TWEETED ATTITUDES TOWARDS WOMEN PARLIAMENTARY CANDIDATES IN KUWAIT: A SOCIAL DOMINANCE PERSPECTIVE

This study explores how Kuwaitis use Twitter to communicate their attitudes towards women Parliamentary candidates (WCs) in a traditionally male-dominated society, and how these tweeted attitudes are thematically constructed, either negatively or positively. The study also explores how these attitudes differ according to gender and evolve quantitatively and qualitatively over subsequent elections. A total of 1744 tweets about all eight women candidates in 2013 posted for 40 days prior to the Kuwait Parliamentary Election were retrieved and analyzed. The tweets posted about the two women candidates of those eight who continued to run in 2016 and 2020 were also analyzed in terms of length, content, and themes. Gender significantly correlated with attitudes in the first election, but not in the subsequent two elections. Tweeted attitudes turned to be more elaborate, information-based, and longer over consecutive elections. The dominant positive theme was generic, whereas the dominant negative theme was specific, and candidate based. Women candidates were praised for acting like men, whereas they are mocked for looking like men.

Politically marginalized groups, such as women in patriarchal societies, are found to be using Twitter as a "counter-public sphere" to share their voices and gain publicity (Hu and Kearney, 2020). Political leadership in the Arab World is dominated by men. No woman has ever been able to attain the position of head of state or Speaker of Parliament in this part of the world over centuries. In addition, the Arab region has the lowest regional average of women parliamentarians worldwide (IPU, 2021). In these historically patriarchal societies, women must overcome cultural hurdles to claim political leadership roles, even though the first Arab woman minister was appointed in 1962 in Egypt (Bier, 2010).
In the traditionally patriarchal society of Kuwait, women were granted the right to vote and stand in Parliamentary elections in 2005. It was a major political and cultural transformation that posed several questions about how the candidacy and potential leadership of women would be evaluated by both the dominant and marginalized genders in this country. Ever since, the participation of women in the Kuwaiti parliamentary elections has not been representative of the percentage of women in the population. In some instances, women candidates have never been able to win a seat in Parliament. This demonstrates that women do not necessarily vote for women in this part of the world, which raises questions about how voters perceive the strengths and weaknesses of women candidates and whether these perceptions vary according to the gender of the voters. A longitudinal analysis of the attitudes expressed by social media users would provide an unobtrusive tool for measuring these attitudes from one election to another. Using a survey to analyze the attitudes of voters toward women candidates in a collectivist culture that values face-saving implies the risk of obtaining less reliable responses.
The present study seeks to deconstruct and hence, thematically categorize the attitudes of Twitter users towards women candidates (WCs) in Kuwait over time, not just the general sentiment. The study also seeks to identify whether those tweeted attitudes toward WCs who keep running in consecutive parliamentary elections change affectively and cognitively over time. In addition, the study fills a gap in the literature about the cultural perceptions of WCs in a traditionally patriarchal society that is passing through a phase of both modernization and democratization.

BACKGROUND
Kuwait gained its independence from Britain in 1961. A year later, this monarchy wrote its constitution and established the first People's Assembly in the Arabian Peninsula. The new constitution denied women the right to either vote or run for parliamentary elections. Women's suffrage had to wait until 2005 when the parliament honored the Amir (ruler) Decree, which granted women the right to elect and be elected. A unique element of the political environment in Kuwait is that this constitutional democracy has never had any official political parties. Tribes and ideological leanings (Islamic versus liberal) represent the functional alternative to what is called a multiparty system in other modern democracies.
With the initiation of commercial exports of Kuwait oil in 1946, the per capita income and the gross domestic product started to skyrocket. This monarchy was ranked as having the 8 th highest per capita income in 2017 (Worldodometer). The foundation of the Kuwaiti economy has been transformed from trading, fishing, and pearling to the extraction and exportation of oil, paving the way for modernization in Kuwait. In recent years, communication technology has become an intrinsic component of the lives of average citizens. Kuwait has become one of the most active countries on Twitter. With the highest number of per capita tweets, Kuwait was known as the "tweetiest" country in the world (Moncanu et al., 2013). In 2016, Kuwait had the most Twitter users per capita in the MENA region (Statista, 2016). Although the population of Kuwait represents only one percent of the population of the Arab world, Kuwaitis were estimated to produce 10% of the tweets in the region (Arab Social Media Report, 2017). In March 2017, Twitter had a penetration rate of 12.6% in Kuwait, which was the second-highest rate after Bahrain (Weedoo, 2017).
In the petroleum era, Kuwait has found itself caught between the two opposing forces of modernization and tradition (Tetreault & Al-Mughni, 1995), which has led to the creation of a political system that encourages the education of women but discourages their civil independence and participation in political leadership. By law, women do not have the same legal rights as men. Kuwaiti women who marry non-Kuwaitis cannot pass on their nationality to their children, whereas the children of a Kuwaiti male are born Kuwaitis. These newly granted rights to elect and be elected did not lead to a significant surge in the political activism of women. Although women represent most of the electorate in Kuwait, only a handful of WCs to none have managed to win seats in Parliament. As reported by the World Bank (2021), the marginalized political status of Kuwait women is not commensurate with their participation in the labor force compared to men. According to the latest report of the Central Statistical Bureau (CSB, 2019-2020) of Kuwait in 2019, Kuwaiti women slightly outnumber Kuwaiti men in the labor market (188,141 women vs. 154,276 men), even though women represented around 75% of the student body at Kuwait University (the only state university in the country) in the same year (National Academies of Sciences, Engineering, and Medicine, 2020).

