Psychometric properties of the Portuguese version of the Global Transformational Leadership (GTL) scale

‘What makes an excellent leader?’ is the first of many questions that have arisen over time, given the focus on leadership in all organizations and societies. There is no society with complete absence of leadership, and leadership itself is influenced by the organizations and the society that surround it ( Bass, 1997 ). In the middle of the 20th century, leadership became a prominent research area ( Judge & Bono, 2000 ), and is one of the most studied subjects in the Social Sciences, with numerous papers, perspectives, analysis levels, and theoretical and methodological orientation lines associated with this concept.

In the middle of the 80s an approach emerged that ended up dominating the interest of a large part of researchers – transformational leadership – which brought about a renewed interest in the study of leadership and changed its underlying paradigm ( Hunt, 1999 ). Until the mid-1980s, according to Bass (1985) , existing theories about leadership focused primarily on the clarification of the goal and the role of the employee and also on the way the leader compensated or sanctioned his behaviour. The concept of transformational leadership was d by Burns (1978) , simultaneously with that of transactional leadership. Later on, these concepts were the object of operationalization and refinement by Bass (1985) through the Multifactor Leadership Theory ( Tejeda, Scandura, & Pillai, 2001 ), triggering the change from the ‘transactional’ paradigm to the ‘transformational-transactional’ one. According to this theory, leadership started to be conceived of not only as a question of contingency reinforcement of employees’ performance by the transactional leader, but also as a way of moving the employees beyond their personal interests towards the wellness of the group, organization, or society by the transformational leader.

The most cited comprehensive leadership theory, the Multifactor Leadership Theory ( Judge & Piccolo, 2004 ), s a number of the leader's behaviours ( Bass, 1985; Bass & Avolio, 1993; Yukl, 1994 ) and, extending Burn's (1978) work, divides leadership into three second-order domains: transformational, transactional, and laissez-faire. Transformational leadership, on which this paper focuses, is characterized as an adaptable and flexible leadership style that encourages followers to do more than they originally expected, broadening and changing their interests and leading to conscientiousness and acceptance of the group's purposes ( Bass, 1985; Bass, Avolio, Jung, & Berson, 2003; Bass & Riggio, 2006 ). The emphasis falls on collective motivations, the main characteristic of this leadership type lies – the mobilization towards directing the interests to a common goal ( Bass, 1999 ). In this context, transformational leadership is a process of influence which is capable of moving social systems, with the objective of making the employees become agents of change for the organization's movement and development ( Burns, 1978).

Podsakoff, McKenzie, Moorman, and Fetter (1990) , through a literature review about transformational leadership, concluded that the concept can be summarized in six fundamental behaviours: identifying and articulating a vision; providing an appropriate model; promoting the acceptance of group goals; developing high performance expectations; giving individual support; and providing intellectual stimulation to staff. Inspired by this model, Carless, Wearing, and Mann (2000) described transformational leaders as those who exhibit the following seven behaviours: they 1) communicate a vision (develop an image of the future of the organization and communicate it); 2) develop staff (diagnose the needs and capacities of each employee and express an individual interest in each one); 3) provide support for them to work towards their objectives through coordinated team work; 4) empower staff (give them authority to implement policies and supporting their decisions); 5) are innovative (use non-conventional strategies to achieve their goals); 6) lead by example (adopt congruent behaviours with the attitudes and values they support); and 7) are charismatic (inspire employees to transcend their personal interests and limitations, develop a conscience of the collective interests and guide them to achieve extraordinary goals).

A large part of contemporary research about leadership focuses on the effects of transformational leadership on a set of individual and group variables. Therefore, at the individual level, the studies performed have supported a positive effect of this kind of leadership on the commitment, motivation, and performance of employees (e.g., Arnold, Turner, Barling, Kelloway, & Mckee, 2007; DeGroot, Kikker, & Cross, 2000; Dumdum, Low, & Avolio, 2002; Fuller, Patterson, Hester, & Stringer, 1996; Judge & Piccolo, 2004; Meyer, Stanley, Herscovitch, & Topolnytsky, 2002; Nielsen, Randall, Yarker, & Brenner, 2008; Wang, Oh, Courtright, & Colbert, 2011; Wolfram & Mohr, 2008 ). In a recent meta-analysis, Wang et al. (2011) showed that there is a stronger effect of this leadership type on employees’ attitudes and motivation than on their performance. At the group level, research suggests a positive association of transformational leadership with the quality of group experience (e.g., Braun, Peus, Weisweiler, & Frey, 2013; Jung & Sosik, 2002; Long, Yusof, Kowang, & Heng, 2014; Nielsen, Yarker, Randall, & Munir, 2009 ) and with group autonomy (e.g., Butler, Cantrell, & Flick, 1999; Wofford, Whittington, & Goodwin, 2001 ).

