Eveland (2002) defines cognitive processing of information as the “general act of movement or manipulation of information in memory,” p. 28). This movement or manipulation of information starts with a person’s media exposure motives as grounded in U&G (Eveland, 2001, 2002). U&G considers the individual’s motives for consuming media messages to satisfy psychological and social needs. Motives determine which media channels and content individuals use, which in turn determines the outcomes of the media (Ruggiero, 2000).

People who consume more information have more information available for cognitive elaboration. Making more information available to cognitions requires retrieval of more mentally stored information to link with newly received news and information (Eveland, 2001). Studies show that consumption of greater amounts of media information is associated with more cognitive elaboration (Ho et al., 2013; Jiang et al., 2021; Zhang and Yang, 2021). In this research, it is expected that a frequent consumption of new news and information will trigger people to elaborate on the issue more often than people who receive little news and information.

People who frequently and exhaustively elaborate an issue’s information may reach elaboration saturation. They start to feel that they do not need to elaborate an issue further because elaborating it may not improve comprehending it. For example, regarding the issue of women’s political rights in Kuwait, individuals who have consumed a lot of information about it and often elaborated on the issue to form conceptualization and schema, may feel that receiving additional information about the same issue does not add anything new to the things that they already known or have conceptualized. In this case, they will directly store incoming information as supplementary information in their memory, without further elaboration (Kim and So, 2018). Also, many individuals use the media for opinion reinforcement. In this case, too, they directly store additional information for opinion confirmation (Cappella et al., 2015). Given the previous literature, it is anticipated that media information will increase people’s awareness about issues, whether by elaborating on it or not. Accordingly, the following hypotheses are formulated:

In the Arab region, government media agencies censor information and act as mouthpieces for authorities (Jamal and Melkote, 2008). Consequently, news information rarely presents a variety of opinions on an issue. This has caused Arab media researchers to long be interested in exploring how U&G is related to consuming of heterogeneous news with multiple perspectives and how it will affect audiences’ views, particularly on issues like free speech. U&G studies have found a link between a motivation to consume the news in order to be exposed to heterogeneous viewpoints and dissatisfaction with Kuwait’s authoritarian government (Jamal and Melkote, 2008), supporting democracy in Egypt (Al-Kandari and Hasanen, 2018), doubting simplistic justifications of conspiracy theories (Al-Kandari, 2010), and greater willingness to express opinions (Al-Kandari et al., 2022a,b). This U&G research indicates that the heterogeneous opinion exposure motive predicts support for progressive politics and discontent with current authoritarian politics. In the West, research also suggests that exposure to a wide range of diverse news sources is related to retaining diverse perspectives on issues (McLeod et al., 1999b) and a greater cognitive sophistication (Eveland et al., 2004). In relation to CMM and SOS, it is expected that people who seek out different perspectives on an issue will eventually hold diversified information about it, encouraging them to hold diverse viewpoints. Such diverse information induces greater cognitive elaboration in order to conceptualize different and conflicting issues. It will result in the emergence of many arguments and justifications that individuals can use in their opinion expressions about an issue with others.

Individuals motivated by a surveillance motive can meet their needs and desires by merely acquiring facts without opinion. In contrast, individuals employing a heterogeneous opinion exposure motive may encounter cognitive conflict or dissonance due to their desire to receive diverse perspectives on issues. People who are exposed to diverse perspectives on issues are likely to encounter news that contrast with their attitudes and beliefs (Taber and Lodge, 2006). Those following a heterogeneous opinion exposure motive will likely be exposed to information that includes opinions because they are seeking diverse viewpoints to give facts context (Taber and Lodge, 2006). Consequently, those pursuing heterogenous opinion exposure motives are more likely to encounter internal conflict than those following surveillance motives. In order to avoid cognitive dissonance from exposure to diverse viewpoints, people will elaborate more as a way of justifying and rationalizing new information that conflicts with their existing opinions (Quinn and Wood, 2004). For example, right wing supporters who select to view a left-wing news network may encounter opinions that oppose what they endorse. Watching these views will be more likely to stimulate the individuals to cognitively elaborate and use mental strategies that mitigate the influence of incoming media’s diverse information that may undermine or challenge their own opinions. Indeed, research by Guo and Chen (2022), Jiang (2022), and Qi et al. (2022) establish this link between information heterogeneity and greater cognitive elaboration. Based on the literature, it is expected that individuals who use the media for a heterogeneous opinion exposure motive will intensively elaborate on issues. Figure 2 indicates the way surveillance and heterogeneous opinion exposure motives influence cognitive elaboration. Thus, the following hypotheses are introduced:

Once exposure to media information is initiated by media motives, two cognitive processes, attention to information and elaboration on information (Zhang et al., 2024), mediate the processing of the learning of news information. Eveland (2001) explains that an ample body of research has demonstrated that attention, or the focus of mental effort, to the media is positively related to learning. Because attention is a process of minimizing mental distractions, it enables a greater mental concentration on an issue, resulting in a better grasp of many information cues and details.

