of internet users from five Arab countries: Saudi Arabia, Tunisia, Lebanon, Qatar, and UAE (N = 4,198). Respondents who said the Internet has increased their contact with politically and religiously similar people also tended to say it increased their contact with politically and religiously dissimilar people. A four-item scale measuring online political and religious homophily and heterophily was reliable (a = .754 overall; .79 among Arab nationals), and is referred to in this article as an ‘idea chamber’ index. Confirmatory Factor Analysis and inter-item correlations of scale items are two additional tests that affirm the internal consistency of the measure. Implications for research on digital communication are discussed.

Keywords

Echo chamber homophily idea chamber political efficacy public sphere

Much prior research has examined digital echo chambers, spaces where Internet users connect with only or mostly similar others (see Colleoni et al., 2014). ‘Similarity breeds connection,’ wrote McPherson et al. (2001, p. 415). The echo chamber hypothesis holds that selective exposure (Barberá et al., 2015) limits ideological diversity and contributes to polarized publics. The limited ideological spectrum of conservative political radio and cable TV was a focus of early echo chamber discussions (Jamieson and Cappella, 2008), though the concept now typically references digital confines in which Internet users are ensconced. Recent network analyses of social networks has found echo chambers increase the spread of disinformation online (Benkler et al., 2018), and how such disinformation contributes to exclusionist ideologies.

The echo chamber phenomenon is presumed to exist on a spectrum, competing with heterophily, and may be strengthened or weakened under certain conditions. Via social network analysis and machine learning, Colleoni et al. (2014) found that homophily on Twitter depends partly on political ideology; liberals on Twitter were less exposed to divergent views than were conservatives, though conservatives who followed high numbers of like-minded Twitter users exhibited greater political homophily. In an analysis of Twitter users that examined followers of 159 members of Congress and 33 policy non-profits, Boutyline and Willer (2017), however, found conservatives and those with extremist views exhibited more political homophily on Twitter than progressives or people with less extreme views.

In both studies, political homophily and heterophily are conceptualized as opposite ends of one spectrum. However, online, can people not simultaneously increase their exposure to like-minded perspectives and divergent ones? In a web-administered behavior-tracking study, Garrett (2009) found that, while opinion congruity is a motivation for online information seeking, once Internet users access a news story, they tend not to abandon it just because it contains disconsonant views. Online environments are messy, and the conceptualization of echo chambers versus public spheres does not represent reality for many internet users, who experience both homophily and heterophily online at the same time.

This study examined data from Internet users in five countries to explore whether political and religious homophily and heterophily coexist, and whether this mix of an individual’s homophily/heterophily tends to increase or decrease concurrently. The concept of ‘idea chambers,’ defined as the collection of online spaces enabling concurrent contact with similar and dissimilar others, appears more apt to describe the realities of what people experience online than the notion of echo chambers alone, which implies a high degree of homophily and avoidance of diverging viewpoints.

We examined whether people who say using the Internet has increased their contact with people who share their political and religious views would, simultaneously, say the Internet has decreased their contact with politically and religiously dissimilar others—the echo chamber hypothesis. Following extant work on digital homophily (see Sunstein, 2001), the authors of this article originally expected data to support current echo chamber assessments, but something else was observed in the data.

A new approach to assessing digital connections emerged, involving how people use the Internet for ideological ties, even incongruous ones, resulting in a scale assessing the extent to which people use the Internet, or perceive that they do, to exchange ideas—both divergent and culturally proximate ones. This study assessed the so-labeled ‘idea chamber’ phenomenon, and examined potential correlates of the construct. The study moves research beyond the discussion of echo chambers as entities that either exist or do not exist, are either common or not common, toward a conversation acknowledging that echo chambers can be present alongside online heterophily simultaneously.

From echo chambers to ‘idea chambers’

Discussions of online homophily tend to support one of two hypotheses. The echo chamber hypothesis holds that Internet users are cocooned in online spaces with similar others (Goldie et al., 2014), to the exclusion of diverse contacts. Homophily generally finds support in prior research. People evaluate others more favorably, and are more likely to maintain contact with them, if their political views are consonant (McPherson et al.). Similarly, people avoid information that differs from their worldview (Festinger, 1962).

The online public sphere hypothesis, conversely, holds that people benefit from diverse networks online (Dahlgren, 2005). The public sphere hypothesis holds that certain online spaces demonstrate diversity of people and ideas, and thus deems the echo chamber hypothesis inadequate (see Vaccari, 2013). While this latter hypothesis is more nuanced, both hypotheses tend to conceptualize online homophily on a spectrum opposite heterophily, so if homophily increases, heterophily decreases.

Many studies of political homophily reflect the dueling spectrum treatment afforded the echo chamber concept, which the titles convey: ‘Open media or echo chamber?,’ which was a content analysis and network analysis of links in comments on Facebook pages of partisan cable TV shows (Jacobson et al., 2016); ‘Cascade or echo chamber?’ (Fieldhouse et al., 2016), an agent-based study of electoral turnout; ‘From echo chamber to persuasive device?’ (Vaccari, 2013), which examined 31 qualitative interviews with operatives from the 2008 U.S. presidential election; and ‘Social media…The ultimate voter engagement tool or simply an echo chamber?’ (Harris and Harrigan, 2015), a descriptive analysis of 2010 U.K. elections. Few such studies insist the echo chamber or public sphere hypotheses are supported absolutely, and rather consider the likelihood of echo chambers and/or the existence of online public spheres among specific groups and under certain conditions (see Del Valle and Bravo, 2018), yet many studies express greater support for one of the two hypotheses.

There is also a duality of approaches in research on echo chamber and public sphere hypotheses. First, some research explores political or ideological homophily on specific networks generally. Such research, for example, seeks to answer the question of how homophilous are, say, Twitter or Instagram. Barberá et al. (2015) found, in 150 million tweets from four million Twitter users, that the echo chamber hypothesis may be overstated, at least regarding Twitter. Another study found more than half of Twitter accounts among a representative sample of U.S. residents follow users from opposing political stances (Eady et al., 2019). Second, there are many studies of online homophily among specific groups of users of social platforms, analyzing things like news sharing among self-identified Republicans and Democrats on Facebook. Jacobson et al. (2016) found evidence of political homophily in comments on Facebook pages among viewers of partisan cable TV programs.

Using social network analysis and network regression modeling, Nahon and Hemsley (2014) studied political homophily in hyperlinking on political blogs, finding evidence of echo chambers, specifically that political blog discussions are often divided off based on partisanship. In another groups-specific study, Gerber et al. (2013) found in a survey of government and planning officials in California that local governments with politically homogeneous constituents were more likely to partner in online planning networks with other governments with similarly homogeneous constituents than with municipalities with diverse constituencies. Dubois and Blank (2018), however, analyzed survey data from 2,000 adult U.K. Internet users, finding that, while a small proportion operate in echo chambers, politically interested respondents and those who consume diverse news media tended to eschew echo chambers.