THEORETICAL FRAMEWORK
In his "resource curse" theory, which seeks to explain gender inequality in the context of the prevailing economic activity in a society, Michael Ross suggests that oil production is the primary reason women lag (Ross, 2008). According to him, oil extraction relies on manual jobs, which are traditionally male-dominated, and that fact lowers the level of participation of women in the labor market. This in turn affects women's identities and self-perceptions negatively, and thus limits their participation in formal political and economic networks. Accordingly, the political influence of women would be diminished because "petroleum perpetuates patriarchy" (Ross, 2008, p. 120). However, this theory falls short of explaining women's empowerment in other traditional but less industrialized societies. Rwanda is the leading country worldwide in terms of women parliamentary representation, with women holding 61.3% of the seats in the Lower House in 2021 (Inter-Parliamentary Union, 2021). On the other hand, women suffer from gender inequality in non-oil-rich countries such as Jordan and Yemen. Pippa Norris argues that Ross does not factor in how the "public feels about men and women in the workforce, society's moral values towards sexuality, and the strength of religious identity" (Norris, 2011, p. 9). Norris suggests that it is "Mecca" rather than "Oil" that explains gender inequality in Arab societies. In her meta-analysis of a series of social change studies, Norris remarked that egalitarian attitudes towards the election of women are stronger in non-Islamic societies. She concluded that "long-standing religious traditions leave an enduring mark on gendered norms and beliefs" (Norris 2011, p. 17). In her view, it is Islam that determines "the division of labor for men and women in the home, family, and public sphere." A substantial body of literature suggests that Islam is the main factor that accounts for the disempowerment of women (Abdalla, 1996;Amawi, 2001;Inglehart & Norris, 2003;Jamali, Sidani, & Safieddine, 2005;Khimish, 2014;Pettygrove, 2011;Saleema & Tlaiss, 2011;Sidani, 2005). However, women have reached the top position (presidency or premiership) in several Muslim nations, such as Bangladesh, Indonesia, Pakistan, Turkey, and Tunisia. Women also have 43% of the seats in the Senegal Parliament, where Islam is the main religion. In contrast, women have 27.6% of the House of Representatives in the U.S. (IPU, 2021). Attitudes towards women in leadership roles are found to be negative in some predominantly Christian societies. In Greece, employed men were not found to hold strong positive beliefs about the ability of women to assume leadership roles (Galanaki, Papalexandris, & Halikias, 2009).
Both the "Resources Curse" and the "Mecca" theories adopt ungeneralizable perspectives by attributing gender inequality in oil-rich Arab countries to historical and ideological factors, respectively. Women claim significant leadership roles in oil-rich, Muslim countries, such as Azerbaijan, which met the world's demand for crude oil in 1890 and gave women the right to vote in 1918 before both the U.K. and the U.S. did (Scott, Dakin, Heller, & Adriana, 2013). The "Resources" and "Mecca" theories lack this cultural/social perspective in explaining women disempowerment, whereas the theory of Social Dominance Orientation (SDO) is more focused on that perspective.
The theory of SDO explains how societal hierarchies and attitudinal orientations cause individuals to either accept or reject the traditional hierarchy of power within their societies. It focuses on "attitudes towards hierarchical relationships between groups and the desire to promote intergroup domination" (Sidanius, Pratto, Laar, & Levin, 2004, p. 858). Individuals who have a high level of SDO tend to adhere to the traditional hierarchy in their society, even if it requires bias to do so (Simmons, Duffy, & Alfraih, 2012). Simmons and his colleagues studied a sample of 89 college students from the U.S. and Kuwait. They found that high levels of SDO were negatively related to favorable attitudes towards women as managers.
The relationship between gender and SDO has been the focus of the Invariance Hypothesis (IH). According to this hypothesis, men tend to have higher levels of SDO than women, regardless of social class and family income (Pratto and Sedanius, 1994). Men, who are historically more dominant politically and economically, will be more reluctant to embrace ideas of egalitarianism and gender equality. They would also use negative, sarcastic attitudes towards WCs as an equality-suppressing and hierarchy-enhancing strategy that serves to legitimize the myth that inequality is normal and, thus, justified.
To the contrary of IH, SDO explains why some men might not overtly oppose women's leadership roles. It predicts that men generally play roles that enhance hierarchy (such as police, military, and business executives), whereas women are "over-represented in roles that accentuate hierarchy, such as teachers and social workers" (Pratto, Sidanius, & Levin, 2006, p. 295). Men will tend to maintain hierarchy if they are in control of the resources of their society. However, men do not tend to show overtly biased behavior towards women, as this will not be necessary when hierarchically driven, long-established oppression within a society is already suppressing women (Simmons, et al., 2012). For instance, all parties emphasized the role of women during the parliamentary elections of 1997 in Jordan. However, all the parties and tribes-except for the Communist Partyrefrained from fielding WCs (Amawi, 2007).
SDO also assumes that the subordinate members of a society will collaborate with the dominant members in order to maintain the hierarchical structure of their society and, hence, participate in their own oppression (Pratto et al. 2006, p. 276). Although the oppressed might tend to support each other emotionally, they will tend to "display greater endorsement of hierarchy-attenuating legitimizing myths" (Pratto et al.,p. 276). This might explain why WCs in traditionally hierarchical societies do not receive the majority of the votes, even though women represent the majority of the voters. This suggests that women voters in Kuwait are less likely to be significantly more supportive of WCs than male voters.
Accordingly, gender might predict the variance in the tweeted attitudes towards WCs in Kuwait, especially after the legal changes in 2012 that led candidates to run individually. Accordingly, voters would no longer vote for one of the collaborative groups of candidates but instead select individual candidates. This represented a challenge for WCs, who would have to run individually instead of collaborating with other well-established male candidates.
The literature does not provide empirical evidence that gender will correlate significantly with attitudes towards WCs following this legal change, especially after the inability of such candidates to win more than two seats in 2013. The theory of SDO predicts that the dominant social group tend to communicate more favorably towards the suppressed group as long as they are capable of maintaining the traditional societal hierarchy. RH1: Gender is expected to correlate more significantly with the tweeted attitudes towards Kuwaiti WCs posted before the 2013 elections than with those posted before the subsequent elections in 2016 and 2020.
Perceptions of gender roles and attitudes towards women leadership are relative constructs that are influenced by a myriad of differing factors. An analysis of such perceptions and attitudes may help in identifying those factors. Kauser and Tlaiss (2011) suggest that negative attitudes towards women managers "reinforce gender role stereotyping that defines women via their domestic and reproductive chores" (p. 38). Similarly, a study analyzing tweets sent to the parliamentary members in Britain found that women members were the targets of gendered abuse and incivility (Southern and Harmer, 2021). In the Arab region, negative attitudes towards working women have been shown to be the biggest obstacle to the career progress of women (Jamali et al., 2005;. In such patriarchal societies, women are perceived primarily as caregivers who are either dependent on or dedicated to the care of others, whereas men are perceived as breadwinners and leaders. In these societies, women rely on the support of their families or tribes in order to succeed in business or politics. Shanahan (2009) reported that the six women who won in the 2005 Parliamentary election in Lebanon succeeded because of their links with patriarchal political families. Madsen (2010) conducted in-depth interviews with six women business leaders in the UAE. These professional women did not report any dissatisfaction with gender roles in their families; they described their fathers as being well-traveled, open-minded, and supporters of women's education, whereas they referred to their mothers as "kind" and "pretty." However, the continued involvement of women in leadership roles might be another factor in changing how they are perceived by other members of society. The attitudes and opinions voiced about WCs via public forums might evolve over time if those candidates maintain their exposure and political involvement. According to the elaboration likelihood model, with repeated exposure to messages from WCs, individuals might process those messages more centrally, where they tend to "bring diverse issue-related thoughts and previous experience to evaluate an argument" (Oh & Sundar, 2015, p. 215). Individuals might also use a central route (issues and policies) in processing information about WCs after gaining more information about them from one campaign to the subsequent. Accordingly, the peripheral stimuli about WCs (non-issue-related stimuli, such as gender, age, or appearance) might be processed more attentively by voters who lack prior knowledge about those candidates. Being exposed to more information about WCs, individuals might apply a more logical reasoning in supporting or rejecting them.
Research also shows that Twitter's decision in 2017 to increase the maximum length of tweets from 140 to 280 characters had an impact on the linguistic aspects of tweets (Gligorić, Anderson, & West, 2018). Accordingly, Twitter users are expected to use more characters (quantitative cues) and express more informationdriven attitudes (qualitative cues) in their tweets about WCs who run for elections frequently. RH2: Tweeted attitudes towards WCs in Kuwait will be associated with peripheral cues in the elections of 2013 and central cues in the subsequent elections. RH3: Tweeted attitudes towards WCs in Kuwait will be significantly longer in 2016 and 2020 compared to 2013.
Identifying the "content" of both negative and positive tweeted attitudes towards WCs in Kuwait is also important to "capture emerging themes" (Barry et al., 2005, p. 239) and uncover how Twitter users react to the candidacy of women. Such thematic analysis is expected to provide a description of the affective and cognitive components of the tweeted opinions and attitudes towards WCs in Kuwait over three consecutive parliamentary elections. RQ: What are the cognitive and affective components that summarize the main themes of the tweeted attitudes towards WCs in Kuwait before the parliamentary elections in 2013, 2016, and 2020?