On the other hand, elaboration is the cognitive “process of connecting new information to other information stored in memory, including prior knowledge, personnel experiences, or the connection of two new bits of information in new ways” (Eveland, 2001, p. 573). Outcomes from CMM research establish that the process of elaborating mass media messages expands mental routes for storing more information (Ho et al., 2013). The more people elaborate, the more they can produce inferences, conclusions and other cognitive outcomes that can be stored in memory (Bao and Lee, 2023; Luo et al., 2025; Zhang et al., 2024). In contrast, people who elaborate less will draw fewer inferences and conclusions. Similar to the literature that establishes that a greater level of elaboration about an issue’s information leads to expanding the routes of mental schemas and information storage about the issue, the current study proposes:

Learning is the final outcome in Eveland’s original CMM. Eveland (2001, 2002, 2004) conceptualized learning as the ability to recall facts from memory. Yang et al. (2017) suggested that perceived familiarity can serve as an alternative indicator of learning from mass media because many facts can be forgotten over time, but the basic idea or concept remains in memory. Many recent studies (e.g., Bao and Lee, 2023; Fang, 2024) have adopted perceived familiarity as an indicator of knowledge about issues instead of using measures of factual knowledge. Findahl (2001) states that, “The fact that so little specific information can be remembered from a news story does not mean that no learning takes place. The informative base from which conclusions are drawn can be forgotten while the conclusions remain” (p. 119).

People tend to consume more news and information to prepare themselves for anticipated social interactions (Eveland, 2004; Payne et al., 1988). Research on SOS indicates that retaining more information about a topic induces greater levels of reflection and deliberation, which are needed in social conversations, and results in a greater willingness to share one’s opinion (Shah et al., 2005). Having information about an issue contributes to greater self-confidence and willingness to express ideas (Chan et al., 2012) and a feeling of reassurance about the ability to exchange ideas and convince others (Sweetser and Kaid, 2008). In summary, this research suggests that people who frequently get involved in social communication tend to consume more media information. This increased consumption contributes to a greater storage of information which, in turn, encourages people to feel they have adequate information to support and defend their opinions should they choose to express them. Indeed, research by Neubaum (2022) indicates that individuals who view media messages are more likely to express opinions on social media. Also, Zhou and Yang (2024) found that an incidental news exposure contributed to online opinion expression. Therefore, the following is explored:

Information diversity is the second learning component this study explores. When exposed to news, people not only receive historical, factual, and background information, they also acquire various perspectives and opinions on an issue (Valenzuela et al., 2019).

SOS research suggests that having more information diversity triggers people to contribute more to conversations, even if those conversations are with people who hold differing opinions (Albuloushi and Algharaballi, 2014; McLeod et al., 1999a; Scheufele et al., 2004). Retaining heterogeneous information on an issue gives a speaker the ability to make sophisticated justifications of opinion positions (Brundidge, 2010; Price et al., 2002). Diversity of information can ensure more opportunity to back up and supplement opinions with various examples and arguments that can be used to refute the views of others (Shah et al., 2005). Individuals holding diverse information can support their expressed opinions and also anticipate the perspectives of others and how they think, which allows a speaker to better estimate what others may say (Sweetser and Kaid, 2008). It is anticipated that individuals with diverse information will be more confident in expressing contrary opinions because they can cope well with alternative perspectives and arguments (Brundidge, 2010). This suggests the following hypothesis:

Scholars have explored concepts, such as political efficacy, political information efficacy, and communicative utility, which have some similarity to information instrumentality. However, Table 2 provides a comparison of the concepts with some differences. Based on the limitations of other concepts found in Table 1, this study advances the concept of information instrumentality as a composition of three elements: (1) a mentally assessed confidence in personal knowledge, (2) a feeling of competence about one’s ability to express opinions based on gained knowledge, and (3) a certainty about the quality of information for opinion expression. Considering these elements together, information instrumentality refers to level of confidence in a person’s retained information about an issue that can be functional for an opinion expression behavior.