Social media platforms generally aim to connect people with like-minded others. Within a given platform, though, some functions may promote heterophily, others homophily, leading to different outcomes (see Campbell, 2015), as perhaps in public posts on Instagram versus direct-messaging on that platform. Instagram as a whole may increase users’ contact with both homophilous and heterophilous others, particularly given that a key motivation to use Instagram is to show affection to others (Phua et al., 2017), rather than to promote one’s political views.

Also, inherent differences between public Twitter use and direct messaging on platforms like WhatsApp highlight structural differences. Twitter posts are public unless accounts are private, or unless specific users are blocked, and yes, some research has found varying levels of political homophily on Twitter. Closed, direct messaging, however, could cultivate comparatively less homophily. Most internet users belong to direct-messaging groups of family or coworkers (Dennis et al., 2017). You can choose your friends, as they say, but not your family and not always your coworkers. Twitter users can choose who they follow, while people have less choice regarding family and coworkers with whom they must connect.

Amichai-Hamburger and Hayat (2011) found in nationally representative surveys from 13 countries that heavy Internet users were more likely than light Internet users to say the Internet increased contact with people who share their political and religious beliefs. They used some of the same homophily questions studied here, but again, homophily was deemed spectral and opposite heterophily; that Internet use might concurrently increase contact with similar and dissimilar people was not considered. The current study may be the first to suggest that online homophily and heterophily can increase/decrease together, such that online spaces need not be strictly an echo chamber or not, but rather are part of larger idea chambers, wherein people use some Internet spaces to engage with homogeneous and heterogeneous people and thinking.

Explication of idea chambers

The notion of idea chambers is defined as: the collective online spaces that concurrently increase both an individual’s contact with like-minded people and ideas, as well as with divergent people and divergent ideas. Someone with a strong idea chamber characteristic uses the Internet to simultaneously connect with similar and dissimilar people and ideas.

The notion of idea chambers does not suggest echo chambers are nonexistent; a person with a strong idea chamber characteristic may, indeed, operate in some online spaces that resemble echo chambers, but they are also exposed to diverse perspectives and ideas in online spaces that are not echo chambers, and where heterophily exists.

The concept of idea chambers acknowledges that some online spaces are structured to encourage homophily, while others encourage, or at least permit, heterophily. Hampton et al. (2011), for example, found that using certain online digital tools was associated with greater network diversity among 2,512 U.S. residents, while use of some social media platforms contributed to reduced contact with certain groups. Twitter may facilitate more exposure to diverse political ideas than, say, Instagram does.

For this reason, media use and news use behaviors may reflect some of the diversity of online spaces. Exposure to news does not create opinion diversity in and of itself, but news reportage is one of the ways many individuals learn about diverse people and ideas, particularly as news is, indeed, newsworthy for being revelatory. Another media use factor that may be relevant is the volume of direct messages individuals send/receive that are to/from a group, rather than one individual. Frequency of getting news via social media is yet another likely relevant correlate of idea chamber scores. Time spent online with family and friends may be another. These factors are among the variables tested in the current study.

The operationalization of idea chambers in the current study specifically assessed political and religious homophily and heterophily, though we do not specify types of homophily in the idea chambers definition, as there are other kinds of homophily/heterophily subsequent studies may assess: economic, educational, ethnic, etc.

Media contexts among countries in this study

Arab-Muslim countries are often deemed one monolith, but they differ in terms of culture, linguistic dialects, religiosity, and economy (Gregorian, 2003). The countries in this study reflect this diversity. Consider the economic mix of countries in this study: Arab Gulf nationals are among the world’s richest persons, and their countries host low-income workers as well as better compensated ‘middle class’ expatriates. Tunisia and Lebanon are lower income countries.

As this study examines online homophily and heterophily, as well as media-use predictors thereof, in Arab countries, differing digital media contexts in Arab countries are considered here. In several ways, Arab countries are more digitally connected than the U.S. or European countries, and, therefore, homophilous isolation among Internet users in Arab states could be even more worrisome than in other locales. Internet penetration in many Arab countries, and not just in wealthy Arab Gulf states, is higher than that in the U.S. and many European countries, and, percentage-wise, more people in many Arab countries own smartphones than do U.S. citizens (Dennis et al., 2017; Poushter et al., 2018).

This study examined numerous social media-related correlates of idea chamber scores. While certain demographics—young, urban, well-educated individuals, many of whom were women—extensively used social media during the Arab uprisings (Howard et al., 2011), today, online social platforms are used by most Arab demographics, including older individuals. It worth highlighting, however, a gender gap in social media use in some Arab countries; in Tunisia, 49% of males use social media while only 28% of females do, and in Lebanon the gap is reversed: 67% of males say they use social media while for females it is 76% (Poushter et al.).

Research finds a positive relationship between social media use and political participation, but has not shown a clear link between social media use and political uprisings, including in the Arab region (Boulianne, 2015). Still, while research demonstrates both the popularity of social media platforms in Arab countries, and their relationships with certain kinds of activism, the links can be overstated. One study using survey data found that Arab bloggers, though renowned for their political activism, actually tend to blog for non-political reasons, and the strongest predictors of blogging in six Arab countries were heavy Internet use and using the Internet for mostly apolitical reasons (Author, 2016a).

Arab countries have fast become digitally connected. In Lebanon, 52% of nationals used smartphones in 2015 and 80% used them in 2017. Similarly, in Jordan, 51% of nationals used smartphones in 2015 but 76% were using them by 2017 (Poushter et al.). The countries examined in this study represent a technologically vibrant part of the world, and with the exception of Tunisia, the samples from the Arab countries examined here feature some of the world’s most digitally connected populations.

The sample of countries in the current study is in some ways the ideal starting place for testing a new social science construct. Because of their large expatriate worker populations, for example, Arab Gulf countries in the study have some of the world’s most diverse populations. India and Pakistan nationals outnumber Arab nationals in Qatar and the UAE (Author et al., 2020), and the countries have sizable expatriate cohorts from Europe and North America. New social science measures are very often fielded first in the U.S. or Europe, and are assumed to be valid and reliable in other parts of the world, which is often not the case (Peng et al., 1997). This study not only tested its new measure in Arabic as well as English, but in countries outside the west.

This study includes an examination of idea chamber scores across the Arab nationalities in the data set. There are significant differences in media systems and regulations of Arab countries (Rugh, 2004), so there may be significant differences in idea chamber scores as well. Also, three of the countries included in this study are locked in a regional crisis, a blockade of Qatar (imposed by Saudi Arabia, UAE, and other countries), and the ongoing debacle involves fake news and distortions (Pinnell, 2018), some of the very concerns raised in discussions about echo chambers.

Just looking at a few elements of their media systems, Arab countries’ differences become clear. Lebanon and UAE have expansive media production, especially for movies, news, and TV, while Saudi Arabia, Tunisia, and Qatar have lower levels of media output, with the exception of news and sports coverage from Qatar (Schoenbach et al., 2016). Saudi Arabia is more censorial than any of the countries in the current study, according to the Committee to Protect Journalists’ (2015) ratings of most-censored nations. The UAE also employs extensive censorship of media (Duffy, 2013), as does Qatar, while Tunisia and Lebanon restrict sensitive speech less frequently.

