Dinh et al. (2014) state that the topic of leadership has generated diverse theories as an outcome of the increase in scholarly research over the last decade. The discussion now illustrates this diversity: Sanda and Arthur (2017) refer to the complexity of the leadership process by which a person influences followers; however, Meindl (1995) and Meindl and Ehrlich (1987) warn against romanticising leadership, as if leadership were a phenomenon, whereby leaders deliver it to followers. Oc and Bashshur (2013) explain that leadership is not something delivered independent of recipients, leadership is co-created. Yukl (2012), in fact, defines leadership in a way that could be performed by multiple individuals. Because there is a difference between people in positions of leadership and leadership itself, leadership is either a matter of location or a matter of interaction and influence. The idea that companies are shaped solely by their leaders is challenged by the followership literature (Schneider, Gardner, Hinojosa, & Marin, 2014).

The current study perceives leaders as individuals who have a differential influence on the initiation, direction and coordination of a group and, followers as those who freely accept the influence of leaders, in line with the definitions of Bastardoz and Van Vugt (2019).

Research into followership, specifically follower behaviour, is gaining momentum (Oc & Bashshur, 2013; Uhl-Bien et al., 2014), and followers are recognised as not being monolithic (Bastardoz & Van Vugt, 2019). Many authors have defined the role of followers as passive in its orientation to the leadership process (Carsten et al., 2010; Oc & Bashshur, 2013; Sronce & Arendt, 2009). Followers are acknowledged as necessary, but not as necessary as leaders, to accomplish goals (Oc & Bashshur, 2013). Ford and Harding (2018) believe that followers are helpless conformists who are only driven once persuaded by their leader.

This unidirectional focus on the influence of leaders on followers perceives followers exclusively as passive recipients of the leadership influence (Oc & Bashshur, 2013).

Earlier definitions of leadership tended to define it in terms of maintaining group cohesion and taking initiative, whereas followership requires decisions about which leaders to follow, as well as when and where to follow them (Vugt, 2006). Fairhurst and Uhl-Bien (2012), however, refer to the interaction between a leader and followers as a ‘relational view’ (p. 1). They posit that leadership is created through social interactions and the relationship between a leader and followers. This relational view, or relational dynamics, is further defined by Uhl-Bien et al. (2014) as a mutual influence process in which followers engage with leaders. Earlier theories, such as the leader–member exchange (LMX) theory, emphasise the engagement model between leaders and followers, and how these relationships culminate in effective leadership outcomes (Graen, 1995; Liden & Graen, 1980). Leader–member exchange theory goes as far as to refer to the relationship as dyadic, or two-way, in which one influences the other.

Theory development across the past decade has, therefore, culminated in an understanding that the leader–member exchange is relational, as opposed to one-sided, and that this relationship is key to driving the outcomes or consequences of the work unit (Oc & Bashshur, 2013). Oc (2018), in more recent research, posits that follower traits are important predictors of leadership creation. Uhl-Bien et al. (2014) suggest that in the process of leadership, without followers and follower behaviour, there is no leadership. We can, therefore, infer that the study of followers is key to just understanding how the leadership process is affected by follower orientation. In this regard, we adhere to the call by Shamir (2007) to reverse the lens in leadership research by addressing the role of followers in the effectiveness of leadership.

Blair and Bligh (2018) emphasise that Africa is one of the regions that are underrepresented in followership theory development. The current study is conducted in South Africa that adds to its contextual contribution within followership literature. Matshoba-Ramuedzisi (2021) also highlighted that because South Africa is grouped under the Sub-Saharan Africa cluster in the GLOBE Study (House et al., 2004), it is important to review followership studies in this region. Matshoba-Ramuedzisi (2021) reports that the studies are limited; however, an interesting study in Uganda (Ofumbi, 2017) revealed that the Acholi people enact their followership through a process of observation and the response can either solicit no actions or actions to uphold or change the status quo. The responses of followers include obedience and deference and participation alongside leaders or intervention. The response of followers is important to investigate especially when leaders are ineffective and when they are vindictive or malevolent towards followers. Singh and Bodhanua (2013) conducted a South African study and found that followers are powerful actors who in turn influence other individuals, the organisation itself and the environment in which the organisation is embedded, and studies on followership should, therefore, consider organisational and environmental contexts that contribute how followers think, act and how they perceive their roles, themselves and their duties. This South African study also found that ‘there is still a tendency to glorify leaders at the expense of recognizing and understanding the people who choose to follow such leaders’ (Singh & Bodhanua, 2013, p. 513). Other Sub-Saharan Africa studies include the studies by Thomas (2014) who applied the Kelley 20-question followership style survey and found that the Rwandan sample had a significant smaller number of exemplary followers than an American sample. Du Plessis and Boshoff (2018) found that followership is a significant influence on employee work engagement.