The present study adopts the concept of information instrumentality. Past research has established the fact that held knowledge about an issue contributes to the frequency of, as well as confidence in expressing opinions more boldly. For example, research by Chan et al. (2012) and Sweetser and Kaid (2008) confirms that individuals who know more about a topic tend to express their opinions more often and more boldly. In their study, Porten-Cheé and Eilders (2015) found that exposure to online user-generated information increased the likelihood of opinion expression online. As a form of feelings of self-confidence and self-competence, political efficacy has been positively to linked political participation within OSROR model (Chen, 2021; Kenski and Stroud, 2006), Finally, Shah et al. (2005) establish that knowledgeable people feel more self-assured about expressing opinions about a topic. In an experiment (Shah et al., 2005), subjects who were provided with more information about a candidate in a mock election tended to express their opinions confidently and more often than subjects who were provided with little information about the candidate. In summary, this research suggests that individuals who view themselves as knowledgeable feel well-informed and confidently express opinions more than people who think themselves ignorant. Thus, it is anticipated that:

FSI is the main inhibitor of opinion expression (Noelle-Neumann, 1977). FSI is the fear of being ostracized from one’s social network for expressing unpopular opinions. It encourages a social consensus that “must be constantly ensured by a sufficient level of agreement on values and goals” (Noelle-Neumann, 1991, p. 158). Much research confirms that FSI moderates speaking in public (Noelle-Neumann and Petersen, 2004).

FSI causes individuals to observe the media and other sources of information looking for cues about the major opinion so that they can assess whether their beliefs are congruent with the majority view. That in turn conditions people’s decisions regarding speaking out: “The effort spent in observing the environment is apparently a smaller price to pay than the risk of losing the goodwill of one’s fellow human beings, of becoming rejected, despised, alone” (Noelle-Neumann, 1993, p. 41).

As a result of the previous arguments, the following is expected:

Table 3 provides a summary of hypotheses in a cause-and-effect format.

For the first study, a sample of 502 students from Gulf University for Science and Technology, Kuwait, responded to a self-administered questionnaire about the recruitment of women into the military. This issue has been controversial since the Ministry of Defense announced that it would admit women recruits into the military (Alkhaldi, 2021). Conservatives used religious arguments and fatwas to justify their opposition arguing that Islamic teaching discourages the recruitment because it allows the unsupervised mingling of women and men in military camps and other facilities. Clerics also argued that the strenuous physical demands are suitable for men more than women. In opposite, the advocates of women military recruitment grounded their discussions on constitutional rights about gender equality and fair treatment (Alkhaldi, 2021) (Table 4).

The students who responded to the questionnaire were enrolled in general introductory courses. Selecting students from general introductory courses resulted in a more representative sample of students because the courses draw students from a variety of academic majors. Of the total, 211 (42%) of respondents were male, and 291 (58%) were female.

While a general population sample is better for generalizability, convenience samples, such as a student sample, can be useful in testing the reliability and validity of new models and testing the validation of new models and variables that are later to be used and expanded on with general population samples (Chan et al., 2022). Also, students represent an appropriate sample that can reflect a generational attitude of individuals who might reject previously held attitudes by older generations on different political and social issues. Such a sample can specifically reflect individuals who are better educated (Hanel and Vione, 2016). In recent years, research from Kuwait has indicated that this segment of society heavily uses online and social media and tend to view American media content as more likely to endorse women’s rights and have liberal outlooks on life (Abdulrahim et al., 2009). As a result, this study employed a student’s sample due to the exploratory nature of Study 1.

For the second study, network sampling was employed. A sample of 826 individuals, of which, 340 (41%) were males and 486 (59%) females, completed a self-administered questionnaire regarding the issue of non-citizen residents. An estimated 100,000 non-citizen immigrant residents (Hagagy, 2021), have long been denied Kuwaiti citizenship because of the country’s restrictive immigration policy. Human rights activists have proposed changing the policy based on legal and human rights. Opponents argue that a policy change would lead to adverse political and economic repercussions (Beaugrand, 2006) (Table 4).