Hypotheses and research question

This study submits the following hypotheses and research question regarding a four-item idea chamber measure that assesses concurrent political and religious homophily and heterophily. Each of the following hypotheses derives from literature on media use and social connections, and prior to each hypothesis below there is a precis of that theoretical foundation.

Increases in one’s online contact with like-minded people and divergent people, the notion of idea chamber, necessitates Internet connectivity. Research has found that Internet use is associated with numerous positive outcomes, and while a high idea chamber score may not be pro-social in and of itself, a mix of homophily and heterophily online is preferable to homophily alone. Shields and Kane (2011) found in a survey of college students that people who heavily use the Internet tend to have more interpersonal interactions. More specifically, most Internet behaviors Hampton et al. (2011) studied were positively associated with network diversity.

H1: Internet use will be positively associated with idea chamber scores, after controlling for media use variables, demographic factors, and media-related attitudes.

H2 expands on prior research like Dubois and Blank (2018), who found that a diverse media diet was negatively associated with online echo chamber tendencies among a representative sample of U.K. adults. While news use and exposure to diverse media are not necessarily the same, the news use measure in this study consists of four items assessing diverse topics: getting news about one’s community, about one’s country, about other Arab countries, and about non-Arab countries.

News coverage frequently reports conflict (see Vliegenthart et al., 2011), and people who debate such coverage online often differ in their beliefs. News is, by definition, new information, so heavy news use should correlate positively with idea chamber scores. Garrett (2009) found that once an Internet user accesses a news story, they do not abandon the news item just because it contains divergent ideas. A primary concern about online homophily is it reduces exposure to news from diverse perspectives (see Sunstein). For this reason, analyses of online homophily often examine variables relating to news use; Jacobson et al., for example, examined homophily on Facebook pages of cable news programs. Moreover, the sameness to which discussions of online homophily refer is often political sameness and ethnic sameness, which are common pluralistic domains that news reportage elucidates. We also consider news use as a correlate of online homophily/heterophily.

H2: News use will be positively associated with idea chamber scores, after controlling for the other variables in the study.

Recent research has assessed online political efficacy (Abraham and Pane, 2014; Ognyanova and Ball-Rokeach, 2015; Velasquez and LaRose, 2015), one’s belief that the Internet can be employed for political change. Classic research found that diverse connections do little to affect political alienation (see Mutz, 1987), but the hypothesis here posits online political efficacy is positively associated with simultaneous connections with similar/dissimilar others. Sasaki (2017) found Internet use better predicted online political efficacy among less educated than among educated Internet users. The items assessing online political efficacy used in this study have been fielded in surveys by The World Internet Project (2009) for over a decade.

Respondents who say Internet use expedites political change should report higher idea chamber scores. Indeed, persons who feel the internet hastens political reform should also report increased contact with allies (similar others), as well as with persons who hold opposing views—individuals they might persuade to change positions. This hypothesis is based on prior research, including by Author (2018), which found online homophily positively associated with online political efficacy (that study only used only the homophily items in the idea chamber measure, not the heterophily items). The measure of online political efficacy in this study is external: one’s sense their Internet activities can effect change in external systems.

H3: Online political efficacy will be positively associated with idea chamber scores, after controlling for the other variables in the study.

A novel aspect of idea chambers is that people who connect with like-minded others online connect with holders of divergent viewpoints at the same time, so higher idea chamber scores should be positively correlated with tolerance for divergent ideas. In one study of survey data from Arab Gulf countries, support for censorship was negatively associated with support for preserving one’s culture (Author, 2016b); that is, support for free speech was associated with openness to cultural change. Moreover, the idea chamber index assesses connectivity with similar and dissimilar others with regard to sensitive topics, religion and politics—the very high-strung subjects free speech laws and policies are designed to protect.

H4: Respondents’ support for freedom of expression and openness to discussing politics will both be positively associated with idea chamber scores, after controlling for the other variables.

Arab and Muslim countries are often lumped together as a monolithic Middle East, though important differences exist both within and between countries (Gregorian). We might observe, then, substantial differences between idea chamber scores among the five primary Arab nationalities in the data examined here. Given the idea chamber index is a new measure, we do not advance a formal hypothesis, but ask the following research question:

RQ1: Do nationals in the five Arab countries differ in idea chamber scores?

Method

This study examined a four-item idea chambers scale, assessing online homophily and heterophily among Internet users in five countries (N = 4,198): Lebanon (n = 922), Qatar (n = 961), Saudi Arabia (n = 886), Tunisia (n = 565), and UAE (n = 865). Most residents in Arab countries are internet users, so not many respondents were excluded. Ninety-one percent or more of respondents in Lebanon, Qatar, Saudi Arabia, and UAE are internet users, and 80% in Jordan and 68% in Tunisia. Data were collected in: Lebanon, 22 February to 25 March 2017; Qatar, 7 February to 23 March; Saudi Arabia, 6 February to 23 March; Tunisia, 8–27 March; UAE, 1 February to 29 March. Survey response rates were robust: Lebanon (49%); Qatar (37%); Saudi Arabia (76%); Tunisia (87%); UAE (73%).

Sampling

This study analyzed data from Media Use in the Middle East, 2017, an annual survey by Northwestern University in Qatar, partly funded by Qatar Foundation (Dennis et al., 2017. The authors were provided with access to the data. Respondents were members of the general population ages 18 or older. The Harris poll collected the data, gathered in multistage random probability sampling of households. Surveys took 30 minutes, and were completed in Arabic or English.

Researchers started at random points in cities and rural areas across countries, and selected homes randomly. Kish grids assisted random selection of individuals within households. Random-digit dialing was employed in Qatar, utilizing sampling frames from the country’s two phone operators. Rim weighting in all countries strengthened representativeness, and factors were gender, age, nationality/citizenship.

Measurement

The idea chambers scale consisted of four semantic differential items, prompted by: ‘Has your use of the Internet increased or decreased your contact with the following groups?’ (1 = greatly decreased; 5 = greatly increased). None of these items required reverse coding.

People who share your political views

People with political views different than yours

People who share your religious beliefs

People with religious beliefs different than yours

The items are not new to this study, but come from The World Internet Project (2009; see also Amichai-Hamburger and Hayat, 2011; Cole et al., 2012). In previous studies, however, while the items that form the idea chamber measure in the current study were worded identically and used the same numerical differentials, they were not combined to form one index measuring homophily and heterophily, but were either fielded and assessed singly or combined into an index of only homophily items (see Lupáč and Sládek, 2008). This study is the first to compute the seemingly disparate items as one index.

Descriptive statistics and inter-item correlations of the scale items are in Table 1. On average, respondents are more likely to say the Internet has increased contact with people who share their political views and people who share their religious views than to say the Internet has increased contact with people who do not share their political or religious views. But the differences were not seismic: all means are between 3.1 and 3.35. Inter-item correlations between the four items were all significant, ranging .37–.56.

Table 1.

