This study employed an exploratory, descriptive, sequential mixed-method design, which had three phases. Phase 1 consisted of two subphases: subphase 1 entailed conducting an integrative literature review to explore and describe the concept of self-leadership of nurse educators and subphase 2 entailed conducting semi-structured focus group interviews with nurse educators from a selected group. Phase 3 entailed the development and validation of guidelines to facilitate self-leadership in nurse educators. This article reports on phase 2, which focused on obtaining information about the self-leadership practices of nurse educators through structured questionnaires.

The population was nurse educators teaching in purposively selected private nursing schools, public nursing colleges, and universities in the Gauteng and KwaZulu-Natal provinces in South Africa. The nursing education institutions in the two provinces consistently had high outputs of registered nurses and midwives for the years 2009-2018.

Convenience sampling was used to recruit nurse educators appointed on a full-time basis at the remaining nursing education institutions in the two provinces, who were not included in the qualitative phase of the study. Participation was voluntary and anonymous, and no incentives were offered. The total number of possible respondents was 443, of whom 265 (59.8%) responded.

A questionnaire was compiled based on the themes that emerged from phase 1 (subphase 1 and subphase 2) and the literature to determine the self-leadership practices of nurse educators in nursing education institutions. This data collection approach is typically that of a sequential mixed-methods study because one set of data would be building on the other [30]. The questionnaire comprised four sections: Section A was designed to elicit the biographical information of nurse educators who participated in the study; Section B comprised questions that attempted to ascertain the nurse educators’ understanding of the concept of self-leadership and its constructs; Section C comprised questions that attempted to determine the self-leadership practices of nurse educators; Section D comprised questions that attempted to ascertain how motivation could influence self-leadership. Some of the questions were further explored by asking open-ended questions to obtain richer information. The questionnaire used a seven-point Likert scale with the responses ranging from strongly disagree (1) to strongly agree (7). This article focuses on findings based on data collected in Sections A and C of the questionnaire. The hand-delivered structured questionnaires were distributed and collected by the researcher between 2018-2019. Sealed and lockable boxes were used to collect the questionnaires from the respondents. Each questionnaire was accompanied by an information leaflet that described the purpose of the study and the ethical issues related to the participation in the study.

Rigour of the study was ensured through validity and reliability. In order to ensure the validity of the instrument, content validity, face validity, and construct validity were measured. Content validity was established through an integrative literature review, discussions, and consultations with the study supervisor to evaluate the relevance of the questions to the subject. Pre-testing of the questionnaire and coding of the question items by a statistician also improved the content validity of the instrument. Face validity was established using the statistician and nurse educators who were not part of the study to review the questionnaire and identify questions that were difficult to interpret and answer.

Construct validity was established by incorporating the theoretical framework and integrative literature review on self-leadership into the questionnaire.

Reliability was enhanced by carefully designing, pre-testing, and checking the questionnaire. Bell and colleagues describe the Cronbach-alpha test as a commonly used test to determine internal reliability [31]. In this study, a Cronbach’s alpha coefficient of 0.6 was used to determine the reliability of the tool. The closer the Cronbach’s alpha is to 1, the higher the internal consistency reliability [31].

Table 1 below presents a summary of biographical data of the respondents, namely gender, age, and type of nursing education institution. Where not all respondents responded to items such as age, the authors calculated the frequencies and percentages according to the number of responses. Therefore, missing values were not included but were indicated in the tables.

Only 15 (6%) respondents were males. Males are a minority in the nursing profession, which is generally dominated by females according to Smith and colleagues [35]. Thus it was not unexpected that females responded to the questionnaire more frequently than males.

The age bracket with the lowest number of respondents was that of 21–30 years at nine (3.4%) and the age bracket with the largest number of respondents was that of ages 51 or older, which comprised 140 (53.2%) respondents, which means that more than half of the respondents were older than 51-years. Some authors have observed that nurse academics are becoming redundant, which may impact the country’s ability to educate and produce next generation nurses [36, 37].

As depicted in the table, 178 (67%) of the respondents were teaching at public nursing colleges, 60 (23%) taught at private nursing schools, and 27 (10%) taught at universities.

Exploratory factor analysis was applied to the responses of the self-leadership activities subscale. The maximum likelihood method was used to extract the factors, and this was followed by a varimax (orthogonal) rotation. The following criteria were used to determine the number of factors to be used for rotation: cumulative percentage variance greater than 50%; Eigen value greater than 1; and a significant decline in the scree plot. In interpreting the rotated factor pattern, an item was said to load on a given factor if the factor loading was 0.40 or greater for that factor and less than 0.40 (absolute value) for the other factors in the rotated factor matrix, as depicted in Table 2. The factor loadings that were 0.40 or greater are highlighted in bold.