For the sampling procedures, students in introductory courses at Kuwait University were asked to distribute the study’s questionnaire to family for extra credits. They were instructed to allow only their first-degree family members (father, mother, and siblings) who were 18 years old or older to complete the questionnaire. Students who were unwilling or unable to complete the activity were given alternative extra credit assignments.

For a culturally conservative Arab and Muslim state, such as Kuwait, the use of network sampling is a novel but useful approach. Allowing students to administer the questionnaire to their family members overcomes a cultural hurdle because social, cultural, and religious restrictions establish barriers to interactions between individuals of different genders (Cohen and Arieli, 2011), which makes it harder for male researchers to recruit female respondents. For example, a Kuwaiti study by Al-Kandari et al. (2022b) indicates that phone surveys report low response rates when the caller is male and the potential respondent is female. Such cultural barriers can be found in many Arab countries, especially those in the Arab Gulf. In addition, the use of first-degree relatives enhances trust and honest responses and participation, while mitigating potential bias (e.g., shared family ideology) due to the frankness that can be found among family members (Hanel and Vione, 2016).

The sample included 493 (59%) respondents who were age 18–25, 128 (16%) who were 26–33, 77 (9%) who were 34–43, 89 (11%) who were 44–55, and 39 (5%) who were 55 or older.

Kuwaiti has a young population.

Items representing the motives for using news media and political information were adopted from Al-Kandari et al. (2022a,b). For the surveillance motive, individuals indicated their use of news and information in order to “be aware of the daily subjects and events,” “follow new occurrences,” “know what events and news are happening,” and “know what happens around the world.” The last item, “know what happens around the world,” was used only in Study 1 because it loaded low (<0.70) in Study 2 on the factor loading test. This is as per the recommendation of Hair et al. (2014) (Study 1: M = 3.83, SD = 0.91, α = 0.902, composite reliability = 0.931; Study 2: M = 3.82, SD = 0.90, α = 0.873, composite reliability = 0.922). For the heterogeneous opinion exposure motive, items were: “view perspectives different from my own,” “know different viewpoints on issues,” “listen to opinions different from my own,” and “obtain diverse opinions about a specific issue.” The last item, “obtain diverse opinions about a specific issue,” was used only in Study 1 because it did not load well (<0.70) on Study’s 2 factor loading test. Study 1: M = 3.77, SD = 0.88, α = 0.875, composite reliability = 0.914; Study 2: M = 3.77, SD = 0.89, α = 0.800, composite reliability = 0.882. Note: Responses for all items used 5-point Likert-type scales of strongly agree as 5 and strongly disagree as 1 unless noted otherwise.

Items of Fear of Social Isolation (FSI) were adopted from Scheufele et al. (2001). They were, “I worry about being isolated if people disagree with me,” “I feel uncomfortable if I disagree with other people,” “I avoid telling other people what I think when there is a risk that they will avoid me if they know my opinion,” and “I feel annoyed if nobody wants to be around me because of my personal opinions” (This item was used only in Study 1 due low factor loading score (<0.70) in Study 2) (Study 1: M = 2.21, SD = 0.1.11, α = 0.853, composite reliability = 0.897; Study 2: M = 2.38, SD = 1.13, α = 0.882, composite reliability = 0.927).

Items for opinion congruence were created for both studies. Different numbers of items and different wording were used for both studies to test the validity of the scales of opinion congruence. For Study 1, respondents were asked to estimate the extent to which their opinions were congruent with the majority view in society: at the present time (current opinion congruence), in the future (future opinion congruence), and in media coverage (media congruence). Respondents answered on a scale of 1 to 10, where 10 reflected “to a great extent” and 1 reflected “to a little extent.” For Study 2, opinion congruence items were: “My opinion about the issue is similar to the opinion of the majority of people in our society,” “I have a similar opinion regarding the issue to most people in our society,” “Media coverage of the issue is similar to my opinion about this issue,” and “My opinion in the future regarding the issue will be comparable to the opinion of the majority of people in our society on this issue,” (Study 1: M = 5.49, SD = 2.28, α = 0.650, composite reliability = 0.792; Study 2: M = 3.22, SD = 1.03, α = 0.805, composite reliability = 0.869).