Cronbach’s alphas for idea chamber scores, and descriptive statistics and inter-item correlations of the index items.

	Alpha	N	M	SD
All respondents	.754	4,198	12.95	2.68
All Nationals	.79	2,439	13.06	2.70
All Non-Nationals	.704	1,759	12.80	2.66
By Language
Completed survey in Arabic	.764	3,239	–	–
Completed survey in English	.74	951	–	–
Qatar All	.649	961	12.82	2.50
Qatar Nationals	.625	235	12.67	2.20
Qatar Non-Nationals	.655	726	12.86	2.59
UAE All	.70	865	12.21	2.36
UAE Nationals	.612	163	12.27	1.81
UAE Non-Nationals	.71	702	12.19	2.47
Lebanon
Nationals	.786	922	13.29	2.44
Tunisia
Nationals	.78	565	11.94	2.54
Idea chamber Index Items ^a		Pearson r	M	SD
Share your political views			3.31	.91
Political views diff. than yours		.374^b	3.1	.95
Share your religious beliefs		.492^b	3.35	.86
Religious beliefs diff. than yours		.380^b	3.13	.90
Political views diff. than yours
Share your religious beliefs		.336^b
Religious beliefs diff. than yours		.561^b
Share your religious beliefs
Religious beliefs diff. than yours		.388^b

___________________________________

^aPrompt: Has your use of the Internet increased/decreased your contact with people… 1 = greatly decreased; 5 = greatly increased.

^bAll r’s have p<.001.

Confirmatory Factor Analysis (CFA) for the index was conducted in SAS. CFA found that the construct has two dimensions (for homophily/heterophily—not political/religious). This makes sense, given that the full index purports to measure homophily and heterophily simultaneously. The CFA (N = 4,198) compared three models (unidimensional; two dimensional political/religious; and two dimensional homophily/heterophily). Generally, a Comparative Fit Index (CFI) >.90 and a Tucker-Lewis Index >.95 are indicators of good model fitness. The CFA yielded CFI = .99 and TLI = .96 for the two dimensional homophily/heterophily model (χ2 = 25.37; p < .001), again, indicating goodness of fit. Both CFIs and TLIs, respectively, for the unidimensional model (.91, .73; χ² = 354.32; p < .001) and the two dimensional political/religious model (.92, .51; χ² = 325.12; p < .001) did not reach the thresholds of interest (again: >.90, and >.95).

Potential correlates of idea chamber scores

This study examined potential predictors of idea chamber index scores: 1) measures of internet use and of news use; 2) attitudes about media; 3) demographics.

Measures of internet use and news use: Hours each week using the Internet (‘During a typical week, how many hours do you spend online?’). Number of years respondent has used the Internet (‘For how many years have you been using the Internet?’). Gets news via social media/direct messaging (Index: ‘For news and information, how frequently do you use the following as a source?’ a. Social media; b. Instant/direct messaging. 1 = Never; 6 = Several times/day. Split-half reliability=.86). Frequency of news consumption (Questions created specifically for the Media Use studies). Index: ‘How frequently do you get news or news headlines about each of the following?’ a. Your community; b. This country; c. Other Arab countries; d. Countries outside the Arab region. 1 = Never; 6 = Several times/day. Cronbach’s alpha=.84). Posting messages or comments on social media/direct messaging (Index: ‘How often do you use the Internet for the following purposes?’ a. Post messages or comments on social media; b. Post messages or comments via instant/direct messaging. 1 = Never; 6 = Several times/day. Split-half reliability=.70). The authors considered including these variables separately in the regression models, though because they correlate so strongly with one another, they were combined.

News apps use (‘For news, how frequently do you use the following?’ 1 = Never; 6 = Several times/day). Newspaper use (‘For news, how frequently do you use the following?’ 1 = Never; 6 = Several times/day). Spending time with others in-person and online likely relates to one’s online homophily/heterophily. Time each week online with family (‘During a typical week, how many hours do you spend online with family?’). Time each week online with friends (same wording as for family online). Time each week in-person with family (‘During a typical week, how many hours do you spend face-to-face with family?’). Time each week in-person with friends (same wording as for family in-person). Percent of direct messages respondent sends to/from group (‘When you use instant or direct messaging, about what percent of messages do you send/receive to/from a group vs. an individual?’).

Attitudes about media: Online political efficacy (Index: ‘Do you think that by using the Internet…a. You can have more political influence; b. Public officials will care more what people like you think?’ 1 = strongly disagree; 5 = strongly agree. Split-half reliability=.80). This measure assesses external online political efficacy: the extent to which individuals feel their internet use can hasten external political change.

Tolerance for dissenting speech may also be associated with one’s contact with similar and dissimilar others. Support for free speech online (‘It is OK for people to express their ideas on the Internet even if they are unpopular.’ 1 = SD; 5 = SA). OK for news to report problems in my community (‘It’s OK for news to report about problems in my community.’ 1 = SD; 5 = SA). Comfortable speaking out about politics (‘In general, I feel comfortable saying what I think about politics.’ 1 = SD; 5 = SA). Media in this country can report news without interference (‘Media in this country can report news independently without interference from officials.’ 1 = SD; 5 = SA). Concern about online surveillance may affect one’s willingness to communicate with dissimilar others. Worry about online surveillance (Index: ‘I’m worried about governments checking what I do online’ and ‘I’m worried about companies checking what I do online.’ 1 = SD; 5 = SA. Split-half reliability=.81). Political ideology has been found to be associated with online homophily (Barberá et al. (2015)). Conservatism (‘Compared to most nationals in this country, how would you describe yourself?’ 1 = very progressive; 5 = very conservative).

Demographics: Nationality (1 = national; 0 = non-national). In Qatar and UAE, non-nationals outnumber citizens, both in census data as well as in the current data set, so we control for nationality in these countries. We also hold nationality constant in Saudi Arabia, where a large percentage of residents are non-nationals, and in Tunisia, where a smaller minority are expatriates. All respondents in Lebanon were nationals, so we don’t control for nationality there. Age (‘What is your age?’). Gender (1 = female; 0 = male). Education (‘What’s the highest level of schooling you completed?’ 1 = no formaleducation; 10 = beyond college degree – master’s, Ph.D., etc.). Income: ‘What is your total monthly family/household income?’ [Lebanon (1 ≤750,000 Lebanese lira, 12 ≥ 10,000,000 lira); Qatar (1 ≤3,000 Qatar riyals, 15 ≥75,000 riyals); Saudi Arabia (1 ≤3,000 Saudi riyals, 12 ≥45,000 riyals); Tunisia (1 ≤100 Tunisia dinars, 12 ≥5,000 dinars) UAE (1 ≤ 3,000 Emirati dirhams, 6 ≥45,000 dirhams)]. Religiosity (‘How often do you attend religious services?’ 1 = never; 9 = once/day or more).

Analyses

Researchers conducted multiple linear regression models in SPSS 25 separately for the five countries. Examined cases were Internet users only. Pairwise exclusion of cases was used. In each model, multicollinearity tolerance was set at .20, though no predictors violated that threshold. One-Way ANOVA was used to compare idea chamber scores among nationals.

