Infante & Wigley (1986) have defined the verbal aggressiveness as an attack in the perception of individual without, or with, simultaneous attack to the attitude that the person takes toward a communication issue. Verbal aggressiveness functions as a catalyst for physical aggressiveness (Infante, Chandler, & Rudd, 1989; Infante & Wigley, 1986; Sabourin, Infante, & Rudd, 1993) . The negative effects of verbal aggressiveness may be more disturbing even than physical aggression because its effects may last longer (Infante, 1995) . The majority of research shows that the verbal aggressiveness is considered to be destructive (Bekiari, 2016; Bekiari, Kokaridas, & Sakellariou, 2005, 2006; Bekiari, Perkos, & Gerodimos, 2015; Infante, 1995; Myers, Edwards, Wahl, & Martin, 2007; Myers & Rocca, 2000) . It has been found that there is a negative relationship between perceived verbal aggressiveness of the teacher with the affective learning (Bekiari, 2012; Bekiari & Manoli, 2016; Myers & Knox, 1999, 2000; Myers & Rocca, 2000, 2001) . It is also negatively correlated with students’ motivation (Bekiari, 2014; Bekiari & Sakellariou, 2002; Bekiari & Syrmbas, 2015; Goodboy & Bolkan, 2009; Myers et al., 2007; Rocca, 2004, 2008) and fair play behaviors (Hassandra, Bekiari, & Sakellariou, 2007) . Also, verbal aggressiveness affects on the discipline in class (Bekiari et al., 2006; Bekiari & Tsiana, 2016) causing behaviours ( Claus, Booth-Butterfield, & Chory, 2012; Kennedy-Lightsey & Myers, 2009) , reducing teacher’s reliability (Infante, 1985; Infante et al., 1992) and interpersonal attraction (Syrmpas & Bekiari, 2015) . Students declare lower levels of satisfaction and learning (Bekiari, 2014; Bekiari, Perkos, et al., 2015; Manoli & Bekiari, 2015; Mazer & Stowe, 2015; Myers, 2002; Myers et al., 2007; Myers & Knox, 1999, 2000; Myers & Rocca, 2001; Schrodt, 2003) and present lower academic achievement scores (Uludag, 2013; Yaratan & Uludag, 2012) . Also, verbal aggressiveness is negatively associated with interest (Weiss & Houser, 2007) , self-esteem (Buford, 2010; Infante & Wigley, 1986; Schrodt, 2003) , behaviour, thinking and motivation (Bekiari, Perkos, et al., 2015; Hasanagas & Bekiari, 2015; Mazer & Stowe, 2015) . Verbal aggressiveness practised by instructors negatively affects the relationship with their students (Frymier & Wanzer, 2006; Manoli & Bekiari, 2015; Martin & Myers, 2006) or athletes in training context (Bekiari, Digelidis, & Sakellariou, 2006; Bekiari, Patsiaouras, Kokaridas, & Sakellariou, 2006) . The use of verbal aggressiveness by teachers has adverse effects for themselves, as they appear to be less communicative (Myers et al., 2007) . Students present a less scientific approach and sociability (Hasanagas & Bekiari, 2015; Myers & Rocca, 2000; Rocca & McCroskey, 1999) while they report burnout (Avtgis & Rancer, 2008; Yaratan & Uludag, 2012) . Instead, the lack of verbal aggressiveness enhances motivation, positive perception of classroom climate and emotional learning (Mazer & Stowe, 2015) . Several studies describe the negative effects of using verbal aggressiveness from their chiefs to subordinates at the workplace (Infante, Anderson, Martin, Herington, & Kim, 1993; Infante & Gorden, 1985a, 1985b; Wheeless, Wheeless, & Howard, 1984) but few are those who associate the verbal aggressiveness with the leadership.