Data
In 2012, the electoral laws were amended to reduce the number of candidates a Kuwaiti can elect from four to one. This represented a challenge for underrepresented groups, especially women, who had to run individually, not in collaboration with long-established candidates. The present study analyzes the tweeted attitudes towards all WCs who ran after this legal modification. First, the tweets about all eight WCs 1 who ran in 2013 were retrieved and analyzed. The tweets posted about the two WCs who continued running in the subsequent two elections in 2016 and 2020 were also analyzed. Only the tweets posted during the legal time of campaigning in those three elections (40 days before voting started) were retrieved. The call for candidates to declare their candidacy is usually announced 40 days before the election. The tweets were searched and retrieved using Twitter Advanced Search. The names of the WCs were the search keywords. Table 1 shows specific dates of data collection and the ratio of participation of WCs in each election. A total of 1877 tweets were retrieved. The tweets posted by either the news media or the WCs themselves were separated from the rest of the tweets and discarded from the analysis. Retweets that included only video and/or news clippings were also excluded, and only text messages were included in the analysis. The actual number of tweets included in the analysis is 1744.

Coding & Analysis
A native Kuwaiti researcher 2 coded 10% of the tweets. A test of inter-coder reliability showed a range of .85 to .95 of Alpha Cronbach's reliability for the categories and themes used in the analysis, indicating an acceptable level of coding reliability.
We coded each tweet according to these categories: 1) Character count (excluding all non-textual elements such as links to videos and websites).
2) Generality or specificity of the tweeted attitude. A tweet was coded as general (Gen) when it generally included a candidate's name, good wishes for a candidate, a general eulogy, such as "the best" or "well-deserving," a pledge to vote for a specific candidate, or ill-wishes, such as, "I wish she wouldn't win." A tweet was coded as "specific" when it included an assessment of a candidate's skills, achievements, opinions, policies, and/or specific reasons for supporting or rejecting a candidate.
3) Peripheral processing in the tweet, which indicates that it is more focused on the personal or demographic characteristics of a candidate rather than her views, policies, and actions. A tweet was coded "peripheral" when it focused on a candidate's gender, caste, appearance, age, tribal affiliation, and social status. 4) Gender of tweeters (identified by reading their profile information). In cases where we did not find information about the gender of the tweeters, we read their tweets to identify whether they used masculine or feminine pronouns (which are easily identified in the Arabic language). Luckily, we did not find any accounts that were accessible only to followers. In previous research, the gender of Twitter users was identified through the user's first name (Bamman, et al. 2014;Burger, et al. 2011;Cunha, Magno, Gonçalves, Cambraia, & Almeida, 2014). However, using only the first name to infer the gender of a Twitter user was not an option for the present study, as it is not uncommon for Kuwaiti Twitter users, especially women, to use pseudonyms as their user names. 5) The general attitude of the tweets towards WCs. Each tweet was coded into three sub-categories: positive, neutral, and negative. The coders read each tweet to determine whether it was favoring a candidate, against her, or just making a neutral reference to her. Neutral references include such things like posting a link to a news story in traditional media, quoting a candidate, and posting an announcement for a candidate's rally or media appearance. Although such tweets could have been indicative of a positive leaning towards a candidate, we focused our analysis only on the explicit, verbalized expressions of the tweeters. Negative tweets were coded according to the existence of any of these aspects: the use of negative descriptions/attributes about any WC (such as liar, hypocrite, greedy, and nepotistic); mockery of the ethnicity, looks, and age of the WCs; criticism directed towards the views of any WC; expressing a preference for male candidates in referring to a WC; and general rejection of the candidacy of a WC (I won't vote for you). Positive tweets included positive descriptions (such as honorable, experienced, and having a good style), overt support and good wishes, appreciation of past political performance, preferring any WC over male candidates, complimenting a WC's looks and young age, and expressing general advocacy for a WC (such as encouraging tweeters to attend a WC's political rally and praying that she will win).
In addition, an extant textual analysis was applied to all the tweets to determine their main attitudinal themes without using a preexisting coding frame (Braun & Clarke 2006). Extant textual analysis is an unobtrusive method that allows more objectivity in data collection (Charmaz 2007) and inhibits the interference of researchers and, hence, the potential contamination of original opinions. This method of analysis in the present study is adopted to interpret the cultural symbols used by tweeters in expressing their attitudes towards WCs and make connections between such symbols to create themes that summarize and describe the data at hand (the tweets).
We compiled all the tweets verbatim and read them several times. Our aim was to provide a thematic description of the entire data set. Those tweets were coded according to the main theme of the views and attitudes expressed. When a tweet had more than one theme, each theme was coded separately. A data-driven, interpretive thematic analysis was conducted to identify the meanings of attributes and descriptions in Kuwaiti culture and determine the subthemes of each theme. The analysis was conducted in two phases. First, each tweet was coded according to its attitude and the properties of such an attitude. For instance, a tweet that read, "She needs a new face," was coded as a negative attitude towards the physical features of the WC. The general code for this tweet was "Criticism of physical traits." In the second phase, those general codes were recoded into main themes, and each main theme was coded into sub-themes (axial coding) depending on its components or how any WC was mentioned. "Criticism towards physical traits" was thematically coded as "Physical traits/negative" with an axial coding of physical traits such as the face, body weight, and age. Tables 2 and 3 provide an illustration of these two phases of textual analysis.

Research hypotheses
In the election held in 2013, only two out of eight WCs made it to Parliament, although women made up more than 53% of the electorate. These two candidates received the highest number of tweets about the WCs. In 2016, only one WC won a seat out of fifteen WCs. None of the 33 WCs won enough votes to be in Parliament in 2020. While the negative tweets outnumbered the positive ones (39% negative versus 33% positive), the women tweeters expressed significantly more positive attitudes towards WCs than the male tweeters did in 2013 (X 2 = 0.21, p = 0.001). Table 4 shows that this association was not significant in either the election of 2016 or 2020, even though the percentage of positive tweets about WCs remained higher among women tweeters in those two elections. This explains the significant positive association between gender and the tweeted attitudes towards WCs shown in Table  4. The data in that table supports RH1 and rejects its null hypothesis.  RH2 predicts a significant association between the use of specific/central cues in the tweeted attitudes towards WCs and the year of election. The data in Figure 1 shows a gradual increase in tweeting about the political statements and actions of WCs from one election to the next, while the emphasis on their demographic characteristics took the opposite direction. Nominal by nominal contingency coefficients between use of either central criteria or central cues in tweeted attitudes towards WCs and years were found significant (X 2 = 0.39 and 0.40, respectively, p< 0.001). The data rejects the null hypothesis of RH2.