Results

This study examined idea chamber scores among Internet users in five Arab countries. Though there was not a formal hypothesis posed on the internal reliability of the idea chamber index, that is the first aspect of the dependent variable reported.

Idea chamber index was internally reliable

Table 1 reports Cronbach alphas for the four-item measure by language of survey, country, and, where applicable, nationality. Reliability was acceptable (>.61) or robust (as high as .79) among subgroups in the samples. Alphas for Arabic and for English surveys were higher than .74. The scale was somewhat more reliable among nationals (.79) than non-nationals (.704). For most national and non-national populations, reliability was >.70 and in some cases reached .80, though coefficients for Qataris and Emiratis nationals were lower (>.61). Taber (2017) highlights that Cronbach’s alpha values >.70 are robust, and alphas >.60 are acceptable in some cases.

Internet use, news use were not consistent predictors of idea chamber scores

H1 said Internet use would be positively associated with idea chamber scores, after controlling for media use, media-related attitudes, and demographics. H1 was mostly unsupported. Both time spent using the Internet and length of time someone has used the Internet were not significant correlates of idea chamber scores in three of the countries. Posting social media/direct messages was associated with idea chamber scores in one country (Lebanon, positively). The models, though, predicted sizeable amounts of variance in idea chamber scores across the countries (from a high of 53% in Saudi Arabia to a low of 13% in Qatar), but the predictive validity came mostly from factors other than Internet use (Table 2).

Table 2.

Correlates of idea chamber scores (standardized betas from multiple regression models).

	Lebanon	Qatar	KSA	Tunisia	UAE
Media Use and Interpersonal Predictors
Hours each week using Internet	.016	–.021	–.105 ^a	–.037	–.097 ^b
No. of years have used Internet	.101 ^a	.094	–.031	.100	.078 ^b
Gets news via social media/DM	–.008	.031	–.038	–.038	–.130 ^a
Frequency of news consumption	.143 ^c	.055	.106 ^c	.030	–.016
Post messages or comments on SM/DM	.125 ^a	.008	.034	.126	.045
News apps use frequency	–.011	–.001	.002	–.194 ^a	.139 ^c
Newspaper use frequency	.004	.136 ^a	.060 ^b	–.123 ^b	–.076 ^b
Time each week online w/ family	–.148 ^c	–.015	.004	–.042	–.067
Time each week online w/ friends	–.069	.039	–.061	.150	.078 ^b
Time each week in-person w/ family	.088 ^a	–.026	.035	–.012	–.011
Time each week in-person w/ friends Q16	–.114 ^a	–.026	–.013	–.026	–.073 ^b
% of direct messages to/from group	–.121 ^c	.021	.244 ^c	–.035	–.181 ^c
Attitudinal Predictors
Online political efficacy Q24 1–2	.227 ^c	.283 ^c	.332^c	.195^c	.407 ^a
Support free speech online	–.067 ^b	.021	.052	–.052	.031
OK for news to report problems in my community	.070 ^b	–.023	.028	.092	.116 ^a
Comfortable speaking out abt. Politics	–.005	.046	.130 ^c	.012	.078 ^b
Media in this country can report news w/o official interference	–.069 ^b	.108 ^b	.117 ^c	–.009	.111 ^a
Worry about online surveillance	–.092 ^a	.023	.008	.130 ^b	.074
Conservatism (a=.81)	–.115^c	.024	.052	.079	.160 ^c
Demographics
National of country	–	.049	–.041	.027	.010
Age	–.050	–.022	–.029	.038	–.025
Gender (Female)	.012	.091	–.009	–.082	.027
Education	.035	.026	.083 ^a	.263 ^c	–.019
Income	.096 ^a	–.103	.108 ^c	.035	–.039
Religiosity	.070 ^b	.091 ^b	.072 ^a	–.020	–.066 ^b
Adjusted R-square (% variance explained)	.279	.131	.531	.158	.326
Total N	877	452	816	334	765

^ap < .01.

^bp < .05.

^cp < .001.

∼ approaches significance: p > .05 < .08.

Likewise, time spent online with family and friends were each predictive of idea chamber scores in only one country, and the association was negative in Lebanon. Time in-person with family and with friends was only associated with idea chamber scores in one and two countries, respectively. Time in-person with friends was negatively associated with idea chamber scores in Lebanon and UAE. The share of direct messages respondents estimate that they send to a group was negatively correlated with idea chamber scores in two countries (Lebanon, UAE), and was a strong, positive predictor of scores in Saudi Arabia.

H2 said news use would be positively associated with idea chamber scores, after controlling for numerous alternative factors, and was also mostly unsupported. While news use frequency was a positive correlate of idea chamber scores in two countries (Lebanon, Saudi Arabia), and newspaper use was significantly associated with scores in four countries (albeit negatively in Tunisia, UAE), most standardized betas for news consumption across countries were either insignificant or were negative predictors. Accessing news via social media/direct messaging significantly predicted idea chamber scores in one country (UAE), and the association was negative.

Online political efficacy was the strongest predictor of idea chamber scores

H3 said online political efficacy (OPE) would be a significant, positive predictor of idea chamber scores, after controlling for other variables. H3 was supported; OPE positively predicted idea chamber scores in all five countries, and was the single strongest predictor of scores in all countries but Tunisia. OPE is the only variable in the models both significantly associated with idea chamber scores in all countries and exhibited positive associations in each country.

H4 said support for freedom of expression and openness to discussing politics would both be positively associated with idea chamber scores. H4 was mostly unsupported. Support for free speech online and political outspokenness were modestly predictive of scores, in a few countries.

Saudis and Lebanese reported the highest idea chamber scores

RQ1 asked about differences in idea chamber scores among Arab nationalities. Saudis and Lebanese reported idea chamber scores that were significantly higher than those of other nationals. Table 3 contains means and LSD post-hoc comparisons for idea chamber scores among the five nationalities. Qataris had the next highest idea chamber scores, followed by Emiratis and Tunisians.

Table 3.

Idea chamber scores among Arab nationals, and LSD post-hoc comparisons.

	M	SD	N
Saudis	14.12	3.00	866
Lebanese	13.29	2.44	922
Qataris	12.82	2.50	961
Emiratis	12.21	2.33	865
Tunisians	11.94	2.54	565
		Mean Diff.	Sig.
Saudi Arabia
	Lebanon	.80	.000
	UAE	1.82	.000
	Tunisia	2.15	.000
	Qatar	1.42	.000
Lebanon
	UAE	1.02	.000
	Tunisia	1.36	.000
	Qatar	.62	.001
Qatar
	Tunisia	.74	.000
	UAE	.40	.124
UAE
	Tunisia	.33	.153

–

Discussion

This study proposes that online homophily and heterophily are not countervailing phenomena—that the Internet can, at once, increase one’s connections to ideologically similar and dissimilar people. At the theoretical level, this calls for reconsidering the tenets of the echo chamber hypothesis, which asserts that, given freedom of choice in online interactions, people seek similar viewpoints and avoid dissimilar ones. At the conceptual level, the present study implies that online homophily and heterophily can be construed as two dimensions of the same construct, rather than as opposite ends of a single continuum. The items in the idea chamber measure—religious and political homophily and heterophily—should not necessarily be correlated with one another, but they are, and this is the very reason the new measure counters existing conceptualizations of echo chambers.