Research on leadership mainly focuses on transformational leaders who could change deep structures, major procedures or the overall organisation (Van Wart, 2003) . Bass (1985) model examines the relationship between transactional and transformational leadership. Transactional leadership based on a relation transaction between directors and members/employees, promoting them with rewards (Burns, 1978; Pashiardis, 2004) . It consists in an administration of minimal interventions (laissez-faire), based on the exceptions management (management by exception) and contingent reward (Robbins & Judge, 2012) . In contrast, transformational leadership is a special leadership style applied by superiors to motivate subordinates to operate at a higher level, offering spiritual challenges and paying attention to their individual needs by creating a supportive environment (Belias & Koustelios, 2014; Gkolia, Belias, & Koustelios, 2014b) . Transformational leadership is based on trust and respect, engaging a parallel commitment to a vision creating incentives for employees (Eliophotou-Menon, 2011) . The transformational leadership is grounded on the following characteristics: idealized influence, inspirational motivation, intellectual stimulation and personalized care (individualized consideration) (Bass, 1991, 1999; Bass & Stogdill, 1990; Robbins & Judge, 2012) . According to the afore-mentioned assumptions, leadership can be regarded as process influencing thought, emotions, attitudes, and behaviours of a group of people from the leader. In this process, the led members are supposed to willingly and appropriately cooperate in order to achieve their best performance (Bouradas, 2005) .

Aim of this research is to suggest determinants and effects of verbal aggressiveness and leadership, measured in term of its effect, leadership, as well as to analyze relations between leadership and verbal aggressiveness. The expected innovation lies in the implementation of complete network analysis which is expected to depict more objectively the structures of leadership and verbal aggressiveness.

Complete network analysis focuses on the interaction of participants and uses a set of metrics for the analysis of their relationships (Scott, 2000; Wasserman & Faust, 1994) . The following network analysis metrics have been calculated by Visone software:

1) In-degree is defined as the percentage of edges which ends in a node. It can be interpreted as an occasional hierarchy position.

2) Katz status expresses chains of successive relations using power series. Thus, it can be interpreted as an accumulative hierarchy position. It signifies situations much deeper than indegree.

3) Pagerank is based on the transferred value from one node to others and interpreted as distributive hierarchy position. It is similar to Katz status but it restricts outliers. Furthermore, it prevents hierarchy deformations induced by Katz status.

4) Authority indicates the nodes who attracting most links from other nodes, among those ones who intensively seek to maintain links.

All aforementioned metrics have been repeatedly used and interpreted in real-empirical context (Bekiari, Hasanagas, Theocharis, Kefalas, & Vasilou, 2015; Bekiari & Hasanagas, 2015, 2016a, 2016b; Bekiari & Spyropoulou, 2016; Hasanagas & Bekiari, 2015; Theocharis & Bekiari, 2016) .

Classes of public secondary schools (Gymnasium and Lyceum) in Trikala have been surveyed as network samples. Each class consisted a network, where the pupils were the nodes and the links among them were their relations (trust, company, conflicts etc). In other words, such a network sampling is a cluster sampling including the survey of links among the population (Farmakis, 2000) . The set was calculated to be composed by 171 individuals. There were 128 pupils (class A: 20 pupils, female: 5 male: 15, class B: 21 pupils, female: 10 male: 11, class C: 22 pupils, female: 8, male: 14, class D: 21 pupils, female: 10 male: 11, class E: 21 pupils, female: 8 male: 13, class F: 24 pupils, female: 15 male: 9). Their age varied from 13 to 17. Also, there were 43 teachers (female: 22, male: 21). Their age varied from 32 to 62. Precondition to participate a teacher was to teach over two hours per week in the class involved in the research.

The questionnaire consisted of two parts: a) non-network variables (e.g. gender, birth year, teaching years etc.), and b) network variables (power dimensions, verbal aggressiveness and leadership).

The part b of the questionnaire was based on the Verbal Aggressiveness Scale (Bekiari & Digelidis, 2015) in order to measure verbal aggressiveness. Preliminary examination Scale (Bekiari & Digelidis, 2015) supported the psychometric properties of the instrument. In particular, confirmatory factor analysis indicated satisfactory fit indices (CFI: .97, SRMR: .02), and internal consistency of the scale (α = .96). The scale consisted of eight items (e.g., “insults students or teachers,” “makes negative judgments of ability”).