The study by Carsten et al. (2017) included respondents from a Chinese company across 366 dyads samples. They state that in Chinese culture, leaders understand that there are harmonious relationships between themselves and their followers, and that authority is granted by the followers through deference (McElhatton & Jackson, 2012). The authors claim that these results can be generalised across the world. However, differences in power distance between China and South Africa might challenge this claim. Power distance refers to the expectation that there is an unequal distribution of power (Hofstede, 2011) between leaders and followers. The GLOBE Study (‘House et al., 2004’) found that the power distance for Chinese culture was considered very high (Chinese sample = 3.1). In South Africa, however, there were varying power distances for each ethnic group (black people = 3.65, white people = 2.64), according to the GLOBE Study (House et al., 2004). This suggests that the results across different countries may be dissimilar as LMX relationships have been revealed to be affected by differences in power distance (Anand, Vidyarthi, & Rolnicki, 2017) across different ethnic groups. Power distance, for example, affects the communication between leaders and followers (Anand et al., 2017), which, in turn, can lead to differences in the strength of the correlations than in the Carsten et al. (2017) China study. The present study, therefore, aims to contribute to theory development in followership literature, by offering results in a different cultural context and offering data on followership from a follower perspective.

The philosophy behind the methodology is positivism, whilst the methodological approach is deductive (Saunders, Lewis, & Thornhill, 2009). The population and unit of analysis are individual employees in corporate businesses in South Africa. The perception of these employees about their own followership was the unit of analysis. The target population sample, therefore, was followers, which means employees who had managers, as the present study investigated role orientations within a hierarchical context. They had to be employees who had work experience at corporate businesses to be able to have the experience of being a follower while reporting to a manager. We were, therefore, interested in a general population of employees who would have work experience and who were reporting to a manager. The researchers worked in corporate businesses and utilised non-probability convenience sampling and targeted their colleagues in their own networks, who had work experience and who were reporting to managers. The focus of the current study was on how the respondent perceived followership, and the questions, therefore, revolved around how the respondents interacted with their manager. The level of management was not part of the research questions. The perceptions of the respondent were important, regardless of the level of management the respondent was at or the level of management of the manager.

The researchers were interested in perceptions around followership throughout the organisation. Except for the Chief Executive Officer (CEO) of a company, the rest of the employees would perceive themselves as having a manager whom they report to and the current study was targeting this broad definition of employees. It is, therefore, difficult to establish the total population from which the sample was drawn. As Johnston, McLaughlin, Rouhani and Bartels (2017) advised, the total number of the employees who had experience of working in organisations and were reporting to managers (called N) was considered to be unknown in this study.

The survey questionnaire included biographical questions to ensure that the respondents met the criteria of the target population. In Table 1, the frequencies and percentages of the sample are displayed for each biographical question. The biographical questions include gender, age, how long the respondents were working for their current employer, number of employees at the company to ascertain the size of the organisation, the duration of working with their current manager and the hours per week interacting with their current manager, the industry the respondent was working in and which race group the respondent belonged to. It was important to note the number of employees, because small organisations might not have a hierarchical structure where there are employees responding to managers. Because the current study considers followership within this hierarchy of manager and follower, the researchers deem it important to ascertain the size of the organisation.

The researchers monitored whether the respondents had in fact answered the questions on their interaction with their manager and the time of reporting to their current manager. All respondents completed these questions, and therefore they were not excluded from the analysis based on these criteria. The researchers requested their colleagues to forward the survey link to their colleagues who met the criteria, namely having work experience and reporting to a manager in an organisation. The sampling method includes purposive, snowball sampling because of the fact that the respondents were asked to forward the link of the questionnaire to those in their networks who had work experience and reported to managers in their organisations. In summary, we focused on the micro-level analysis of an individual employee on followership.