Echo chambers may still exist, but they may be subcomponents of idea chambers, which themselves contain spaces of diversity, disagreement, and divergent views. And of course, online ideological homophily is not necessarily inherently unfortunate; even individuals of a similar ideological stance can disagree, even vehemently, on specific policies, news coverage, and language, as Vaccari et al. (2016) found in surveys of German and Italian Twitter users. As with the notion of networked individualism (Rainie and Wellman, 2012), people with high idea chamber scores likely have partial membership in multiple networks simultaneously.

Having strong idea chamber scores, that is, simultaneously communicating with convergent and divergent others, may offer a new way to conceptualize the public sphere hypothesis, which maintains the Internet is a marketplace of diversity. In that conceptualization, presumably, people interact with both similar and different individuals and lines of thought. But again, prior research has positioned the notion of the public sphere and the echo chamber as opposing concepts. The idea chamber notion allows that the idea marketplace can simultaneously offer homophily as well as diversity, similar to what Flaxman et al. (2016) found in their analysis of web browsing behaviors of 50,000 U.S.-based Internet users.

While models explained sizeable amounts of variance in idea chamber scores in several countries, three of four hypotheses were mostly unsupported; Internet use, news consumption, and attitudes supportive of free expression were not consistent predictors of idea chambers scores. Still, the current findings provide new insights as well as pathways for future research. For example, it is possible that other psychographic characteristics not exclusively related to media use may better predict idea chamber scores. Openness to new experience (McCrae, 1987), for example, may be associated with idea chamber scores, as may a concept like need for cognition (Cohen et al., 1955). More recently, Gil de Zúñiga et al. (2017) found that extroversion, agreeableness, and conscientiousness were all positively associated with use of social media, while openness negatively predicted social media use. Personality characteristics, then, may better predict idea chamber scores than internet use and other media use variables.

Internet use was not consistently predictive of idea chamber scores, countering the study’s first hypothesis, perhaps because one’s reasons for using the Internet (enter online political efficacy) are better predictors. Likewise, the hypothesis that news use would positively predict idea chamber scores was unsupported in most countries. It could be that news consumption may affect what individuals discuss online with homogeneous and heterogeneous others, but is not associated with overall online connectivity with others. That support for freedom of speech would positively predict idea chamber scores was another largely unsupported hypothesis. Again, given that online political efficacy strongly predicted idea chamber scores, the latter may be better explained by what people intend to do online (say, political mobilization) than by the latitude they believe others should have to speak freely online.

The main implication of idea chambers is that online spaces needn’t be either echo chambers or diverse public spheres; they combine to form idea chambers, wherein people use some Internet spaces to engage with diverse ideas, and they simultaneously may abscond in other, more homophilous spaces. What types of interactions, individuals, topics, platforms, tasks or experiences lend themselves to the formation of idea chambers can be investigated in future studies. We contend that idea chambers can be measured using a simple index that was internally valid across large datasets of nationally representative samples from five countries, regarding two domains: politics and religion.

Whether or not future research employs the exact four item-scale tested in this study, the key finding is that online homophily and heterophily need not be studied as opponents. Internet use can simultaneously increase contact with politically and religiously proximate and divergent people.

The scale is internally reliable in English and Arabic, two of the world’s most-spoken languages. While Internet use and news consumption were not consistent predictors of idea chamber scores, the totality of predictors explain sizeable variance in three countries—a quarter or more of the variance in two countries and more than half of the variance in Saudi Arabia. A questionnaire could be constructed for future surveys to examine potential correlates of idea chamber scores not tapped here: attitudes on tolerance, religious affiliation, online gaming, online porn use, and online gambling, to name a possible few.

The positive end of the idea chamber scale measures online connectivity with other people, both ideologically proximate and not. However, low idea chamber scores may indicate individuals who use the Internet less to connect with others, and more to perform mundane tasks like, say, shopping and appointment scheduling. Future studies of concurrent online homophily and heterophily might examine the extent to which people use the Internet for thematic, ideological exchange versus practical tasks. Future work on idea chambers may also dovetail with research on media repertoires, or ‘how media users combine different media contacts into a comprehensive pattern of exposure,’ (Hasebrink and Popp, 2006), as idea chambers, not unlike media repertoires, are conceptualized as the collective spaces consisting of dissimilar and similar communication contacts.

Online political efficacy, one’s sense that the Internet is useful for political influence, was the strongest predictor of idea chamber scores, and was positively associated with scores in all countries. OPE was a measure of online external political efficacy, and so it makes sense that respondents who feel the internet can help them enact political change would have increased interactions with both heterogeneous (whom they might persuade of their political views) and homogenous others (their political allies). Perhaps hearteningly, individuals who are likely to use the Internet to effect political change have higher idea chamber scores, as opposed to being siloed political partisans. Political efficacy generally involves a sense of self-confidence (Balch, 1974), as one believes their political actions matter. OPE, too, is one’s sense that they can use the Internet to prompt political change. Perhaps individuals who feel they can invoke political change online feel confident enough to maintain contact with ideologically divergent people.

Three countries in this study are embroiled in a regional crisis; the Saudi and UAE-led blockade of Qatar, which was predicated upon fake news that nearly led to war (DeYoung and Nakashima, 2017). Saudi Arabia planned to invade Qatar and overthrow its ruler but was stopped by the then-U.S. Secretary of State (Emmons, 2018). The fear of echo chambers and of fake news that can flow within them, then, may be as heightened in the Arab Gulf as anywhere. In this context, some findings in this study may offer sanguinity, as idea chamber scores were highest among Saudis, and, after Lebanese, Qataris reported the next highest.

It may surprise some readers that Saudis reported the highest idea chamber scores, which defies stereotypes that presume Saudis are homogenous, isolated, and lack cosmopolitanism. Since 2013, the Media Use in the Middle East surveys have often found that, despite such stereotypes, Saudis often report the most progressive attitudes on measures such as support for free expression and concerns about online surveillance (see Dennis et al., 2019). Saudi Arabia has trended online in recent years for its war on Yemen, the murder of journalist Jamal Khashoggi, and its imprisonment of women’s rights advocates (Hawley, 2020). Many observers outside Saudi Arabia have been critical of the regime in Riyadh, while Saudis themselves frequently defend the country on internet platforms. Internet use could, then, simultaneously connect Saudis to compatriots sticking up for their country as well as the country’s detractors. Tunisians reported the lowest idea chamber scores, which may relate to Tunisia having the lowest internet penetration among countries in data used in this study; Tunisians may use the internet less for political and religious discussions than other Arab nationals, and more for quotidian transactions and communication.