Figure 1. Use of Cues in Tweeted Attitudes towards WCs According to Year of Election
RH3 predicts that the tweeted attitudes towards WCs in Kuwait will be significantly longer (in terms of the number of characters used) in 2016 and 2020 compared to 2013. The Kruskal-Wallis test shows a statistically significant association between the mean number of characters used in the tweeted attitudes towards WCs and the year of election (H = 29.91, df = 2, sig < 0.001). To test the statistical difference between each pair of election years, pairwise comparisons have been computed to test these expected differences. As shown in Table 5, the tweeted attitudes in 2020 were significantly longer than those used in either 2013 or 2016. However, the average number of characters in the tweeted attitudes during the 2016 elections was not statistically different from the average in 2013. This data partially supports RH3.

RQ: Tweeted qualities/traits of WCs
The data demonstrated five positive and five negative attitudinal themes. The coders counted the frequency of each theme and its subthemes to assess the weight of each attitudinal component. Figures 1 and 2

Figure 2. Visual Representation of the Weights of Positive Tweet Themes (Left) and Subthemes (Right) Figure 3. Visual Representation of the Weights of Negative Tweet Themes (Left)
and Subthemes (Right)

Positive Themes
The components of the positive attitudes expressed did not differ significantly between men and women, except for the use of emotion. Although the emotional component of the "general acceptance" theme was minimal, women were the main generators of this component. Six out of the seven emotional attitudes towards WCs were posted by women tweeters. This concurs with research done in the West, which shows that women are more prone on social media to using terms that convey affection, whereas men tend to use terms that assert their power (Bamman et al., 2014;Ottoni, Pesce, Las Casas, Franciscani, & Meira, 2013). The major theme of the positive attitudes towards WCs is too general, with a minimal cognitive component. Tweeters with positive attitudes primarily expressed their "best wishes" for a candidate by saying, "may Allah grant her victory" (Wish), expecting them to win (Expect), evaluating them in broad strokes, such as "she is the best" or "she deserves to win" (Eval), and expressing an emotional disposition, such as "I love her" and "happy she is running" (Emotion). This dominant positive theme reflects a subjective, tweeter-based general acceptance of a candidate rather than a candidate-based acceptance. Tweeters favored candidates out of love, support, and/or good wishes for them. All these are subjective favorability criteria that are not based on any specific objective traits of the WC.
The second dominant theme, however, did represent an objective, candidatebased acceptance, in which tweeters were favorable toward a candidate because of her political performance (Actions) and communication (COM) skills. Those skills of a WC have been demonstrated ether in a previous political job or during a previous parliamentary term. They were also expected to be demonstrated after the WC won the election. Satisfaction with both past political performance and communication skills were major subthemes of this positive theme. The tweeters placed a special emphasis on the communication skills of the candidates, praising them for their "eloquence," "sophistication as speakers," and "cogent logic while talking to the media." Some tweeters described the speaking style of one candidate as "making the liver of the listener cooler" (a Kuwaiti colloquial expression connoting a good mood and a positive reaction).
The third positive theme relates to how tweeters evaluated the personal traits of WCs. The most prevalent subtheme under this theme was the perceived sagacity and knowledge of a WC (Sage), followed by the moral traits of the WC, such as being "modest," "patriotic," and/or "religious." One tweeter addressed a WC by tweeting, "I wish you an early win of a seat; your mind is astute, and you have solutions and administrative skills." Other tweeters (especially women) favored the good looks of some WCs or praised them for being young and bringing new blood to the House.
The fourth attitudinal theme pertained to general advocacy for WCs, as several tweeters posted vows of voting for a WC or calls urging users to vote for a specific candidate (Rally) and/or to watch for her media and YouTube appearances. This reflects an active, positive attitude towards WCs in the hopes of garnering more support for them and helping them gain wider exposure in the public eye.
Gender-based tweets represent the least prevalent positive theme. Tweeters expressed a positive attitude towards a WC because she is "a daughter of men," "a daughter of Kuwait," "the pride of Kuwaiti women," or "better than many men." One candidate was supported because she was "more honorable than the mustache of her opponent." A man described one candidate as being "better than 50 mustaches" (the mustache used to be a common symbol of manhood in Arab culture). This emphasis on the superiority of WCs was not observed in the tweets posted by women. A smaller percentage of their positive tweets maintained that certain candidates were the best representatives of women.