The survey strategy ensured participant confidentiality as personal information that could be used to identify respondents was not requested (Saunders et al., 2009). An Internet-mediated questionnaire tool (Saunders et al., 2009), Survey Monkey (‘Survey Monkey’, 2018), was used to create, publish and gather responses for the survey. This allowed for ease of access by respondents and real-time updates on the progress of sampling. This was not an open survey, and only those employees in the network of the researchers who met the criteria of the sample population (being followers reporting to managers and having work experience in corporate businesses) received the survey link. Incompleteness was handled during data analysis by the removal of incomplete responses (Hair, Black, Babin, & Anderson, 2010).

A pilot test was administered to 17 respondents to understand the suitability and representativeness of the questions, test for face and content validity and to ensure that scale items were clear and understandable (Saunders et al., 2009; Zikmund, Babin, Carr, & Griffin, 2013). This number was deemed suitable because of the expected power effect in the range of 0.2–0.5 (Cohen, 1992). Some responses to the pilot study revealed errors that included incorrect scale item labelling in certain questions and members of the pilot study reported language errors. These were corrected prior to the distribution of the final survey.

The researchers did not send out the survey link to these pilot respondents and therefore assumed that they did not form part of the final survey respondents. However, the researchers acknowledge that because of the fact that the respondents were asked to forward the survey link to their colleagues, it could have caused the initial pilot respondents to complete the questionnaire again and which in turn could have caused response bias. This is, therefore, a limitation in the survey design of the present study. Because of the fact that personal information, like names and surnames and companies’ names, was not requested, the researchers could not control for this potential limitation.

The survey consisted of questions that took approximately 15 min to complete. Respondents were told the purpose of the study and then informed of their voluntary participation and the ability to exit the survey at any time. Nineteen questions relating to the constructs were asked. The final section of the survey requested demographic information as described above. No information concerning the date of birth, identity number, contact details or other personal information was requested. A total of 310 responses were received (N = 310).

After the response period lapsed, the data were exported into an Excel file. The data were assessed using IBM Statistical Package for the Social Sciences (SPSS) Statistics Version 24 (IBM SPSS, 2016). Whilst the original study by Carsten et al. (2017) was conducted in one industry, the current study spanned various industries.

The research instrument is a self-completed, structured survey questionnaire (Zikmund et al., 2013). The questionnaire used a Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree), with questions relating to co-production orientation, passive role orientation, voice behaviour and upward delegation.

Co-production: This construct was measured by five items from the scale of Carsten and Uhl-Bien (2012), with Cronbach’s alpha, 0.90. Sample items included, Followers should be on the lookout for suggestions they can offer to superiors, and followers should communicate their opinions, even when they know leaders may disagree.

Passive role-orientation: This construct was measured using four items from the scale of Carsten et al. (2017). Sample items included:

Because one is a follower, he and/or she does not have to worry about being involved in decision-making; and being a follower means that you don’t have to think about changing the way work gets done.

The Cronbach’s alpha of the scale was 0.93.

Voice behaviour: The construct was assessed using six items from the scale of Van Dyne and LePine (1998). Sample items included:

I communicate my opinions about work issues to my manager even if my opinion is different and others disagree with me; and I develop and make recommendations to my manager concerning issues that affect my work.

The Cronbach’s alpha of the scale was 0.89.

Upward delegation: Four items were used to measure this variable adapted from the scale of Carsten et al. (2017). Sample items for the upward delegation included:

How often do you pass problems along to your manager rather than taking care of them yourself?; and how often do you pass responsibility for problems to your manager?

The Cronbach’s alpha of the scale was 0.92.

The questionnaire was similar to the one used in the study by Carsten et al. (2017), except that those concerning the impact of followership on leadership were not included. This is because the present study focuses only on the antecedents of followership, particularly how followers’ own beliefs or orientations influence their follower behaviour, within the cultural context of South Africa.

To understand the sample, initial information was presented as descriptive statistics. The demographic characteristics of the dataset were reported as frequencies. The results of the questionnaire were tested for consistency, validity, reliability and inter-item correlation using Cronbach’s alpha test (Tavakol & Dennick, 2011) (See Table 2 with Cronbach’s alpha values). Confirmatory factor analysis (CFA) was used to confirm convergent validity. Comparative fit index (CFI) was also used to measure the variables and their relation to the construct. To complete an exploratory factor analysis (EFA) on the constructs, the Kaiser–Meyer–Olkin (KMO) coefficient was used to test for sampling adequacy. The Bartlett test of sphericity was used to test the suitability of data reduction.