For Lebanese, who arguably live in one of the most politically and religiously diverse Arab countries (Ghattas, 2017), high levels of online contact with similar and dissimilar people could be an extension of interactions Lebanese have with others in daily life. Qatar has the smallest national population of countries in this study, and its nationals put a high premium on privacy (Fromherz, 2012), so low idea chamber scores among Qataris may reflect that they communicate less online generally with other people, whether those individuals have similar religious or political views or not.

Limitations

While the idea chamber index was reliable in all countries, the study’s hypotheses were mostly unsupported, with the exception of in Lebanon, and, at times, in Saudi Arabia. Given the hypotheses were supported in Lebanon, we might expect that media use predicts idea chamber scores in countries with more open media and political systems, but the absence of support for the hypotheses in Tunisia, and support for hypotheses in Saudi Arabia, challenges that notion. Indeed, Tunisia has the most open media and political system not only among Arab countries in this study, but among all Arab countries (Masri, 2017).

A countries’ openness, then, might not help explain the relationship between media use and contact with similar/dissimilar others, and, again, perhaps individual characteristics like extroversion and openness will be found explanatory. Still, there may be something unique to Lebanon—its confessionalist religio-political system, for example, in which Lebanese have exposure to more religious diversity than people in many other Arab countries—that renders the relationship between media use and idea chamber scores stronger.

While respondents’ online homophily and heterophily tend to increase or decrease together, their actual online behavior may differ from their perceptions. Another limitation of the self-report data analyzed in this study is that just because someone says the Internet increases their contact with people who share and do not share their political and religious views does not mean that the Internet has equally increased the contact with people in these groups. Subsequent research into idea chambers might examine digital networks of Internet users who report different idea chamber scores.

Ultimately, that Internet use does not seem to increase contact with like-minded people while decreasing exposure to diverse viewpoints, is a good thing. The idea chamber index is, however, broad, and examines political and religious homophily/heterophily, to the exclusion of ethnic or socioeconomic homophily. The measure is also not platform specific. Future research may employ the idea chamber scale by asking respondents the extent to which their use of specific platforms, like, say, WhatsApp or Snapchat, promotes simultaneous homophily and heterophily, as some online platforms and spaces surely resemble an idea chamber more than others. But evidence here that the Internet for many people is not a homophilous black hole, is encouraging. Future research using the idea chamber index may also benefit from revising question wording, which asks ‘Has your use of the Internet increased or decreased…,’ by changing ‘Has’ to ‘Does.’ Current language suggests each internet user was once a non-user, which may not be the case for digital natives.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Justin D Martin

References

Abraham

Pane

(2014) The mediating role of social competition identity management strategy in the predictive relationship between susceptibility to social influence, internet privacy concern, and online political efficacy. In: Information and Communication Technology – Eur Asia conference, Bali, Indonesia, 14?17 April 2014, pp. 492?499. Berlin, Germany: Springer.

Amichai-Hamburger

Hayat

(2011) The impact of the internet on the social lives of users: A representative sample from 13 countries. Computers in Human Behavior 27(1): 585–589.

Balch

(1974) Multiple indicators in survey research: The concept “sense of political efficacy”. Political Methodology 1: 1–43.

Barberá

Jost

Nagler

, et al. (2015) Tweeting from left to right: Is online political communication more than an echo chamber? Psychological Science 26(10): 1531–1542.

Benkler

Faris

Roberts

(2018) Network Propaganda: Manipulation, Disinformation, and Radicalization in American Politics. New York, NY: Oxford University Press.

Boulianne

(2015) Social media use and participation: A meta-analysis of current research. Information, Communication & Society 18(5): 524–538.

Boutyline

Willer

(2017) The social structure of political echo chambers: Variation in ideological homophily in online networks. Political Psychology 38(3): 551–569.

Campbell

(2015) Mobile communication and network privatism: A literature review of the implications for diverse, weak, and new ties. Review of Communication Research 3: 1–21.

Cohen

Stotland

Wolfe

(1955) An experimental investigation of need for cognition. The Journal of Abnormal and Social Psychology 51(2): 291–294.

10.

Cole

Suman

Schramm

, et al. (2012). The World Internet Project. World Internet Project. Los Angeles, CA: University of Southern California.

11.

Colleoni

Rozza

Arvidsson

(2014) Echo chamber or public sphere? Predicting political orientation and measuring political homophily in Twitter using big data. Journal of Communication 64(2): 317–332.

12.

Committee to Protect Journalists (2015) 10 most censored countries. Available at: https://cpj.org/2015/04/10-most-censored-countries/ (accessed 30 January 2021).

13.

Dahlgren

(2005) The internet, public spheres, and political communication: Dispersion and deliberation. Political Communication 22(2): 147–162.

14.

Del Valle

Bravo

(2018) Echo chambers in parliamentary Twitter networks: The Catalan case. International Journal of Communication 12: 21.

15.

Dennis

Martin

Hassan

(2019) Media use in the Middle East, 2019: A seven-nation survey. Available at: www.mideastmedia.org/survey/2019. (accessed 30 January 2021)

16.

Dennis

Martin

Wood

(2017) Media Use in the Middle East, 2017: A Seven-Nation Survey. Northwestern University in Qatar. Available at: www.mideastmedia.org/survey/2017 (accessed 30 January 2021).

17.

DeYoung

Nakashima

(2017, July 16) UAE orchestrated hacking of qatari government sites, sparking regional upheaval. Washington Post. Available at: www.washingtonpost.com/world/national-security/uae-hacked-qatari-government-sites-sparking-regional-upheaval-according-to-us-intelligence-officials/2017/07/16/00c46e54-698f-11e7-8eb5-cbccc2e7bfbf_story.html (accessed 30 January 2021).

18.

Dubois

Blank

(2018) The echo chamber is overstated: The moderating effect of political interest and diverse media. Information, Communication & Society 21(5): 729–745.

19.

Duffy

(2013) Cultures of journalism. Arabic- and English-Language Newspapers within the United Arab Emirates 9(1): 22.

20.

Eady

Nagler

Guess

, et al. (2019) How many people live in political bubbles on social media? Evidence from linked survey and Twitter data. SAGE Open. Epub ahead of print. DOI: 10.1177/2158244019832705.

21.

Emmons

(2018, August 1) Saudi Arabia planned to invade Qatar last summer. The Intercept. Available at: www.vox.com/world/2017/7/17/15983460/qatar-uae-hack-boycott (accessed 30 January 2021).

22.

Fieldhouse

Lessard-Phillips

Edmonds

(2016) Cascade or echo chamber? A complex agent-based simulation of voter turnout. Party Politics 22(2): 241–256.

23.

Festinger

(1962) A Theory of Cognitive Dissonance ( Vol. 2). Palo Alto, CA: Stanford University Press.

24.

Flaxman

Goel

Rao

(2016) Filter bubbles, echo chambers, and online news consumption. Public Opinion Quarterly 80(S1): 298–320.

25.

Fromherz

(2012) Qatar: A Modern History. Washington, DC: Georgetown University Press.

26.

Garrett

(2009) Echo chambers online?: Politically motivated selective exposure among internet news users. Journal of Computer-Mediated Communication 14(2): 265–285.