Different measures can be found in the literature that assess transformational leadership behaviours, such as the Multifactor Leadership Questionnaire (MLQ; Avolio, Bass, & Jung, 1995 ), the Conger-Kanungo scale ( Conger & Kanungo, 1994 ), the Leadership Practices Inventory (LPI; Kouzes & Posner, 1990 ), and the Transformational Leadership Questionnaire (TLQ; Alban-Metcalfe & Alimo-Metcalfe, 2000 ). These scales have in common the fact that they are relatively long and therefore time-consuming to complete. In order to overcome this difficulty, Carless et al. (2000) proposed a short, practical, and equally valid instrument to measure transformational leadership – the Global Transformational Leadership scale (GTL) – which is based on the seven behaviours of transformational leadership defined by the authors and referred to above. The GTL is composed by seven items (one item for each behaviour) that evaluate the frequency of transformational leadership behaviours exhibited by the leader, based on a 5-point Likert scale (1 = rarely or never to 5 = very frequently or always ). In the original study, the authors, based on a sample of 1,506 participants of an international bank organization, performed an Exploratory Factor Analysis (EFA) that led to the emergence of one single factor. The unidimensional structure was later confirmed by a Confirmatory Factor Analysis (CFA). The authors also evaluated the convergent validity, via the comparison of the GTL results with LPI and MLQ, and the discriminant validity, via the comparison of results of groups of managers in which different scores were expected. The results obtained support the scale's convergent and discriminant validities. Finally, the internal consistence, estimated via Cronbach's alpha (? = .93), revealed the presence of a scale with a reliability level much higher than the cut-off point of .70 proposed by Nunnally (1978).

Results Statistical Procedure

Firstly, missing values were analysed and no survey contained a number of non-responses equal to or greater than 10% ( Bryman & Cramer, 2004 ). The total sample was randomly divided into two sub-samples (each one with 228 participants) in order to obtain evidence of cross-validity. In the first sub-sample, composed of 48 teams, EFA was performed by applying the principal component analysis. In the second sub-sample, consisting of 22 teams, two procedures were adopted in order to validate the factorial structure obtained from the first sub-sample: a new EFA was conducted and Tucker's (1951) congruence coefficient, first developed by Burt (1948) and popularized by Tucker, was estimated, and a CFA was performed. Reliability, measured via Cronbach's alpha, was computed for each of the sub-samples and also for the total sample.

Psychometric Analysis of the GTL Items deive analysis

The deive statistics of the items were computed for the first and second sub-samples and for the total sample ( N = 456) (mean, standard deviation, minimum and maximum, frequencies for response option, and skewness and kurtosis), as can be seen in Table 1 . It was observed that the subjects’ answers were distributed over the five response options in each of the scale items and that the percentage of responses in one option was less than 40% in all items. Results concerning the univariate normality revealed no severe normality violations, since absolute skewness (Sk) and kurtosis (K) values were lower than 3.0 and 8.0, respectively ( Kline, 2005 ) (in absolute values, the highest value for skewness was 0.84 on item 7 from the first sub-sample, and for kurtosis it was 0.73 on item 2 also from the first sub-sample). These data indicate an adequate overall variability of the answers given by the subjects and an adequate discriminative power of the items.

Table 1

GTL: Items Deive Analysis.

Item n°	First sub-sample				Second sub-sample				Total Sample
My team leader...	Mean	SD	Sk	K	Mean	SD	Sk	K	Mean	SD	Sk	K
1. communicates a clear and positive vision of the future.	3.90	0.84	?0.44	0.14	3.41	1.04	?0.49	?0.03	3.66	0.94	?0.46	0.05
2. treats staff as individuals, supports and encourages their development.	4.00	0.87	?0.78	0.73	3.49	1.05	?0.55	?0.03	3.75	0.96	?0.67	0.35
3. gives encouragement and recognition to staff.	3.89	0.94	?0.60	?0.06	3.42	1.08	?0.38	?0.32	3.66	1.01	?0.49	?0.19
4. fosters trust, involvement and cooperation among team members.	3.98	0.91	?0.63	?0.02	3.52	1.12	?0.51	?0.34	3.75	1.02	?0.57	?0.18
5. encourages thinking about problems in new ways and questions assumptions.	3.91	0.93	?0.63	0.09	3.55	0.99	?0.42	?0.14	3.73	0.96	?0.53	?0.03
6. is clear about his/her values and practices which he/she preaches.	4.07	0.87	?0.62	?0.16	3.46	1.10	?0.49	?0.33	3.77	0.99	?0.55	?0.25
7. instils pride and respect in others and inspires me by being highly competent.	3.99	0.95	?0.84	0.44	3.44	1.12	?0.51	?0.29	3.72	1.04	?0.67	0.07

First sub-sample analysis

EFA was performed with the SPSS program (IBM SPSS Statistics 22). The viability of EFA was assessed by means of the following three indicators: Kaiser-Meyer-Olkin's Measure of Sampling Adequacy (KMO); Bartlett's test of sphericity; and the Anti-image Matrix. Results from KMO (.93); Bartlett's test, ? 2(21) = 1118.957, p < .001; and Anti-image Matrix (the values on the main diagonal vary between .92 and .95 and all values outside this diagonal were small) confirmed the adequacy of the matrix for factor analysis.

