The antecedents of leader-member-exchange (LMX) relationships in African context: the in ﬂ uence of the supervisor ’ s feedback delivery-tactic

Purpose – The current paper aims to analyse the antecedents of leader – member exchange relationships (LMX) by speci ﬁ callyfocusing onthein ﬂ uenceof thesupervisor ’ s feedbackdeliverytactic. Design/methodology/approach – This study uses qualitative research methods with primary interviews as the main data source. Primary interviews with 40 managers from top supermarkets in Nigeria, SouthAfrica andtheUKwere undertaken. Findings – The authors found that both high-quality positive feedback and constructive criticisms produced the same feelings – more positive interpersonal relationships with their supervisors, higher levels of commitment to their organisations, higher job satisfaction and thus, high-quality LMX relationships. Where criticisms were delivered without greater interpersonal treatment, feedback was perceived as negative, and participants revealed lack of job satisfaction, lack of commitment to their organisations, poor interpersonal relationship with their supervisors, high turnover intentandthus low-qualityLMXrelationship. Originality/value – To the best of the authors ’ knowledge, the current paper is one of the ﬁ rst studies to highlight the consequences of different feedback delivery tactics on subsequent LMX quality particularly in African context. The authors speci ﬁ cally develop a process-based model of enhancing high-quality LMX, which shows the role of the supervisor ’ s feedback delivery tactic in the process. The authors also develop a process-based model thatillustrateshownegative/unconstructivefeedbackcould result in a low-quality LMX. Finally, to the best of the authors ’ knowledge, this paper is also one of the ﬁ rst to offer a comparative assessment between African and British (the UK) empirical settings and highlight some interesting dynamics concerningLMXqualityandrole of supervisor ’ s feedback deliverytactic.


Introduction
As an emerging market with a wealth of opportunities (George et al., 2016, p. 377), Africa needs a large body of management research that places the continent in the mainstream management literature (Beugré, 2020;George et al., 2016;Jackson, 2015;Nkomo, 2017;Zoogah et al., 2015aZoogah et al., , 2015b. Consequently, there is a burgeoning interest in management research in Africa (Arslan et al., 2021;George et al., 2016;Nkomo, 2017;Zoogah, et al., 2015a), including studies that explore the application of existing organisational theories in the African context (Beugré, 2020). In this context, George et al. (2016, p. 386) has stressed the importance of analysing un-researched management topics (e.g. feedback and leadermember exchange [LMX] quality) in African context, as such research can extend and enrich the existing management theories.
In this vein, the current study examines the influence of various feedback delivery tactics on subsequent LMX quality in selected top supermarkets in Nigeria, South African and their UK counterparts. Selecting supermarkets from these three countries fits well into the increased interest in examining how various leadership dimensions and LMX can complement and influence each other (Boer et al., 2016;Carter et al., 2013), especially in various settings. This decision to draw these supermarkets from three different countries was also to help provide the researchers with the much-needed data for comparing the application of existing organisational theories in the UK's top supermarket and the African contexts (Beugré, 2020), which will help to boost management research in Africa and thus place the continent in the mainstream management literature (ibid).
Although follower perception of supervisory feedback has been linked to LMX quality (Sparr and Sonnentag, 2008;Thuan, 2021), we extend the focus of the literature linking feedback with LMX relationship beyond the confines of follower perception of feedback quality to incorporate the importance of the supervisor's feedback delivery tactic. Based on a study of 40 feedback recipients (supermarket store managers) in Nigeria, South Africa and the UK, this study explores what influences different tactics of feedback delivery have on subsequent LMX relationship.
Feedback delivery tactic is any chosen approach to convey a supervisor's reaction to a subordinate's job performance. This could be crucial in a dyadic relationship, as criticism delivered with greater interpersonal fairness can result in a more positive feeling towards the supervisor, increased feedback acceptance and more positive reactions towards the organisation (Leung et al., 2001) and thus high LMX quality. Therefore, supervisors use feedback to regulate subordinates' emotions (Qian et al., 2016a(Qian et al., , 2016b. Nevertheless, positive feedback has been found to inspire employee knowledge-sharing behaviour (Li et al., 2014), while the frequency of negative feedback is linked to job anxiety (Sparr and Sonnentag, 2008). Constructive and timely feedback has also been linked to continuous learner improvement (Retna and Cavana, 2013;Jada and Mukhopadhyay, 2019). Earlier LMX studies (Chun et al., 2014) have also examined how leader credibility (as a dyadic characteristic) influences subordinates' decisions on seeking performance feedback from the leader. Later, LMX studies (Arain et al., 2020) link abusive supervision to feedback avoidance and, which in turn, leads to help seeking from co-workers. Given that the supervisor's feedback delivery manner can influence followers' emotions, it is surprising that little is known about how the supervisor's feedback delivery tactic can influence the relationship between the dyadic partners and which these authors suggest is a significant omission in the literature.