Cognitive elaboration was composed of an index adopted from Eveland (2001). The items were: “I often think about the news I receive from media about the issue,” “I think about how what I receive from the media about the issue links to other things I know about this issue,” “I try to relate the news and information I receive from the media about the issue to my own past experiences,” and “I frequently link new news about the issue to my own prior personal experiences.” The last item, “I frequently link new news about the issue to my own prior personal experiences,” was used only for Study 1 due to low factor loading score (<0.70) in Study 2, (Study 1: M = 2.90, SD = 1.05, α = 0.841, composite reliability = 0.904; Study 2: M = 3.29, SD = 1.04, α = 0.878, composite reliability = 0.916).

Two question items, adopted from Yang et al. (2017) were used to assess perceived familiarity. They were: “How extensive is your knowledge about the details of the issue?” and “How extensive is your knowledge about the background of the issue?” Respondents used a 5-point scale (Very high = 5, Very low = 1), (Study 1: M = 2.50, SD = 1.18, α = 0.897, composite reliability = 0.951; Study 2: M = 3.35, SD = 1.05, α = 0.887, composite reliability = 0.946).

Information diversity was measured using the following items: “To what extent do you know about other people’s opinions, especially those that contradict your own opinion on the issue?” “How extensive is your knowledge of the different perspectives regarding the issue, especially those that differ from your own?” and “To what extent are you aware of the other viewpoints that are different from yours about the issue?” This last question was asked only in Study 2 due to the low factor loading score (<0.70) in Study 1. Response options were on a 5-point scale (5 = Very great extent, 1 = Very little extent). Study 1: M = 2.90, SD = 1.05, α = 0.876, composite reliability = 0.942; Study 2: M = 3.12, SD = 1.04, α = 0.824, composite reliability = 0.895.

Information instrumentality was measured using the following question items: “To what extent do you trust that your information about the issue will enable you to convince others about your viewpoint in conversations about this issue?,” “To what extent are you certain that the information you know about the issue will benefit you in your discussions with others about this issue?,” and “To what extent do you trust that your knowledge about the issue will help you to exchange your opinions in conversations with others about this issue?” [This question was asked only in Study 2 and dropped from Study 1 due to low factor loading score (<0.70)]. Response options were on a 5-point scale (5 = Very great extent, 1 = Very little extent) (Study 1: M = 2.92, SD = 1.15, α = 0.877, composite reliability = 0.942; Study 2: M = 3.22, SD = 1.09, α = 0.868, composite reliability = 0.919).

For Study 1, four items were employed to measure the likelihood of opinion expression in four different opinion incongruent conversation climates. An opinion incongruent setting presented hypothetical situations in which a majority of people expressed an opinion that differed from that of the respondent. The respondents used a 5-point scale (Extremely likely = 5, Extremely unlikely = 1) to estimate their likelihood of opinion expression in settings where they did not know people. The settings were offline, online, offline where the majority in the setting were of the opposite gender, and offline where the majority of people were 20 years older than the age of respondents. These last two settings were necessary to assess the likelihood of expressing an opinion in a gender segregated and patriarchal situation, such as those frequently found in the Kuwaiti cultural context.

For Study 2, the likelihood of expressing an opinion was assessed using five items. The respondents were first introduced to a hypothetical scenario reflecting an incongruent opinion climate in which the respondents did not know other people in the setting and in which a majority of people expressed an opinion that differed from that of the respondents. Then, the respondents were asked to estimate their likelihood of expressing an opinion to “Show the incorrectness of the other opinion,” “Defend my viewpoint,” “Prove that an opinion is right,” “Convince others of a viewpoint,” and “Persuade others of an opinion.” For each item, responses were recorded using a 5-point scale (Extremely likely = 5, Extremely unlikely = 1). Study 1: M = 2.86, SD = 1.35, α = 0.794, composite reliability = 0.868; Study 2: M = 3.65, SD = 1.06, α = 0.782, composite reliability = 0.851.

For all of the above scales, construct reliability was established through the tests of Alpha Cronbach and composite reliability. Convergent validity was established using average variances extracted (AVE). Finally, discriminant validity was established using the tests of Fornell-Larker and Heterotrait-Monotrait ratio (HTMT). Those criteria tests were employed based on the suggestions of DeVellis (2016), Henseler et al. (2015), and Fornell and Larcker (1981). Details are provided in the upcoming Results section.

Cronbach alpha scores for construct items were from 0.615 to 0.90 and ranges of composite reliability were from 0.79 to 0.94. Even though the Cronbach alpha for one variable was less than 0.615, DeVellis (2016) considers 0.6 for Cronbach alpha and 0.7 for composite reliability as acceptable thresholds (see Table 5). Also, since Study 1 is exploratory in nature, it is acceptable to allow slightly lower alphas in exploratory research since survey items can be later modified to increase alpha levels in a later model validation study (e.g., Nunnally, 1978).