27.

Gerber

Henry

Lubell

(2013) Political homophily and collaboration in regional planning networks. American Journal of Political Science 57(3): 598–610.

28.

Ghattas

(2017, September 12) Lebanon’s shining island of example is shrinking. Foreign Policy. Available at: https://foreignpolicy.com/2017/09/12/beirut-cosmopolitan-example-arab-world-middle-east-peace/ (accessed 30 January 2021).

29.

Gil de Zúñiga

Diehl

Huber

, et al. (2017) Personality traits and social media use in 20 countries. Cyberpsychology, Behavior, and Social Networking 20(9): 540–552.

30.

Goldie

Linick

Jabbar

, et al. (2014) Using bibliometric and social media analyses to explore the “echo chamber” hypothesis. Educational Policy 28(2): 281–305.

31.

Gregorian

(2003) Islam: A Mosaic, Not a Monolith. Washington, DC: Brookings Institution Press.

32.

Hampton

Lee

Her

(2011) How new media affords network diversity: Direct and mediated access to social capital through participation in local social settings. New Media & Society 13(7): 1031–1049.

33.

Harris

Harrigan

(2015) Social media in politics: The ultimate voter engagement tool or simply an echo chamber? Journal of Political Marketing 14(3): 251–283.

34.

Hasebrink

Popp

(2006) Media repertoires as a result of selective media use. A conceptual approach to the analysis of patterns of exposure. Communications 31(3): 369–387.

35.

Hawley

(2020, November 21) Saudi Arabia’s human rights problems that won’t go away. BBC. Available at: www.bbc.com/news/world-middle-east-55002921 (accessed 30 January 2021).

36.

Howard

Duffy

Freelon

, et al. (2011) Opening closed regimes: What was the role of social media during the Arab spring? Available at: https://papers.ssrn.com/abstract=2595096. (accessed 30 January 2021)

37.

Jacobson

Myung

Johnson

(2016) Open media or echo chamber: The use of links in audience discussions on the Facebook pages of partisan news organizations. Information, Communication & Society 19(7): 875–891.

38.

Jamieson

Cappella

(2008) Echo Chamber: Rush Limbaugh and the Conservative Media Establishment. New York, NY: Oxford University Press.

39.

Lupáč

Sládek

(2008) The deepening of the digital divide in the Czech Republic. Cyberpsychology: Journal of Psychosocial Research on Cyberspace 2(1).

40.

Martin JD and Schoenbach K (2016) Predictors of blogging activity in six Arab countries. International Communication Gazette 78(8): 733–754.

41.

Martin JD, Martins RJ and Wood R (2016) Desire for cultural preservation as a predictor of support for entertainment media censorship in Saudi Arabia, Qatar, and the UAE. International Journal of Communication 10: 3400–3422.

42.

Martin JD, Martins RJ and Naqvi SS (2018) Media use predictors of online political efficacy among internet users in five Arab countries. Information, Communication and Society 21(1): 129–146.

43.

Martin JD, Naqvi SS and Arwah I (2020) Attitudes about censorship and internet surveillance among South Asians and nationals in the Arab Gulf: Predictors of digital self-expression values. International Communication Research Journal 55(2): 2–17.

44.

Masri

(2017) Tunisia: An Arab Anomaly. New York, NY: Columbia University Press.

45.

McCrae

(1987) Creativity, divergent thinking, and openness to experience. Journal of Personality and Social Psychology 52(6): 1258–1265.

46.

McPherson

Smith-Lovin

Cook

(2001) Birds of a feather: Homophily in social networks. Annual Review of Sociology 27(1): 415–444.

47.

Mutz

(1987) Political alienation and knowledge acquisition. Annals of the International Communication Association 10(1): 470–498.

48.

Nahon

Hemsley

(2014) Homophily in the guise of cross-linking: Political blogs and content. American Behavioral Scientist 58(10): 1294–1313.

49.

Ognyanova

Ball-Rokeach

(2015) Political efficacy on the internet: A media system dependency approach. In: Robinson L, Cotten SR and Schulz J (eds) Communication and Information Technologies Annual. Bingley, UK: Emerald Group Publishing Limited, pp. 3–27.

50.

Peng

Nisbett

Wong

(1997) Validity problems comparing values across cultures and possible solutions. Psychological Methods 2(4): 329–344.

51.

Phua

Jin

Kim

(2017) Gratifications of using Facebook, Twitter, Instagram, or snapchat to follow brands. Telematics and Informatics 34(1): 412–424.

52.

Pinnell

(2018, June 3) The online war between Qatar and Saudi Arabia. BBC. Available at: www.bbc.com/news/blogs-trending-44294826 (accessed 30 January 2021).

53.

Poushter

Bishop

Chwe

(2018, June 19) Social network adoption varies widely by country. Pew Research Center. Available at: www.pewglobal.org/2018/06/19/3-social-network-adoption-varies-widely-by-country/ (accessed 30 January 2021).

54.

Rainie L and Wellman B (2012) Networked: The New Social Operating System. Cambridge, MA: The MIT Press.

55.

Rugh

(2004) Arab Mass Media: Newspapers, Radio, and Television in Arab Politics. Westport, CT: Greenwood Publishing Group.

56.

Sasaki

(2017) Does internet use provide a deeper sense of political empowerment to the less educated? Information, Communication & Society 20(10): 1445–1463.

57.

Schoenbach

Wood

Saeed

(2016) Media Industries in the Middle EAst, 2016. Northwestern University in Qatar. Available at: www.mideastmedia.org/industry/2016/ (accessed 30 January 2021).

58.

Shields

Kane

(2011) Social and psychological correlates of internet use among college students. Cyberpsychology: Journal of Psychosocial Research on Cyberspace 5(1).

59.

Sunstein

(2001) Republic 2.0. Princeton, NJ: Princeton University Press.

60.

Taber

(2017) The use of Cronbach’s alpha when developing and reporting research instruments in science education. Research in Science Education 48: 1273–1296.

61.

The World Internet Project (2009) The digital future project. USC Annenberg School Center for the Digital Future. Available at: www.digitalcenter.org/wpcontent/uploads/2013/02/2009_digital_future_report-year8.pdf (accessed 30 January 2021).

62.

Vaccari

(2013) From echo chamber to persuasive device? Rethinking the role of the internet in campaigns. New Media & Society 15(1): 109–127.

63.

Vaccari

Valeriani

Barberá

, et al. (2016) Of echo chambers and contrarian clubs: Exposure to political disagreement among German and Italian users of Twitter. Social Media + Society 2(3).

64.

Velasquez

LaRose

(2015) Social media for social change: Social media political efficacy and activism in student activist groups. Journal of Broadcasting & Electronic Media 59(3): 456–474.

65.

Vliegenthart

Boomgaarden

Boumans

(2011) Changes in political news coverage: Personalization, conflict and negativity in British and Dutch newspapers. In: Brants K and Voltmer K (eds) Political Communication in Postmodern Democracy. London, UK: Palgrave Macmillan, pp. 92–110.