Negative Themes
For the most part, the negative tweeted attitudes towards WCs employed a general (Gen) theme of sarcasm and vowing not to elect a WC or calling on others not to choose her (NoNo). Some tweeters posted reactionary non-word cues to ridicule a candidate by tweeting a political statement of hers and commenting on it by using an expression that mimicked the sound of laughter. The following tweets represent this theme: "When a TV show host asks candidate Awatef Elkallaf what her goals are if she wins in the elections, she replies, "When I win, I will think about that." Loooool." Other tweeters ridiculed some WCs for looking so helpless or even having hope of winning. One tweeter said mockingly, "Hope in life is sweet, and she [a candidate] will win 50 years after her death." As illustrated in Figure 3, the second dominant negative theme was directed towards the perceived ineptitude of WCs by using a more specific approach. This included under-evaluating their communication skills (COM) or terming their speeches "too loud," "too dramatic," "too arrogant," or "signifying dumb selfconfidence." One tweeter suggested that a WC needs to receive some tutoring to master the Kuwaiti colloquial. Other tweeters doubted the political competency of some WCs because of their "failed" political past. A few tweeters even wished that women's suffrage had never been granted in Kuwait. They also wished that a specific WC would never win the election.
WCs were also the targets of negativity based on their physical traits. This third negative theme had two main subthemes: one that critiqued the general appearance of the WCs (Looks), and another that lashed out at them because they looked like men (Manlike). Accordingly, some WCs were called "ugly," "junk-like," and even "dog-like." One woman wondered why all WCs "were so unattractive!" A video showing the false teeth of a candidate falling out during her televised interview was the object of mockery by some tweeting men. Others joked that the candidate belonged in a junkyard, called her "Ms. Slippers," or inquired about the "Pakistani barber who does her hair" (in Kuwaiti culture, Pakistani barbers are perceived as having the cheapest saloons, catering to clients of lower social status). One candidate was described as "having the mouth of an ugly Indonesian domestic servant" (Asian female domestic workers are a source of cheap labor found in most Kuwaiti households).
In the same vein, some negative tweets referred to WCs as having a man-like appearance or as being "manly" women. One candidate was called "Rambo" or "brother" and was ridiculed for looking like a famous Egyptian actor who is known for his film roles as a bully and a thug. A few tweeters even suggested that the same candidate needed to "trim her mustache." Other tweeters said that they were "having a problem identifying a WC among men." They suggested that men who vote for one WC will not incite the jealousy of their wives since that candidate can hardly be considered a woman.
The fourth negative theme pertains to some moral evaluations of WCs. A few of these tweets cast doubt on the morality of the actions and political statements of some WCs. Such tweets generally label some candidates as "corrupt" and "lacking integrity." Others referred to them as being hypocrites, inconsistent, and fake, such as, "Never heard her talk about buying out votes in her constituency," "You are just words, no action," and "What about your son getting a job with a tempting salary?" Notwithstanding, the least common negative theme pertains to the demographics of the WCs: their gender roles, religion, age, and social status. Some tweeters suggested that "grocery shopping," "staying home," "taking care of the kids," and "cooking" should be the more appropriate jobs for these WCs. One male tweeter thought that women make a crucial contribution to the fabric of society "because what they do is just give birth." Another tweeter imagined a winning WC "taking her stove to the House." Some men expressed their negative attitudes towards the ethnicity and religious sects of a specific WC. One Shiite WC was described as being more loyal to (Shiite) Iran than to (Sunni) Kuwait. One tweeter said about one WC that "her name is Safavid," (referring to the Shiite Dynasty that ruled Iran for centuries). A few tweeters criticized a candidate for wearing "non-Islamic" clothes. Others urged unveiled candidates to cover up, and veiled candidates to put on their hijab (head cover) properly. Age and social status were also two components of the negative "demographics" theme. Some tweeters, especially men, thought that certain WCs were "old hags" who should give youth a chance. Other tweets read, "I wish my auntie would withdraw and give youth a chance; she is a grandma born in 1946." "You are an oldie from hell and one of the setbacks of Kuwait." Several candidates were also called "Aunties" and were demeaned as "cheap" or coming from a lower social stratum or ridiculed as being "less sophisticated Bedouins."