The part b of the questionnaire was based on the Principal Leadership Questionnaire which consisted of twenty four items (Gkolia, Belias, & Koustelios, 2014a) in order to measure transformational leadership. Preliminary examination supported the psychometric properties of the instrument. In particular, confirmatory factor analysis indicated satisfactory fit indices (CFI = .97, RMSEA = .06), and internal consistency of the scale (α: from .60 to .81). The scale consisted of four factors (leader models behavior, fosters commitment, provides individual support, and holds high expectations) (e.g. “treats me as an individual with unique needs”, “leads by ‘doing’ rather than simply by ‘telling’)”.

Additional questions of the power dimensions (part b of the questionnaire) concerning trust (e.g., advising about humanities and natural Sciences), socialization patterns (e.g., companion within and outside faculty) and study and general cooperation. The power dimensions based on Popitz model (1992) has been used in previous researches ( Bekiari & Hasanagas, 2015, 2016a, 2016d; Bekiari & Spyropoulou, 2016; Hasanagas & Bekiari, 2015) .

The network data were processed by Visone 2.16 software. Several indicators were computed (indegree, status, pagerank, authority, density). Also, different networks structure was depicted.

Thereafter, both the network and non-network data were entered into SPSS 21. Spearman bivariate correlation (to avoid outlier effect) and principal components analysis (for formulating typology) were conducted.

In Figures 1-3 several structures (status of Katz, pagerank and authority) of verbal aggressiveness and leadership are presented. Differences can be observed between the structures of verbal aggressiveness and leadership networks. The networks do not have the same density and networks of leadership are denser than those of verbal aggressiveness. Density was the measure depicting the intensity of the relations between individuals in classes. Density can be utilized as a measure of cohesion, at least 0 case of non-conflicting relations (Borgatti, Everett, & Johnson, 2013) and has been used to examine leadership relations in classes (Carson, Tesluk, & Marrone, 2007) . More specifically, the density of each network is the portion of the potential connections in a network that the

actual connections (Sparrowe, Liden, Wayne, & Kraimer, 2001) . The observed conclusion can be confirmed and as shown in the Table 1, the average density of verbal aggression was .030 while leadership has average density .090 and the proposed as president was .091. Additionally, the average indegree-outdegree of leadership networks was bigger than verbal aggression one as shown in Table 2.

Additionally, differences can be observed in the networks observing various indicators (pagerank, status, authority etc.). These differences were expected because various indicators reveal no identical properties and meanings. So, it is useful to apply several indicators in social network analysis and not only one (Bekiari & Hasanagas, 2015).

In Table 3 it is noticeable that numerous significant correlations occur between preferred class president and leader characteristics (.230 to .564) or of verbal aggressiveness attributes (.174 to .303) or of social interactions (.268 to .370).

All leadership variables were significant correlated with the property of being preferred for president. The results showed a positive correlation between this property and the various characteristics of leadership (leader acts, leader is an example, does not believe in mediocrity, personal approach and communication). No significant correlation had been found between the property of being preferred president and “have high expectations of a leader”. This can be explained by the nature of the position of the class president. It is sure that anyone of classmates expects the president to be the most capable but also knows the limitation of the position, so they don’t have high expectation from him. Also, all

**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

other characteristics of power were significant correlated with the preferred class president. The cooperation, the socialization, possible future business cooperation, advisory on social matters or technical themes is positive correlated with the preference of pupils for the class president. Also, the property of being preferred for class president is positive correlated with some verbal aggression variables. If someone has threatened a classmate or keeps him out of his social surrounding he creates a profile of detached person. This profile can be enforced if the person believes that the other one use negative comments for him in his surroundings. Additionally, the use of verbal aggression in run around situations is positive correlated with the preference of class president. This can be explained as a result of the fact that pupils use verbal aggression as a technique for choosing someone from their friends as a president.