The feedback literature distinguishes between two major categoriespositive and negative/ unconstructive feedback. Positive feedback provides new information that reduces followers' uncertainty in the organisational environment (Daft and Lengel, 1986;Morrison, 1993). Therefore, it should improve the dyadic relationship between leaders and followers and is thus a precursor of high-quality LMX relationships. Negative feedback or unconstructive criticism can have an undesirable effect on employees' attitudes and behaviours (Tata, 2002) and which thus could be an antecedent of low-quality LMX relationships. Yet, there is a lack of empirical evidence examining the supervisor's approach to delivering criticisms to produce a more favourable (employee) reaction towards the superior, hence high-quality LMX relationship.
The LMX literature emphasises various antecedents in determining the nature of LMX relationships van Breukelen et al., 2006). The relationship between feedback-seeking behaviour (FSB) and LMX relationship quality has also been explored (Lam et al., 2007(Lam et al., , 2017Lam et al., 2017;Lam et al., 2007). Although the feedback environment has been linked with LMX relationships (Peng and Lin, 2016), the influence of the supervisor's feedback delivery tactic as an antecedent of LMX relationship quality has been relatively overlooked and which we suggest is a significant omission in the theoretical development of the LMX literature.
Although the LMX literature (Lam et al., 2017) has examined how LMX moderates the relationship between FSB and performance, they focused on how LMX could improve subordinates' job performance through their improved FSB towards leaders, rather than how the supervisor's feedback delivery tactic affects the quality of the LMX relationship. Peng and Lin (2016) also explored links between feedback and LMX but focused on how positive LMX can help supervisors to provide favourable feedback that could enhance organisational citizenship behaviour (OCB) and minimise workplace deviant behaviour (WDB), rather than how the supervisor's feedback delivery tactic can affect the quality of the LMX relationship. Although research linking LMX to human resource development (Sommer and Kulkarni, 2012) has linked constructive feedback to predicted better mood at work, greater job satisfaction and stronger OCB intentions, there is limited research examining how constructive feedback can influence subsequent LMX relationships. The current paper contributes to filling such a research gap.
Our study highlights the consequences of different feedback delivery tactics on subsequent LMX quality, especially in the African context. It offers three major contributions to the extant leadership and management literature streams. Firstly, based on the raw data, we develop a process-based model of enhancing high-quality LMX, which shows the role of the supervisor's feedback delivery tactic in the process. Secondly, we also develop a process-based model that illustrates how negative/unconstructive feedback could result in a low-quality LMX. Finally, our paper is also one of the first to offer a comparative assessment between African and British (the UK) empirical settings and highlight some interesting dynamics concerning LMX quality and role of supervisor's feedback delivery tactic.
The rest of the article is structured as follows. We begin with a review of the relevant literature, before turning to a discussion of the methodology used. After that the discussion on data collection and analysis is presented. We conclude by considering implications for future research on leadership behaviour, leadership development and employee development more generally.
2. Literature review 2.1 Feedback as relational feature of work Relational leadership theory (RLT) provides a valuable framework for examining leaderfollower relationships. Uhl-Bien (2006) theorises two perspectives to RLTan entity perspective (with a focus on identifying and unpacking the leadership qualities of managers as they engage in interpersonal relationships with subordinates) and a relational perspective Leadermember exchange relationships (which explores leadership as a social construction process that produces particular outcome)to help our understanding of the theory. She suggests two main lines of inquiryhow do relational dynamics contribute to structuring LMX relationships, and how are leadership relationships produced? Drawing upon RLT, the present study examines how the supervisors' approach to delivering feedback can influence subsequent LMX relationships.