DISCUSSION
The results partially agree with previous research in showing that Twitter users frequently tend to express sarcastic, negative attitudes towards WCs (Mejova et al., 2013) and that gender determines political attitudes (e.g., Sidanius, 1997a;Sidanius & Pratto, 1999;Southern & Harmer, 2021). The analysis shows that women have more positive and less negative attitudes towards WCs; this is also in line with the literature, which shows that women tend to possess more positive attitudes towards the issue of leadership by women than men (Adeyemi-Bello & Tomkiewicz, 1996;Al-Khoori, 2011;Alibeli, 2015;and Mostafa, 2010). However, this association between gender and attitudes was only significant in the election of 2013. In the subsequent two elections, no statistically significant association between gender and attitudes was detected. The results suggest that gender is not expected to remain a significant factor in predicting the attitudes of Kuwaiti tweeters towards WCs. The results also show the importance of adopting a longitudinal to examine the association between gender and political attitudes in patriarchal societies. A closer examination of the components of the tweeted attitudes towards WCs in Kuwait over more than one election term shows that these attitudes took a "cognitive" turn, especially in the latest election in 2020. In the first election, both positive and negative tweeted attitudes towards WCs included primarily generic terms, such as expressing good wishes, welcoming the nomination of a candidate, expecting their victory, or vowing not to vote for a WC. In the following two elections, those attitudes turned out to be more elaborate cognitively. Tweeters became more cognizant of the performance of a specific WC in parliament and more critical of both the communication skills of WCs and their political statements. Tweeters also reflected more on the integrity and credibility of WCs in the last election. Even though the credibility of the source has been studied as a peripheral cue in the literature (Wu & Schaffer, 1987;Zhu, Xie, & Gan, 2011), it is an indication that tweeters pay more attention to the objective cues in WCs (such as their political performance) than their subjective cues, such as their looks and appearance. Several tweeters announced their support of specific WCs because of their patriotic political speech that "reveals corruption" and "focuses on development." Tweeters tended to show a growing interest in the message of the WCs over time and a decreasing interest in their physical, and gender-based characteristics. In other words, the longer the WCs maintain their civic engagement, the less likely they will be evaluated according to gender-based criteria or "othered" by the socially dominant groups.
This increasing prevalence of "objective" evaluation of WCs in the tweets explains the significant growth in the length of those tweets during the second and third elections. The results showed that the tweeted attitudes posted in 2020 were significantly longer (in terms of characters) compared to those posted in either 2013 or 2016. Several tweeted attitudes towards WCs in 2013 consisted of a few words or sometimes one word to express either general good wishes or negative descriptions of a WC. Although the tweeted attitudes posted in 2016 were longer than those posted in 2013, this difference was not statistically significant. However, the length of the tweeted attitudes in 2020 was significantly higher than the posts in both 2013 and 2016. This difference is hardly attributed solely to the growing interest of tweeters in the political communication and performance of WCs. This could also be attributed to the decision made by Twitter in November 2017 to increase the maximum allowable length of tweets from 140 to 280 characters. Accordingly, tweeters in 2020 could abandon this "telegraphic" writing style dictated by the maximum limit of 140 characters and ramble more freely after this maximum has been doubled. However, a second examination of the data reveals another factor that might explain the varying lengths of tweets across the three elections. The two WCs who ran in the three elections (Safaa and Anwar) received unequal lengths of tweets. Figure 4 shows that Safaa, who won in both 2013 and 2016, had a gradual increase in the length of the tweeted attitudes towards her. Anwar never made it to Parliament in any election, and the length of the tweets posted about her took an irregular pattern.

Figure 4. Average No. of Characters in Tweeted Attitudes towards Two WCs in Three Elections
The length of both negative and positive tweeted attitudes towards Safaa (a PM for two terms) in 2016 (one year before Twitter increased the maximum length of tweets from 140 to 280 characters) was significantly larger than in 2013 (t = 6.0, adj sig. = 0.04). This seems to demonstrate that the formal, continued civic engagement of this WC (rather than the length constraint of tweets) triggered more central processing of her political communication, which was reflected in the longer, information-based tweeted attitudes about her from one election to the subsequent. The other candidate (Anwar) had intermittent exposure to the public eye, mainly during the election campaigns, which limited the amount of information available about her. On the other hand, research comparing the impact of Twitter's move to increase its length constraint on the linguistic aspects of tweets found that tweets constrained by a 140-character limit contained fewer hashtags, more abbreviations, and more symbols (Gligorić et al., 2018), but the authors did not report any significant increase in the quantity or quality of opinions expressed. This suggests that the continued political engagement of WCs determines both the length and cognitive elaboration of the tweeted attitudes towards them.