The values of factor loading ranged from 0.61 to 0.96 and AVE ranged from 0.56 to 0.90. Both met the threshold values of acceptable factor loading of 0.60 and AVE of 0.50, respectively (Henseler et al., 2015). In this regard, the AVE score was 0.773 for the surveillance motive, 0.727 for the heterogeneous opinion exposure motive, 0.686 FSI, 0.561 opinion congruence, 0.758 cognitive elaboration, 0.907 perceived familiarity, 0.89 information diversity, 0.89 information instrumentality, and 0.625 opinion expression (see Table 5).

For cross-loadings, each construct’s indicators loaded higher than that of its corresponding construct. The Fornell-Larker criterion was satisfied as the square roots of all AVEs exceeded their correlations with other constructs (Fornell and Larcker, 1981). As for HTMT ratio, the statistics indicate that none of the construct values exceeded the threshold value of 0.85 (Henseler et al., 2015) (see Table 5). It should be noted that HTMT of < 0.85 is considered a conservative threshold (Henseler et al., 2015).

The first study included a sample of 502 respondents. Males were 211 (42%) and females were 291 (58%) (see Table 4). The results showed that all hypotheses were supported in Study 1, with only one exception. The results indicated that: a surveillance motive caused cognitive elaboration (H1a) (β = 0.21, t = 4.33, p < 0.001) and perceived familiarity (H1b) (β = 0.22, t = 5.94, p < 0.001), a heterogeneous opinion exposure motive caused information diversity (H2a) (β = 0.17, t = 3.88, p < 0.001) and elaboration (H2b) (β = 0.18, t = 3.60, p < 0.001), elaboration predicted perceived familiarity (H3a) (β = 0.50, t = 13.82, p < 0.001), and information diversity (H3b) (β = 0.33, t = 7.44, p < 0.001), perceived familiarity caused information instrumentality (H4) (β = 0.52, t = 12.90, p < 0.001), information diversity predicted information instrumentality (H5) (β = 0.22, t = 5.04, p < 0.001), information instrumentality triggered opinion expression (H6) (β = 0.39, t = 10.43, p < 0.001), and FSI negatively predicted opinion expression (H7a) (β = −0.24, t = 6.10, p < 0.001). Only H7b was not supported, indicating that a person’s opinion congruence with that of the public did not predict opinion expression (β = 0.06, t = 1.33, p = 0.182) (see Table 6; Figure 3).

This section presents the significant mediating paths of the model starting from CMM’s news and information inputs (surveillance and heterogeneous opinion exposure motives) to the behavioral outcome (opinion expression) (see Table 7 for all mediations). In total, there were six significant paths. Below is a discussion reporting the strength of paths according to level of Beta statistics.

The strongest path started with the surveillance motive mediated by elaboration to perceived familiarity then instrumentality that ultimately predicted opinion expression (β = 0.05, t = 4.78, p < 0.001). The second strongest path was through the surveillance motive going through elaboration and mediated by information diversity then instrumentality and finally ended predicting opinion expression (β = 0.02, t = 3.59, p < 0.001). The third strongest path was that of the heterogeneous opinion exposure motive being mediated by elaboration through perceived familiarity and then instrumentality to predict opinion expression (β = 0.02, t = 2.98, p = 0.003). The weakest path was from the heterogeneous opinion exposure motive that was mediated by elaboration then diversity and instrumentality that stimulated opinion expression (β = 0.01, t = 2.35, p = 0.019) (Table 7; Figure 3). Regarding the weak path of heterogenous motive, probably having diverse perspectives on an issue distracts individuals. This is unlike people who use the surveillance motive to seek specific information that helps them to remain focused and able to effectively be aware of what to discuss in conversations.

The following is the results for Study 2. It is a replication of the previous study to validate the model. All study variables used in Study 1 were used in Study 2. The only exception as mentioned in the Methods section is that some items were dropped from Study 1 or 2 due to low factor loading scores (<0.70). Also, while a student sample was used in Study 1 because it was an exploratory model study, Study 2 was conducted using a network sample to validate the model.