Furthermore, relational mechanisms are also crucial in understanding how individuals' perceptions of their role performance can result in thought processes that influence connections to work, colleagues and supervisors (Grant and Parker, 2009;Kahn and Heaphy, 2014). Relational mechanisms emphasise how relational characteristics (e.g. mutual trust, relational features of work and relational commitment) can affect dyadic partners' behaviour and organisational effectiveness (Lavie et al., 2012). Performance feedback is a relational feature of work, and how subordinates perceive supervisory feedback reflects both the degree of satisfaction with role performance and connections to work and colleagues and how the two members of a dyad are affiliated. Studies on the relational environment have found employees' perception of a supportive and trusting managerial environment enabling interpersonal relationships (Kahn and Heaphy, 2014;Newman et al., 2015), which helps define manager-subordinate affiliation. Positive feedback confirms that a subordinate's actions are acceptable to the supervisor, thus fostering a positive experience and minimising emotional exhaustion for an individual (Grant, 2007). It further leads to a high-quality LMX relationship. Subordinates can perceive negative feedback as either highly valuable and relatively unthreatening to their self-concept or as valuable but also highly threatening (Zingoni and Byron, 2017). Negative feedback is also perceived as a sign of a strained relationship with their supervisors (Dulebohn et al., 2012;Holmvall et al., 2019) and thus can be a precursor of low-quality LMX relationship. This paper builds upon these studies and proposes that employees' perception of positive feedback and constructive criticism indicates the level of prosocial influence they feel they have on their supervisor (Grant, 2007(Grant, , 2008 and will also positively influence subsequent LMX relationships.

Theorising performance feedback
In the context of the LMX relationship, feedback can be seen as valuable information provided by leaders to their followers, or sought by followers from their leaders, to identify whether and how well followers are meeting their performance targets and with a view to facilitating improved follower performance (Chun et al., 2018). This definition stresses the importance of constructive criticism as a performance improvement technique. The use of feedback as a performance improvement technique is also well-established in the human resources management (HRM) literature (Pichler et al., 2020), but the application of feedback is sometimes flawed, resulting in subjective and unreliable outcomes (Hensel et al., 2010). The use of performance feedback has been examined widely, and recent studies have linked effective feedback to improvements in care quality in nursing homes (Hoben et al., 2017), computer-based training simulation (Kim, 2018), basic cognitive tasks (Moret-Tatay et al., 2016) and favourable appraisal reaction (Pichler et al., 2020). Nevertheless, despite continued growth in the performance feedback literature, we lack consensus on what techniques are crucial for achieving effective feedback (Pichler et al., 2020;Johnson, 2013) or how feedback influences the behavioural change of followers (Bechtel et al., 2015;Johnson, 2013). While feedback has been linked to performance improvement (Porta and Last, 2018), empirical studies have produced contradictory findings. Jawahar (2010) argued that performance is influenced by reactions to feedbacknot feedback per seand called for organisations to design effective appraisal systems while training their managers on how to conduct effective feedback discussions. The current study has the potential to inform organisational interventions aimed at training their managers on how to conduct effective feedback discussions.
2.3 Linking performance feedback to leader-member exchange relationships quality A fundamental tenet of LMX theory is that leadership effectiveness is dependent on the quality of interaction between the dyadic partners. Studies (Burns and Otte, 1999) have found a heterogeneity of dyadic relationships. Due to leaders' limited power, choices and decision-making time, they tend to focus their attention and interactions on those subordinates whose roles are likely to maximise the team's goals. Earlier work in LMX theory posited that supervisors effectively categorise their subordinates into two clustersin-groups and out-groupsbased on their degree of closeness with them (Lam et al., 2017;Liden et al., 2006;Nishii and Mayer, 2009). While in-group members have higher-LMX relationships, they are expected in return to exercise higher discretionary efforts to perform above what is required (Brouer and Harris, 2007). The burden of these higher expectations is buffered by greater support and feedback opportunities (Dulebohn et al., 2012), mutual trust, respect, liking and greater attention from the leader (Lam, et al., 2017;Liden et al., 2006), reflecting high-quality LMX relationship (McNatt and Judge, 2004;Wayne et al., 1997). A leader's relationship with in-group members increasingly engenders feelings of social exchange, mutual obligation and reciprocity (Chang and Cheng, 2018;Liden et al., 1997), and hence, a high-quality LMX relationship is established. Conversely, Blau (1964) suggested the relationship between a leader and out-group members is a mere economic exchange relationship, based strictly on adherence to formal rules, employment contracts and performance-related payhence a low-quality LMX relationship. More recent work suggests that rather than treating subordinates as a group, leaders form and maintain a separate dyadic relationship between themselves and each subordinate (Lam et al., 2017;Liden et al., 2006;Nishii and Mayer, 2009), with variation between dyads in terms of the levels of interactions which prevail (Dulebohn et al., 2012).
