4 Research Methodology
4.3 Core action research projects measures
4.3.3 Data evaluation
This section describes the different steps undertaken to evaluate the gathered data. This includes a description of how the TMLQ data was gathered and evaluated, as well as a description of the tools used for the SNA.
The TMLQ data was evaluated using the evaluation sheet provided by Mindgarden (2015). For the purpose of this research study, an official TMLQ survey was purchased (including the evaluation sheet). The TMLQ survey was purchased from Mindgarden (Mindgarden, 2015). A Microsoft Excel sheet was developed, and the data was manually entered for the paper surveys, as illustrated in the sample evaluation in Table 7. For each TMLQ attribute, for the team the mean of the respective answers was computed. The answers that needed to be considered for each attribute are listed in parentheses (e.g. ‘Idealised Attributes’ is computed by taking the mean of answers 2, 12, 22, 32, and 42).
Table 7: TMQL Data Evaluation in Microsoft Excel (Mindgarden, 2015) TMLQ attribute E lle n K ar l Spe nc er S imo n Ul i El en a Ma rt in a Te a m Idealised Attributes (2, 12, 22, 32, 42) III 3.20 3.60 2.40 2.20 3.00 2.60 3.20 2.89 Idealised Behaviours (4, 14, 24, 34, 44) IIB 2.60 3.20 1.80 1.20 2.00 1.80 1.50 2.01 Inspirational Motivation (6, 16, 26, 36, 46) INSP 2.20 3.20 1.80 1.00 2.20 2.20 2.20 2.11 Intellectual Stimulation (8, 18, 28, 38, 47) IS 2.40 3.60 2.40 1.80 3.20 2.40 2.80 2.66 Individualised Consideration (10, 20, 30, 40, 48) IC 3.40 3.40 2.60 2.40 2.60 3.00 3.20 2.94 Contingent Reward (7, 15, 25, 35, 45) CR 3.20 2.40 2.00 1.40 2.40 2.60 1.80 2.26 Management-by-Exception (Active) (5, 13, 23, 33, 43) MBEA 2.80 2.40 1.80 2.00 1.80 2.40 2.40 2.23 Management-by-Exception (Passive) (3, 11, 21, 31, 41) MBEP 1.40 0.00 0.60 1.00 0.60 1.20 1.50 0.90 Laissez-faire (1, 9, 19, 29, 39) LF 1.40 0.40 0.00 1.40 0.60 0.60 1.00 0.77 Extra Effort (17, 27, 37) EE 2.67 3.00 2.67 1.33 2.00 1.67 1.67 2.14 Effectiveness (61) EFF 3.00 2.00 2.00 2.00 3.00 3.00 3.00 2.57 Satisfaction (62) SAT 3.00 2.00 3.00 2.00 1.00 3.00 3.00 2.43
The TMLQ proposed by (Bass, 1985) (henceforward, ‘the Bass TMLQ) used a standard Likert scale 0 = ‘Not at all’; 1 = ‘Once in a while’; 2 = ‘Sometimes’; 3 = ‘Fairly often’; 4 = ‘Frequently or always’. In addition to the mean value for each TMLQ attribute, box plots for each attribute were created (Figure 9). These box plots depict the mean value (triangular symbol) and the median (horizontal line); the bottom and top of the box represent the first and third quartiles of the data distribution; the whiskers depict the full data range (i.e. max and min value). In cases where the team members have similar evaluations of one another, the full data range is narrow; otherwise (i.e. in cases of outliers) the data range spread is correspondingly large. Both cases involve additional information that would be lost if only the mean were considered.
Figure 9: Box Plot for TMLQ Data
Calculation of decentralisation
For calculating the decentralisation, the UCINET tool (Borgatti, 2002) was used. For each of the completed surveys on shared leadership competencies rated by each member, the lowest value could be 0 and the highest value could be 4 (see Table 8). The answers/values given to the different areas were averaged as suggested by Pastor and Mayo (2002).
Table 8: Sample Matrix for Shared Vision (following Pastor and Mayo (2002)) Matrix of Shared Vision
Ellen Karl Spencer Simon Uli Elena Martina Total
Ellen 0 3 3 3 3 3 2 17.0 Karl 3.5 0 4 2.5 4 3 2.5 19.5 Spencer 1.5 2.5 0 1 2.5 1.5 1.5 10.5 Simon 2 2 2 0 2 1.5 1.5 11.0 Uli 1 2 2.5 2.5 0 1.5 1 10.5 Elena 2.5 3 3 2.5 3 0 1 15.0 Martina 2 2.5 2.5 2 2.5 2 0 13.5
In the example of the shared vision attribute in Table 8, each cell represents the vision attributed to the other members of the team. The total in the rows can be used as a measure of the shared leadership’s coordination influence attributed to each member by his/her peers. A higher value represents a higher influence on other team members.