Five factors exhibited with Eigen values above 1 and 50.9% of cumulative variance. The results of a scree plot also suggested that the five factors were meaningful. Thus, the five factors were retained for rotation. Table 2 below displays the matrix of rotated factor loadings for the subscale: self-leadership activities.

Six items loaded on the first factor labelled “autonomy-supportive environment”; six items loaded on the second factor labelled “continuing professional development”; six items loaded on the third factor labelled “role modelling”; three items loaded on the fourth factor labelled “shared leadership”; and one item loaded on the fifth factor, possibly labelled “mentoring”.

Item C68 that was cross-loaded for factor 1 and factor 5 was not retained for any of the two factors. Although it loaded the highest on factor 1, it did not make logical sense, nor would it make logical sense if retained in factor 5.

Similarly, item C54 which was cross-loaded for factor 1 and factor 5 at 0.45 and 0.40, respectively, was not retained for either of the factors. The fifth factor in this subscale had two items, which were cross-loaded with other factors in the rotated factor matrix and these items were not retained. As a result, there was only one item, C63, which was about the institution providing new nurse educators with mentors, remaining under factor 5. Since factor 5 only had one item, a score could not be calculated.

Table 3 below presents the results from the analysis of the questionnaire subscale of self-leadership activities. As depicted in the table, the reliability testing was measured on the items that loaded as follows: six items loaded on the first factor labelled “autonomy-supportive environment”; six items loaded on the second factor labelled “continuing professional development”; six items loaded on the third factor labelled “role modelling”; and three items loaded on the fourth factor labelled “shared leadership”. The Cronbach alpha coefficient of all the factors was greater than 0.6, which means that the items of the constructs were reliable. Thus, the results indicate the internal consistency of the self-leadership activities.

The responses on the questionnaire ranged from a 1, strongly disagree, to a 7, strongly agree. The composite construct scores for factors in the subscales were calculated by taking the average of the reliable items that loaded onto that factor. Histograms were used for showing distribution and descriptive statistics. Distribution was considered skew if the skewness value was outside the range of -1 and +1. Where the distribution was skew (skewness outside the range of -1 and +1), the median was used in the interpretation of results.

The composite scores of each of the constructs for subscale self-leadership activities are discussed below:

4.4.1. Construct 1: Autonomy-Supportive Environment (Items C67, 69, 70, 71, 72, 75)

The composite construct “autonomy-supportive environment” measured whether nurse educators perceived that the nursing education institution and its management provided an environment that facilitated self-leadership by providing autonomy-related aspects, such as giving nurse educators room for failure and encouraging them to take risks; involving nurse educators in decision-making processes; supporting them in innovative and creative behaviours and empowering nurse educators by sending them to training programmes that stimulate their self-leadership.

Fig. (1) below depicts the distribution of composite scores on the construct “autonomy-supportive environment”. The composite scores for each respondent ranged from 3.3 to 7 (maximum score). The histogram shows a skewed distribution of composite scores to the left (negatively skewed), with a skewness value of -2.28 and a mean of 6.54. Therefore, a median of 6.8 was used for interpretation. The median (6.8) is greater than the mean of 6.54, with a standard deviation of 0.62. This means that the respondents perceived that an autonomy-supportive environment could facilitate self-leadership by strongly agreeing with this construct.

Table 3. Reliability testing on the subscale self-leadership activities.

Subscale	Constructs	Items	Items left out	Cronbach	Reliability
Self-leadership practices	Autonomy-supportive environment	C67, 69, 70, 71, 72, 75	None	0.82	Good reliability
	Continuing professional development	C46, 47, 48, 50, 51, 60	None	0.78	Acceptable reliability
	Role modelling	C56, 59, 57, 55, 58, 66	None	0.75	Acceptable reliability
	Shared leadership	C61, 65, 53	None	0.69	Acceptable reliability

Fig. (1). Distribution of composite scores for Construct 1: Autonomy-supportive environment (n=264).

4.4.3. Construct 3: Role Modelling (Items C56, 59, 57, 55, 58, and 66)

The composite construct “role modelling” measured the nurse educators’ perceptions of aspects perceived as promoting role modelling to students and colleagues. The aspects related were meeting deadlines on tasks, giving timeous feedback to the students, adhering to teaching schedules (timetables), commencing with work on time, developing lesson plans in their preparations for teaching, and promoting ethical attitudes towards colleagues, students, and society.

Fig. (2). Distribution of composite scores for Construct 2: Continuing professional development (n=264).