Those tweeted attitudes were also thematically rich. The data used for the present study provides limited information about the tweeters. Though there is no reliable way to measure their SDO levels, their tweets reveal how they perceived and assessed WCs. Those WCs were substantially supported based on general, Safaa Anwar affective reactions to them. Among the five themes that summarized tweeted positive attitudes, only one pertained to their skills in politics and communication, and yet that one was not the most dominant theme. A few male tweeters supported WCs because they would be "the best to represent the women of Kuwait." This implies that WCs, if elected, will not topple the long-standing societal power hierarchy existing in the country, but will have a "segregated" position of power to represent their fellow female voters effectively. Other men supported specific WCs because they believed that they acted in as "manly" a fashion as their male counterparts. To justify their public support for those candidates, those tweeters had to redefine the WCs to perceive them as acting similarly to men; this suggests that WCs were (subconsciously perhaps) perceived as belonging to the powerful, dominant level of the social hierarchy in order for them to be supported. This perception might have neutralized the inclination of individuals high in SDO to deny women a higher social status and/or leadership roles. The acceptance of WCs by male tweeters on the basis that those candidates are compared favorably to men suggests a subconscious legitimization of the consensually shared belief about male dominance/superiority in Kuwait. Several male tweeters even expressed their support for a WC because they felt she had been "victimized" in Parliament when she was previously serving in it. Accordingly, it is not just gender, age, clan, ethnicity, and social class that determine SDO, as was suggested by Sidanius and Pratto (1999), but also perceptual and cultural processes that explain how the dominant gender could end up supporting politically the opposite, subordinate gender. In this light, the tweets favoring WCs based on feminism-related ideas were limited even among female tweeters. This finding is not in line with previous research, which shows that women are more likely to vote for WCs who are perceived as "profeminist" (Plutzer and Zipp, 1996;Campbell & Heath, 2017). As some tweeters have suggested, WCs in patriarchal societies are expected to look feminine, but not to adopt a feminist agenda. To the contrary, they are praised on the basis that they act more "manly" than their male counterparts. This dilemma facing women leaders seems to be universal. Masculinity and power seem to be perceived as concomitant. Mavin (2009) argued that women leaders who seek to be credible will not identify with the gender stereotype of women who are stereotypically linked to both appearance and materialistic artifacts, and not to ideas. The media reconstructions of women politicians significantly criticize their appearance and sometimes lash out at them for looking "too sexy" or "too feminine" (Mavin, Bryans, & Cunningham, 2010).
The data also shows that tweeters tend to reject WCs in Kuwait because of who they are (subjective cues) or how they act or communicate (objective reasoning). In return, tweeters accept WCs mainly because of how they feel about those candidates (subjective reasoning). This seemingly unjustified, disengaged public support represents "low-commitment active citizenship" (Stern, Dietz, Abel, Guagnano, & Kalof, 1999, p. 82). Given this weak commitment, the chances for attitudes to be reflected in behavior are not high. With less motivation and less elaborate cognitive processing of the "qualities" of WCs, attitudes are less expected to be translated into behavior, viz., voting.
The "social role" component of the "demographics" negative theme is supported by the stipulations of SDO. Some tweeters identified the role of family care-taker as an alternative role for WCs. Although women in patriarchal societies are still oppressed by such stereotypical depictions, some tweeting women take part in this societal oppression. Voters associated care-giving attributes, nurturing roles, and appearance-related traits with the WCs. In doing so, they have subconsciously dissociated WCs from the attributes of competence and leadership and tended to maintain the long-standing hierarchy in society. This unspoken agreement between superior and inferior genders over the social and occupational roles of women represents another legitimizing force for the "stereotypes and cultural ideologies" (Pratto et al., 2006, p. 275) in Kuwaiti society.
The negative theme of demographics also had a religious component. A few tweeters rejected some candidates based on their affiliation with the minority religious caste in the country. Other tweeters criticized WCs for either not wearing the Islamic head covering, or not wearing it properly. Kuwait falls within the Muslim "patriarchal belt" that extends from North Africa through the Muslim Middle East and to South and East Asia (Offenhauer, 2005). While gender segregation in all stages of education is legally enforced in Kuwait, wearing a veil is not obligatory. A few tweeters rejected the potential political leadership of these WCs based on religious teachings that might deny such a leadership. The Islamic religion has been suggested to be the determining factor in the division of labor and the acceptance of gendered norms and beliefs (Norris, 2011). However, this argument ignores the social and cultural individuality found among various Islamic populations. For instance, anthropologist Mounira Charrad (2001) analyzed the status of women in the Arab nations of Morocco, Algeria, and Tunisia and reported that women's legal rights and political participation in these countries are determined more by kinship relations than by the Islamic faith. Yet, this is not to say that religion does not remain a potentially significant factor in the determination of leadership and power relationships within traditional, patriarchal societies. To examine this issue more fully, we need a more detailed empirical test of the relationship between religiosity and SDO.
Overall, the political tweets about WCs in Kuwait reveal the complexity of the attitudes towards them and explain some of the gender differences and similarities in accepting or rejecting them. Those political tweets have also shown that they might be used as a predictor of power in Kuwait, as the two WCs who won the election in 2013 and the WC who won in 2016 also received the highest number of mentions on Twitter, compared to the other WCs. This replicates some findings in the previous literature (O'Connor et al., 2010;Tumasjan et al., 2010).