Cronbach’s alpha scores were between 0.78 and 0.89 and composite reliability were between 0.85 and 0.95. For both, they were above the acceptable threshold (Table 8).

Values of factor loading were from 0.66 to 0.95 and AVE from 0.53 to 0.89. They met the threshold values. In this respect, the AVE score was 0.798 for the surveillance motive, 0.715 for the heterogeneous opinion exposure motive, 0.808 FSI, 0.627 opinion congruence, 0.732 cognitive elaboration, 0.898 perceived familiarity, 0.740 information diversity, 0.791 information instrumentality, and 0.534 opinion expression (see Table 8).

For cross-loadings, each indicator’s loading on its intended construct was higher than on other constructs. For Fornell-Larker, the square roots of all AVEs exceeded their correlations with other constructs. For HTMT, none of the construct values exceeded the threshold value of 0.85.

Study 2 included a sample of 826 respondents. Males were 340 (41%) and females were 486 (59%). Individuals who belonged to 18–25 age group were 493 (59%), 26–33 were 128 (16%), 34–43 77 (9%), 44–55 89 (11%), and 55 or older were 39 (5%) (Table 4).

In Study 2, all hypotheses were confirmed. The hypotheses were: surveillance motive predicted cognitive elaboration (H1a) (β = 0.30, t = 7.90, p < 0.001) and perceived familiarity (H1b) (β = 0.17, t = 5.11, p < 0.001), a heterogeneous opinion exposure motive predicted information diversity (H2a) (β = 0.15, t = 4.41, p < 0.001) and predicted cognitive elaboration (H2b) (β = 0.20, t = 5.21, p = 0.001), cognitive elaboration predicted perceived familiarity (H3a) (β = 0.46, t = 14.96, p < 0.001) and information diversity (H3b) (β = 0.39, t = 11.52, p < 0.001), perceived familiarity predicted information instrumentality (H4) (β = 0.55, t = 20.36, p < 0.001), information diversity predicted information instrumentality (H5) (β = 0.26, t = 8.67, p = 0.001), information instrumentality predicted opinion expression (H6) (β = 0.29, t = 8.24, p < 0.001), FSI negatively predicted opinion expression (H7a) (β = −0.12, t = 3.21, p < 0.001), and a person’s opinion congruence with that of the public predicted opinion expression (H7b) (β = 0.14, t = 4.09, p < 0.001) (see Table 9; Figure 4).

There were seven significant paths for indirect effects. The strongest indirect path was from the surveillance motive through elaboration and perceived familiarity to instrumentality and ultimately to opinion expression (β = 0.02, t = 4.78, p < 0.001). The second strongest path was from the heterogeneous opinion exposure motive mediated by elaboration then perceived familiarity through instrumentality and ending with opinion expression (β = 0.02, t = 4.23, p < 0.001). The third strongest path was by the surveillance motive going through perceived familiarity to instrumentality that predicted opinion expression (β = 0.03, t = 4.21, p < 0.001). The weakest path started with the surveillance motive that went through diversity then instrumentality to ultimately trigger opinion expression (β = 0.01, t = 2.71, p = 0.007) (Table 9 reports all paths). The results showed that this weak path did not go through cognitive elaboration. It is likely that people who cognitively elaborate on an issue create a greater number of mental paths in which information can be stored to be used later in opinion expression contexts.

In both studies, all hypotheses were confirmed except for H7b in Study 1, which predicted that FSI will cause people to avoid opinion expression. The rest were supported in both studies. In this regard, individuals who often used news media for the surveillance and heterogeneous opinion exposure motives were more likely to cognitively elaborate the news that they received. Also, individuals who viewed news media for the surveillance motive reported higher levels of perceived familiarity while individuals who viewed them for the heterogeneous opinion exposure motive reported a higher level of information diversity. In addition, individuals who often elaborated issues were more likely to report greater levels of perceived familiarity and information diversity. Furthermore, those with greater levels of perceived familiarity and information diversity indicated greater levels of self-perception of information instrumentality. Moreover, individuals with greater levels of self-perception of information instrumentality tended to express their opinions about issues more often. Finally, individuals with greater FSI tended to avoid expressing their opinions about issues (only in Study 2), while those who felt their opinions were congruent with those of the public tended to be more expressive of their opinions (Tables 6, 9).