While leadership quality is found to be central in determining LMX quality, the influence of followers in the process must not be ignored (Lapierre et al., 2006;Martin et al., 2005). Leader characteristics and behaviours directly influence follower attitudes and behaviours and vice versa (Dulebohn et al., 2012;Holmvall et al., 2019), as both members of a dyad form opinions about their dyadic partner, which in turn influences individual feedback on each other (Engle and Lord, 1997;Lord and Maher, 1999). The mutual evaluations of dyadic partners are found to be an antecedent of LMX quality (Snodgrass et al., 1998;Maslyn and Uhl-Bien, 2001;Sin et al., 2009). Feedback would logically capture many of these evaluations, but there is a lack of empirical studies that examine how a supervisor's approach to feedback delivery can influence subsequent LMX quality . Although researchers have extensively examined the role of feedback within HRM, until recently, studies mainly focussed on feedback techniques (Hensel et al., 2010), frequencies (Pichler et al., 2020) and targeted performance (Ammons, 1956;Jordan and Audia, 2012), ignoring how feedback can influence the quality of LMX (Van Breukelen et al., 2006). Earlier feedback studies focussed mainly on feedback received following formal performance appraisals (Lam et al., 2017), but the feedback environment has been redefined to incorporate related aspects of daily manager-subordinate and employee-employee relationships and organisational commitment (Peng and Lin, 2016). There have been renewed attempts to link feedback to outcome variables of value to the organisation, such as OCB (Han et al., 2018;Norris-Watts and Levy, 2004;Rosen et al., 2006;Whitaker et al., 2007).
Given the benefits attributed to the development of high-quality LMX (Lee et al., 2019;Tse et al., 2008), it is surprising how the supervisor's feedback delivery tactic can influence Leadermember exchange relationships the relationship between leaders and followers lacks the desired attention. One of the few relevant studies was undertaken by Seifert and Yulk (2010), who examined how feedback influences supermarket managers' behaviour changes, based on a small-scale study in a regional grocery store chain in the Northeastern United States. They call for further research, including multi-site case studies, and the present paper responds to this call.

Feedback environment, feedback culture and leader-member exchange
To define and conceptualise the feedback environment, Steelman et al. (2004) have developed the feedback environment scale, which consists of the following dimensions for measuring the effectiveness of the feedback environment. Firstly, the recipients must be convinced that the source is credible and knowledgeable about the feedback topic. Secondly, the quality of the feedback received must be high and delivered thoughtfully. Thirdly, there must be a blend of positive and negative feedback when necessary. Then, the source must be available and ready for consultations and conversations regarding the feedback on a regular basis. The source must also be actively promoting and encouraging feedback seeking. The scholars have also stressed the influence of a supportive supervisory environment on recipients' perception of the feedback received (Rosen et al., 2006) in fostering a positive behavioural change in the feedback recipients (Parker and Collins, 2010).
Although the feedback environment is still a relatively new concept (Dahling and O'Malley, 2011), earlier studies on feedback culture -"the environment in which the feedback is sought and received" (Levy et al., 1995: 24) link positive feedback culture to high-quality LMX relationship. Consequently, leaders are encouraged to create a feedback culture where new and useful information that fosters rapid performance improvement is delivered (Lam et al., 2017). The feedback literature also examines the role of leadership in setting goals, maintaining good working relationships (Bass, 2008) and providing timely and appropriate feedback (Ashford et al., 2016). Linked to formal performance appraisal (Ashford and Cummings, 1983), timely feedback shows subordinates how well they are meeting their performance targets (Geister et al., 2006). Although failure to deliver feedback in a timely fashion could represent a missed opportunity for performance improvement (Mahsood et al., 2018), subordinates still face difficulty in receiving feedback in a timely fashion (Anseel et al., 2015). This highlights the need for more studies examining how the feedback environment can be (re)designed to influence performance improvement, subordinate's perception of the feedback received and thus a high-quality LMX relationship.