The seven items of the GTL were then submitted to a principal component analysis, with a free ion of factors, which conduced to a one-factor solution responsible for 70.23% of the variance and with an eigenvalue of 5.76. As can be seen in Table 2 , all items presented loadings above .78 and communalities equal to or above .61. As regards internal consistency, the Cronbach's alpha obtained was .93.

Table 2

GTL: Factor Loadings and Communalities (h 2).

Item n°	Factor Loading	h2
My team leader...	Factor 1
1. communicates a clear and positive vision of the future.	.783	.613
2. treats staff as individuals, supports and encourages their development.	.832	.693
3. gives encouragement and recognition to staff.	.867	.752
4. fosters trust, involvement and cooperation among team members.	.849	.720
5. encourages thinking about problems in new ways and questions assumptions.	.832	.692
6. is clear about his/her values and practices which he/she preaches.	.827	.685
7. instils pride and respect in others and inspires me by being highly competent.	.872	.761

Second sub-sample analysis

In order to quantify the degree of similarity between the factor solutions from each one of the samples, a new EFA was performed in the second sub-sample and Tucker's congruence coefficient ( Tucker, 1951 ) was estimated. The solution obtained from the second sub-sample revealed a high degree of congruence with the solution obtained from the first sub-sample (Tucker's congruence coef?cient = .999).

The one-dimensional structure obtained from EFA was then tested with CFA, using the maximum likelihood estimation method, with AMOS software (IBM SPSS AMOS 22.0). According to Byrne (2010) and Kline (2005) , the evaluation of the model fit should derive from a variety of perspectives and be based on the adequacy of the model as a whole and on the parameter estimates.

In the evaluation of the model as a whole, the principal goodness-of-fit statistics presented in the literature were considered (e.g., Byrne, 2010; Kline, 2005): a) ?2 statistic; b) ?2/gl , in which values between 2 and 3 are recommended, with values up to 5 being admissible; c) the Comparative Fit Index (CFI), which ranges from zero to 1.00, with a value >.95 being considered representative of a well-fitting model; the Root Mean Square Error of Approximation (RMSEA), with values less than .05 being indicators of good fit, and values as high as .10 being considerable acceptable. The results obtained from CFA indicated that the model's goodness-of-fit was acceptable. Thus, although the ? 2 statistic was significant, ?2(14, N = 228) = 43.888, p < .001, the ?2/gl (3.14) was below the reference value, and CFI presented a very good value (CFI = .98). Additionally, RMSEA, with a value of .097 and a statistically significant associated confidence interval between .065 and .130, indicated an acceptable fit of the model to the population (though very close to the threshold proposed in the literature). Since, overall, the fit indexes presented acceptable values, we decided to not re-specify the model.

Concerning the evaluation of the adequacy of the parameter estimates (feasibility and statistical significance of the parameter estimates and appropriateness of the standard errors), results were very favourable. Figure 1 shows the retained factor structure, in which all the estimated parameters are statistically significant and the loadings of the items are much higher than .50.

Figure 1

GTL Factor Structure (standardized solution in values concerning errors are proportions of variance not explained).

The evaluation of the internal consistency in the second sub-sample, as in the first sub-sample, presented adequate values (? = .96).

Nomological Validity

Nomological validity was evaluated, considering the total sample, via the analysis of GTL correlations with quality of group experience and group autonomy. Since, in the present study, the unit of analysis was the group rather than the individual and data were obtained from self-administered surveys answered by individuals, it was necessary to aggregate individual responses to the team level. To justify aggregation, we computed the average deviation index (AD M Index) developed by Burke, Finkelstein, and Dusig (1999) . Following the authors’ recommendations, and since all the scales have 5 response options, we used the criterion AD M ? 0.83 to aggregate, with confidence, individual responses to the team level. The average AD M values obtained for transformational leadership, quality of group experience, and group autonomy were, respectively, .59, .57, and .66. These values were below the upper-limit criterion of .83, revealing that the level of within-team agreement was sufficient to aggregate team members’ scores.

Table 3 presents the intercorrelations between the variables studied, as well as deive statistics and reliability. As expected, the GTL presented high positive correlations ( Cohen, 1988) with group autonomy (r = .55, p < .001) as well as with quality of group experience ( r = .53 p < .001), and these results support the nomological validity of the scale. All the scales presented adequate values of internal consistency.

Table 3

Deive Statistics, Internal Consistency and Intercorrelations between Transformational Leadership, Group Autonomy, and Quality of Group Experience.

	M	SD	?	1	2	3
1. Transformational Leadership	3.88	0.59	.94	--
2. Group Autonomy	3.46	0.56	.89	.55 **	--
3. Quality of Group Experience	3.95	0.53	.87	.53 **	.41 **	--

**

p < .01