To test for correlation, the Karl Pearson’ coefficient of correlation test was used (Winter, Gosling, & Potter, 2016). This correlation test makes use of Cohen’s effect size, whereby an effect size of 0–0.2 is termed small, 0.2–0.5 is termed medium and greater than 0.5 is termed large (Cohen, 1992). Regression analyses were undertaken to offer a more robust analyses of the relationships and an Analysis of Variance (ANOVA) to analyse strengths of relationships. Control variables were included in the correlation test. Sample characteristics and descriptive statistics are provided below, and Table 1 offers a summary of the frequencies and percentages of specific sample characteristics.

Of the 310 responses that were received (N = 310), 11 respondent questionnaires were rejected from the final analysis as 50% or fewer questions were answered. Twelve respondents’ questionnaires were removed as they did not answer the racial demographic question. The reason for this decision was that the researchers assumed that in the South African context, race is an important diversity dimension; however, in the discussion of limitations of the current study, this decision is evaluated. Any responses with between 50% and 99% completion rate were imputed by taking the average of that question and using that for the missing value (per question). So, if a question had an average of 5.2, any missing value was replaced with 5.2. This was done per question for missing responses. Analysis was conducted on the remaining 287 samples (N = 287). In terms of race, the white population represented 24.7% of the sample, whilst the black population represented 13.9% and the Indian population represented 52.6% of the sample. The mixed race population was 5.2% and Asian, 2.1% and Others 1.4%. More than 65% of the sample worked in organisations with 500 or more employees. More than 50% of respondents were between the ages of 31 and 40 years. Males made up 59.04% of the sample, whilst females made up 40.96% of the sample. Working for one’s current manager for between 0 and 2 years represented 51.92% of the sample, whilst between 3 and 5 years represented 30.7% of the sample. The remainder worked for their manager for 6–10 years and represented 11.5% or more of the sample (11–15 years: 2.4% and more than 15 years: 3.5%). This question was not a reflection of work experience; instead, it referred to the current manager only. Because most respondents were more than 31 years of age, they would have had adequate work experience to answer questions about their experience of reporting to managers and therefore their perception of followership. With regard to the duration of working for their current employer, 23.7% of respondents indicated that they have been working for 0–2 years for their current employer, whereas 26.5% of respondents indicated that they have been working for 3–5 years and 24% for 6–10 years and 17.8% for more than 15 years. These findings show that the respondents had adequate work experience to report on their perceptions around followership.

The demographic questions included a list of 12 industries where respondents had to indicate in which industry they were working. These industries included mining, telecommunications, manufacturing, financial services, forestry, etc. There was also an ‘others’ category where respondents could indicate their industries, and therefore there were 22 industries in total. This ‘other’ category is a limitation in the current study as it made it difficult to ascertain and code the category of industry the respondent was working in. Nonetheless, the current study wanted to source respondents from various industries, and it was not essential to identify the specific industry that the respondents were working in for the purposes of the study on followership. The researchers distributed the link to their contacts within the telecommunications and financial industries; however, because of the fact that the researchers requested these contacts to distribute the link to their colleagues (using snowball sampling), it made it difficult to ascertain exactly in which industries these respondents were working. Under the section ‘limitations and future research’, these limitations are discussed.

The University Pretoria Gordon Institute of Business Science Ethics Committee granted the ethical clearance, after reviewing our questionnaire and all ethical considerations. See attached ethical clearance letter from our committee. No Ref No. 20 June 2018.

Response bias may be prevalent when data are collected from hierarchical structures as questions are typically answered to protect relationships (Furnham, 1986; Zikmund et al., 2013). Pilot studies also pose a concern as respondents may respond differently to a survey if they have previously experienced it, resulting from social desirability bias (Furnham, 1986). The authors also acknowledge the possibility of respondent carelessness (Huang, Curran, Keeney, Poposki, & DeShon, 2012). Further limitations include that the study made use of a single source of respondents to assess both the independent/predictor and independent variables and relied on a cross-sectional study and because of the fact that the present design was not an experimental design, the regression analysis presented an inability to test for causality.

As mentioned under the sampling section, the current study’s design poses important limitations, and researchers would need to take caution in interpreting the results, and the limitations are decreasing the probability of generalising the results to the total population. The convenience non-probability sampling method is a limitation, and the snowball sampling of respondents sending the survey link to their colleagues could have caused the pilot respondents to be part of the final respondents, which the researchers could not have control over. The researchers also did not use the 12 respondents’ data from the initial 310 respondents, who did not complete the question about their race. Whilst the researchers considered race to be important in the context of South African’s history, race as control variable was not significant, and in retrospect, the exclusion of these 12 respondents’ data poses a limitation of the current study. In future studies, the researchers would recommend not to exclude respondents’ questionnaires from a total sample when they do not complete the question about race.