Sociograph of shared leadership
As suggested by Pastor and Mayo (2002), the represented shared leadership values of the vision attribute data were dichotomised. Practically, values less than 2 were considered as 0, and values greater than and equal to 2 were assigned the value 1. By doing so, the value-based network data was transformed into a binary network where only the presence was counted and not the strength of the relationship. The dichotomised network for the example in Table 8 is presented in Table 9.
Table 9: Dichotomised Matrix for Shared Vision (following Pastor and Mayo (2002)) Matrix of Shared Vision
Ellen Karl Spencer Simon Uli Elena Martina Total
Ellen 0 1 1 1 1 1 1 6 Karl 1 0 1 1 1 1 1 6 Spencer 0 1 0 0 1 0 0 2 Simon 1 1 1 0 1 0 0 4 Uli 0 1 1 1 0 0 0 3 Elena 1 1 1 1 1 0 0 5 Martina 1 1 1 1 1 1 0 6
To calculate the decentralisation for the shared leadership vision attribute based on the dichotomised example in Table 9, the matrix was entered and saved into the Matrix Editor of the UCINET tool as shown in Figure 10. In the UCINET tool, using the menu selections ‘Network’ -> ‘Centrality and Power’ -> ‘Degree’, the file that was previously saved for the shared coordination attributes was opened to calculate the decentralisation measures. The output for the centralisation measure is presented in Figure 11.
Figure 10: Matrix Editor in UCINET (Borgatti, 2002)
Figure 11: Output File for Decentralisation in UCINET (Borgatti, 2002)
The calculation printout gives an overview of Freeman’s centrality measures for the evaluated network matrix. For example, the number of nodes (seven in this example) and the standard deviation are provided. Further, the indegree of network centralisation is given. In an
undirected network, indegree refers to how prominent a node is. For the example above, the network centralisation is 27.779, which means the decentralisation is 0.72.
Visualising social network diagrams
For visualising the social network diagrams, the tool NODEXL (CodePLex) was used to create the social network diagrams. The previously dichotomised matrixes were used (see Table 9) as input data. The same example is used to show how the graphs and the respective statistics were gathered with the NODEXL tool. One advantage of the NODEXL tool is that it automatically calculates the network density and the number of unique edges for the social network.
Figure 12: Microsoft Excel Template for NODEXL (CodePLex)
Once the data is entered in the NODEXL tool, a social network diagram can be easily created, and the power relation of the team can be visualised (Figure 12). The entire data gathered during the core action research projects was collected using paper-based surveys. Subsequently, this data was manually processed as described in these steps in order to evaluate the data.
5
Development of Research Instrument on Shared Leadership
As described in the previous chapter, combining social network analysis with a shared leadership model requires a stringent shared leadership research instrument. To address this need, a shared leadership research instrument was developed in this study .For the development and validation of the shared leadership instrument, a step-wise approach was followed as proposed by Gehlbach and Brinkworth (2011). In step one, the shared leadership competencies that a person needs to be productive in a self-organising team were identified; (these competencies were described in section 3.4). In step two, interviews were conducted with expert focus groups (see section 5.2) to assess the experts’ experience in these competencies. In the third step, the findings of the literature review were synthesised with the expert interview data. In step four, the scale items were developed (see section 5.3). In step five, an expert face validation was performed (see section 5.4). Finally, in step six, a pilot and the evaluation of the validity and reliability of the developed instrument were undertaken (see section 5.5).
The purpose of validating the shared leadership instrument is to ensure that the developed research instrument measures what it is supposed to measure by using mathematical and statistical methods. Following these six steps for the development of the shared leadership research instrument, would ensure that the instrument is reliable, i.e. it is accurately measured and validated via statistical methods so that the instrument scale items measure the shared leadership competencies as expected.
Hinkin and Schriesheim (1989) and Rattray and Jones (2007) proposed a new, conceptually consistent theoretical definition of the constructs of a scale development as discussed in section 3.4. To underpin the shared leadership characteristics identified during the literature review with a more practical view on shared leadership competencies, face validity was performed (Easterby-Smith et al., 2008a). Face validity refers to expert opinions regarding whether or not the developed scale items represent the relevant domains and the concept of the survey (Ferguson and Cox, 1993; Rattray and Jones, 2007). Face validity is an initial step to validate the theoretically identified shared leadership competencies using expert opinions.