CONCLUSION & IMPLICATIONS FOR FUTURE RESEARCH
The present study shows that gender partially determines political and social attitudes and explains how dominant groups may undergo perceptual modifications that will allow them to support the political leadership of the subordinate group. The political longivity and civic engagement of women in this patriarchal culture is expected to normalize their acceptance as candidates rather than "other" them based on their gender. The results also underscore the important role of attitudes in explaining gender inequality. The dominant perception of WCs within this techsavvy, educated Kuwaiti citizenry is that these candidates are mainly women whose first obligation is their family and whose most important trait is demographic, rather than intellectual. Consequently, women leaders in patriarchal societies are praised for acting like men. Such long-established, basic inequality in this traditional society has maintained the distribution of power and reinforced its acceptance, even by less powerful groups. In order for WCs to be accepted by those voters who have higher levels of SDO, they must identify with the dominant hierarchy rather than present themselves as advocates for feminist rights and gender equality. If they wish to seek leadership roles, they should not act as feminists.
More components of SDO could be explored by studying transitional societies that are passing through modernization processes while still maintaining their long-standing cultural and religious foundations. Such components might prove more consequential in predicting and explaining ingroup inequality and power distribution. These components include, but are not limited to, the use of information technology, study abroad, interaction with different cultures through travel, and the adoption of liberal ideas. An understanding of the impact of such dynamics will contribute to the degree to which the social dominance theory may be generalized and applied to understanding intergroup relations within a dynamic, modernizing society.

LIMITATIONS
As with any content analysis study, the research point, variables, and analysis are restricted by the content of the tweets in the present study. Innately, tweets are "telegraphic" forms of expression that are limited by a specific number of characters, which further constrains the depth of information and reflection shared. This datarelated limitation makes it inevitable for the researcher to analyze the tweeted attitude in isolation from the demographic and psychographic traits of the tweeters. Significant variables that might predict attitudes towards the leadership of women in Kuwait are tribal and religious affiliations. It was hard to decipher such variables from the data used in the present study, which has limited the researcher's ability to discuss and explain the results from a broader perspective. Even though the theory of social dominance was adopted as a theoretical framework, the data inherently did not provide information that could be used in measuring the tweeters' levels of social dominance, which has represented another limitation for the study. Moreover, the lack of previous studies in the same field conducted in Kuwait restricted the scope of the present study.