As the original theorising on LMX emphasises a dyadic relationship that is built on mutual trust, respect and liking between the dyadic partners (Graen and Uhl-Bien, 1995), recent research (Chun et al., 2018) links LMX quality to subordinates' evaluation of leader effectiveness, based on feedback received. Similarly, a positive relationship and favourable perceptions (by the followers) enhance leaders' performance evaluations and reputations within the organisation (Bass, 2008). Yet, there is a paucity of research examining whether this is a sign that a leader is actual fulfilling their part of the mutual obligation (Wilson et al., 2010) or is the leader using impression management to take advantage of a high-quality LMX relationship (Wayne and Green, 1993). The literature linking feedback environment with LMX quality (Peng and Lin, 2016) also links a reduced WDB and increased followers' positive expression of OCB to high-quality LMX. In particular, when a leader and a follower form a good relationship, the leader would likely provide favourable feedback that may result in more OCBs and less WDBs (ibid). Research also links a follower's decision whether to seek feedback (or not) from a leader to the leader's credibility (Chun et al., 2014). This is due to the daily effects of shame on employee well-being and performance (Daniels and Robinson, 2019), owing to negative feedback received from abusive supervisors (Shen et al., 2019;Xing et al., 2021). The negative effects of such an organisational shame (Daniels and Robinson, 2019) on employee commitment (Audenaert et al., 2021), FSB and LMX quality (Auh et al., 2019;Audenaert et al., 2021) have been emphasised.
2.5 Linking feedback-seeking behaviours to leader-member exchange: role of leadership Given the wide range of benefits of feedbacksurvival, adaptation and performance improvementsubordinates are strongly encouraged to seek actively, instead of wait for, feedback (Qian et al., 2017). Yet, there is still a lack of consensus on the factors that influence FSB. Although feedback seekers' personal characteristics and other contextual factors are argued as antecedents of FSB (Anseel et al., 2015), given the crucial role of a leader in individual followers' lives (Chen et al., 2015), recent studies have been examining the role of leadership in employees' feedback seeking (Qian et al., 2012(Qian et al., , 2016a(Qian et al., , 2016bAnseel et al., 2015;Dahling et al., 2015). In particular, the positive influence of the transformational (Anseel et al., 2015;Fenwick et al., 2019), transactional (Ng and Feldman, 2014), authentic (Qian et al., 2012) and ethical leadership (Qian et al., 2017) on both the feedback climate as a whole and on subordinates FSB have been examined. For instance, linked to the social exchange theory, as the ethical role models (Qian et al., 2017), the ethical leadership style can influence subordinates FSB, as receipt of ethical treatment can inspire a feeling that a follower is trusted by their leaders and thus can trigger FSB. Furthermore, a perception of integrity, trustworthiness, safety and fairness (Qian et al., 2017) indicates a high-quality LMX relationship. Therefore, the follower is more likely to reciprocate such a favourable treatment (Brown and Treviño, 2006;Brown et al., 2005), which they do through improved performance (Walumbwa et al., 2011a(Walumbwa et al., , 2011b and enhanced FSB (ibid). Given such influence of the ethical leadership on FSB (Qian et al., 2017), there are calls for more research that examine the influence of other contextual factors (including feedback delivery tactics) on feedback quality (Eisenbeiss and van Knippenberg, 2015) and subsequent LMX quality. Our study responds to such calls.
Due to the motivation to attain the desired outcome (Ashford, 1986), recent research (Lam et al., 2017) has found that subordinates now go beyond formal organisational channels to seek feedback through various informal sources. Earlier research (Aarons, 2006;Aarons et al., 2012) also examined how a supervisor's leadership behaviour can influence followers' attitudes towards feedback seeking. Through a supportive supervisory relationship (Lam et al., 2017), supervisors can encourage employees to be self-regulatory (Ashford and Tsui, 1991) and proactive in seeking useful and valuable feedback (Parker and Collins, 2010), which can make the difference in individual and group performance. In particular, transformational leaders inspire followers' pursuit of a particular course of action (Bass et al., 2003), including feedback seeking. Nevertheless, the charismatic appeal of transformational leaders also lures followers to interaction, which is key in nurturing higher quality LMX (Wang et al., 2005). Although the LMX theorising view leadership as key in influencing followers' perceptions about feedback quality (Ashford, et al., 2016), there is a lack of empirical evidence examining whether various feedback delivery tactics can make a difference in how followers perceive the quality of feedback received and which these authors argue is a significant omission in the theoretical development of the feedback literature.