Because the current study targeted respondents in a range of industries, the respondents were spread across industries. The inclusion of an industry category called ‘other’ is another limitation in the current study, which resulted in uncertainty around the category of the industry that a respondent represented. Future research could use a standardised list of industries and rather not offer the ‘other’ category as a choice to respondents in the demographic questions.

Constructs, other than voice behaviour and upward delegation, such as taking initiative, attitude, constructive and dysfunctional resistance and obedience, could be considered in future studies. Furthermore, other mediating factors such as organisational context and environment might be included in research designs going forward. Another suggestion for further research is to use demographic information to present a different perspective on the constructs, based on followers’ tenure, size of organisation and possibly the duration of interaction with a manager. Whilst the current study did not find significant differences in how different race groups in South Africa perceived followership role orientation and subsequent follower behaviours, future studies could explore with longitudinal qualitative studies how followers socially construct their perceptions and ultimately contribute to the leadership process.

The current study points to the limitations of role theory in explaining followership and therefore contributes that the lens used in future studies might require a more flexible approach. The theory used as a departure point could influence the hypotheses and the instruments used in assessing these and therefore, future studies could use social constructionism (Uhl-Bien et al., 2014), social exchange theory (SET, Blau, 1964) and/or social-cognitive theory of Bandura (1977) to emphasise the agency of followers in co-producing leadership.

Deconstructing the influence that followers contribute to the leadership process offers a great deal of potential for leadership and management studies. For example, this knowledge has benefits for organisational change and development. The results of the study assist managers and subordinates to better understand the traits of followers (co-produce outcomes or remain passive in their role orientation) and their links to employee voice behaviour or upward delegation. Armed with this understanding, leaders could adopt different leadership styles, based on the follower’s orientation, to better manage and delegate work. For example, in cases where a follower is prone to deflect responsibility by upward delegation, the leader could consider the detrimental impact on productivity over the longer term and therefore adapt his or her approach. The leader could give the follower feedback on the consequences of his and/or her behaviour for the effective running of the business. Without being prescriptive, a leader could, for instance, profile a specific follower, based on his or her orientation, and then use this understanding to communicate more effectively with this subordinate.

The findings also have implications for understanding the organisational fit of employees, based on their orientation and behaviour. They could assist managers, change practitioners, human resource teams and even executives to identify which follower behaviours determine and affect their orientation and contribution towards organisational outcomes. The fit between leaders and followers is also influenced by the follower’s orientation and is, therefore, an important variable to consider when matching leaders and followers.

During recruitment, understanding these behaviours can influence the selection of candidates, especially if they display passive role orientation behaviours. The questionnaire used in this study could be used in recruitment as pre-screening exercises to assess orientation so that more informed decisions can be made about suitable candidates. The research instrument used in this study could even be administered to existing employees as part of regular follower assessments. This would give organisations a view of the prevalent behaviour orientations and, where necessary, encourage them to adapt to, or transform, the way in which followers orient themselves. By identifying these behaviours, organisations could even reward or recognise employees who exhibit voice behaviour and refrain from upward delegation.

There is also value for followers themselves as the findings give them a better understanding of their sometimes-subconscious behaviours. Understanding their own orientations may encourage them to adopt or do away with certain behaviours to be more effective within their organisations. As the research was conducted with a sample featuring diverse demographics from across South Africa, without any industry specificity, it is applicable to most leader–follower or manager–subordinate relationships within organisations across the country. Organisations should also pay attention to reorienting leaders towards constructive leadership that is not resistant to followers voicing their opinions.

Whilst the sampling method has limitations, the research presented in this article highlights the importance of follower behaviours and their effects on the bi-directional leadership process. This study extends the work of Carsten et al. (2017) beyond China to South Africa. Within this population, it was revealed that voice behaviour and upward delegation are associated with followers with co-production and passive role orientations, respectively. An understanding of these behaviours and orientations can impact the way change practitioners, leaders and followers interact with one another. Constructive followership, characterised by the co-creation of solutions that actively contribute towards effective leadership processes, should be encouraged.