Fig. (3) below depicts the distribution of composite scores on the construct “role modelling”. The composite scores for each respondent ranged from 3.3 to 7 (maximum score). The histogram shows a skewed distribution of composite scores to the left (negatively skewed) with a skewness value of -2.79 and a mean of 6.7. Therefore, a median of 6.8 was used for interpretation. The median (6.8) is greater than the mean of 6.7, with a standard deviation of 0.452. This means that the respondents perceived that role modelling in the nursing education institution facilitated self-leadership by strongly agreeing with this construct.

4.4.4. Construct 4: Shared Leadership (Items C61, 65, 53)

The composite construct “shared leadership” measured the nurse educators’ perspectives as to whether shared leadership through collaboration, succession planning, and feedback on performance could facilitate self-leadership. Fig. (4) below depicts the distribution of composite scores on the construct “shared leadership”. The composite scores for each respondent ranged from 3 to 7 (maximum score). The histogram shows a skewed distribution of composite scores to the left (negatively skewed) with a skewness value of -2.34 and a mean of 6.56. Therefore, a median of 6.6 was used for interpretation. The median (6.6) is greater than the mean of 6.56, with a standard deviation of 0.619. This means the respondents perceived that self-leadership could be facilitated through shared leadership in the nursing education institutions by strongly agreeing with this construct.

Fig. (3). Distribution of composite scores for Construct 3: Role modelling (n=264).

Fig. (4). Distribution of composite scores for Construct 4: Shared leadership (n=264).

Table 4 below depicts a comparison of the minimum and maximum composite construct scores, their average mean, median, and standard deviation for the subscale self-leadership activities.

Table 4. Summary of construct composite scores for subscale self-leadership activities.

The comparison in Table 4 above indicates that the minimum construct score for all the constructs under the subscale self-leadership activities was for “continuing professional development” at 2.3, and the maximum construct score was 7 for all constructs. Both “autonomy-supportive environment” and “role modelling” had the highest median at 6.8, which means that of all the constructs in the subscale self-leadership activities, the respondents had a more positive perception of the constructs, “autonomy-supportive environment” and “role modelling”. The construct with the lowest median in the subscales was “continuing professional development” at 6.5, meaning that this construct was less important for the respondents when compared to the others in this subscale.

The questionnaire had open-ended items (items 67; 68; 69) in Section C on self-leadership practices (activities).

The questionnaire had 67, 68, and 69 as follow-up items with open-ended questions that explored certain self-leadership practices (activities) and related aspects. In item 67, the respondents were requested to justify why collaboration between nurse educators at a nursing education institution should be encouraged. Of the 206 that responded, 71 (34%) indicated that collaboration facilitated the sharing of ideas, experiences, and best practices, leading to a change in individuals’ teaching practices. A total of 40 (19%) respondents indicated that collaboration helped improve relations and understanding among nurse educators, thus ensuring that nurse educators worked towards a shared common goal.

In the literature, the management of the educational institution is expected to encourage quality and well-structured and resourced collaboration in teams to improve the teaching performance of the educators, students’ achievements, and ultimately the institution [38]. An academic institution could promote self-leadership by creating a supportive atmosphere conducive to collaborative learning, innovation, and creativity, which may ultimately lead to improved student performance [39]. However, it is the responsibility of the individual nurse educator to proactively seek out and take advantage of available collaborative resources in the institution [38].

With regard to item 68, respondents had to indicate examples of training programmes that stimulate their self-leadership. Of the 150 that responded, 121 (81%) respondents listed training programmes related to their continuing professional development needs for effective curriculum delivery, such as innovative teaching strategies, discipline-specific courses, updates on clinical practice, research, and e-learning. Thirty-two (21%) respondents listed training programmes related to self-awareness and emotional intelligence, such as time management, mentoring, project management, personal development, and self-leadership strategies.

In item 69, the respondents were required to indicate which supervisor leadership styles could enhance the nurse educators’ self-leadership. The respondents indicated their preferences of supervisor leadership styles as follows: democratic: 104 (52%); participative: 45 (23%); transformational: 44 (22%); situational: 18 (9%); and autocratic leadership: 16 (8%). Katewa and Heystek (2019) [40], as well as Kolzow (2014) [41] assert that leaders that enhance the self-leadership of their educators are those that use the democratic and participative leadership styles, since these leadership styles emphasize the inclusion of followers in consultation and decision-making, as well as teamwork, and setting a good leadership example fortheir educators by leading from the front. The leaders’ transformational leadership enhances idea generation, exploratory thinking, creativity, and the innovative ideas of educators [42, 43]. However, autocratic leadership was not the preferred leadership style in literature since this style does not promote that consultation and decision-making resides with the leader; thus, undermining educator motivation and performance [44, 45].