In both studies, the strongest predictor, as indicated by high Beta values (Study 1 = 0.52, Study 2 = 0.55), was the positive relationship of information instrumentality by perceived. The second strongest positive prediction was cognitive elaboration predicting perceived familiarity in both studies: Study 1 (β = 0.50) and Study 2 (β = 0.46). Those that had moderate effects in both studies were elaboration predicting information diversity, with Beta values in Study 1 of 0.33 and Study 2 of 0.39 and instrumentality predicting opinion expression in Study 1 of 0.39 and Study 2of 0.29. The rest of the hypotheses were significant but indicated moderate to weak prediction levels and values (see Tables 6, 9).

In Study 1, the surveillance motive paths were stronger than the heterogeneous opinion exposure motive paths in predicting opinion expression at the end in an indirect way. In both studies, the strongest indirect path was that surveillance motive, which was mediated by elaboration through perceived familiarity then instrumentality that at the end predicted opinion expression. The heterogeneous opinion exposure motive path was stronger in predicting opinion expression indirectly in Study 1 than Study 2. Those ranks are evaluated by Beta values (see Tables 7, 10; Figures 3, 4).

In addition, all paths were fully and positively significant in both studies, except the heterogenous motive that predicted familiarity through instrumentality that ultimately predicted opinion expression. Another exception was only in Study 1. The surveillance motive predicted diversity, which was mediated by instrumentality that ultimately predicted opinion expression. This path too was not significant (see Tables 7, 10; Figures 3, 4).

Future research in political communication can use CDMM by coupling information instrumentality with Bandura and Adams’s (1977) self-efficacy. Individuals who trust that their information is instrumental, and empowering may feel that they are efficacious and can contribute to the political process. Together, they can have a predictive value of political engagement and ways of involvement in campaigning and voting. Information instrumentality in the CDMM can be related to cognitive dissonance, knowledge gap and other public opinion theories as well. Also, CDMM can be related to the concepts of opinion leaders and opinion followers. Having a stronger sense of information instrumentality, it is expected that opinion leaders would disseminate views and ideas to opinion followers.

Future research also could apply CDMM to explore demographics as moderators. Research could compare how old and young people mediate information. Older people, who are expected to have stored more information, may find their information instrumentality affects their decisions regarding different communication outcomes more than younger people. In relation to gender, women may be found to elaborate and store information about women’s issues more often than men, and this will affect their information instrumentality. People with low incomes may elaborate and seek information about the cost of living, which may influence their information instrumentality more than people with higher incomes (Alsabah et al., 2025). In this regard, CDMM can be explored in relation to knowledge gap by emphasizing the interaction effects between income and education as indicators of socio-economic status (SES) on cognitive internalization. Researchers can adopt multi-group analytical and statistical techniques to study how people as different groups react to media issues differently. This would probably provide a more specific approach that deals with different publics instead of using a generalized approach that fits all.

Future research may relate information instrumentality to attitude certainty. While information instrumentality indicates a person’s confidence that retained knowledge contributes to a behavior, attitude certainty reflects the unwavering and firmness about the correctness of an embraced attitude on a social phenomenon (Al-Enzi et al., 2017; Matthes et al., 2010). Attitude certainty was found to associate with the hardcore who expressed their opinions regardless of opinion climate (Matthes et al., 2010). Future research may explore whether information instrumentality contributes to attitude certainty, which, in turn, contributes to the likelihood of opinion expression or in relation to other communication behaviors. It is expected that people who know more can be more certain about the correctness of their attitude than those who know little. In this sense, information instrumentality can reinforce an opinion that eventually strengthens attitude certainty that will eventually predict opinion expression. Future research can explore if weak or strong levels of attitude certainty can be attributed to information instrumentality. Figure 5 is an expected model of a path starting from information instrumentality, going through attitude certainty, which is to predict opinion expression at the end.

Finally, future studies could explore the CDMM regarding different issue orientations. As we discussed, cultural and religious issues make little room for cognitive elaboration than politically oriented issues. With cultural and religious issues, we argue that people are more likely to receive information to store right away in memory. They will be subject to less elaboration. Future studies also could explore the paths between elaboration and memory as learning components. If a study finds that the weight value of elaboration predicting learning is higher than learning predicting elaboration, then this means little information is retrieved from memory for elaboration. In this case, information is mainly stored in learning components after little elaboration. If the weight value of learning mediating elaboration is higher, then in this case, people frequently retrieve information from memory (learning) to elaborate on the issue more. This can be tested in a study using many issues for comparisons.