Although research links FSB to performance improvement (Fenwick et al., 2019), successful change implementation (Douglas et al., 2016) and follower loyalty (Bearman et al., 2017), individual attitudes towards feedback may vary (Fenwick et al., 2019), which highlights the need for supervisors to influence individual followers' willingness to seek out and apply feedback (Anseel et al., 2015;Ashford, et al., 2016). Yet, employees may sometimes not know the appropriate sources of feedback. Although the feedback literature views the Leadermember exchange relationships leader and co-workers as the two primary sources of feedback (Morrison, 1993;Williams et al., 1999), due to their key influence in the organisational control system (Green et al., 1986), leaders, are, evidently, the more significant source (Greller, 1992;Lam et al., 2013;Larson, 1989). Recent research (Lam et al., 2017) also links supportive supervisory relationships, FSB, feedback quality, improved job performance and LMX. Yet, followers also pay attention to how leaders differentiate between followers through varying levels of communication and feedback quality (Lau and Liden, 2008). The LMX literature also recognises such differentiated relationships with followers (Graen and Uhl-Bien, 1995;Liden et al., 2006;Nishii and Mayer, 2009;Sparrowe and Liden, 1997), through the varying degree of leader support and feedback (Lam et al., 2017). This links to varying qualities of employee performance and LMX relationship within each group. Despite these theorising linking leadership, feedback and LMX quality, there has been a renewed call for more research that examines how various leadership dimensions and LMX complement and influence each other (Boer et al., 2016;Carter et al., 2013). By examining how leaders' varying feedback delivery tactics can influence subsequent LMX quality, this study responds to these calls.

Research methodology
The present study gathered data from the store managers of supermarkets in Nigeria, South Africa and the UK. Feedback is seen as central to performance in this sector, with the Supermarket Experience Study (Retail Feedback Group, 2015) attributing the high level of customer satisfaction in supermarkets (an average of 4.44 out of 5.0) to several retailing fundamentals, but especially effective feedback (Food and Beverage Close-Up, 2015).
The methodology adopted is qualitative. Standard criteria for evaluating the validity and reliability of qualitative research include dependability, confirmability, credibility and transferability (Gibbert and Ruigrok, 2010;Lee, 1999;Lincoln and Guba, 1985), all of which require researchers to provide a comprehensive analysis of the concrete research actions undertaken (Gibbert and Ruigrok, 2010). In this section, we detail the research activities performed in the processes of the data collection and analysis, demonstrating the standard of rigour involved and thus the validity and reliability of this study (Creswell, 2007;Gioia et al., 2012;O'Reilly et al., 2012).
With a well-defined target group of participants (in this case, store managers at the top supermarkets in Nigeria, South Africa and the UK), purposive sampling is the recommended approach (Wong, 2008). We chose only supermarkets recognised as among the top performers in their respective countries, thus ensuring they were highly competitive and of international standard. In a similar manner, we developed criteria to guide the selection of the store managers. Based on each supermarket's managerial hierarchy, we selected managers whose role was clearly a store manager who receives feedback from the senior/top management. Data were gathered through interviews with 40 store managers from top supermarkets in Nigeria, South Africa and the UK. The authors believed it is important that the supermarkets are drawn from three different countries to help ascertain whether different objectives, staffing compositions, feedback processes and operational strategies would show variance in the supervisors' feedback delivery tactic and/or the participants' perception of feedback received. The interview participants were as follows: Nigeria, 14 participants across eight stores, all from different chains; South Africa, 16 participants across three stores, all from the largest chain in the country; and UK, ten participants across nine stores, drawn from the UK's four biggest chains.
All the Nigerian participants were drawn from Lagos State, the country's commercial nerve centre. The South African participants were drawn from three different stores, while all the British participants were drawn from North East England. The British supermarkets have been chosen due to their colonial legacies (Barnes-Dabban et al., 2017), as the British system of education has been the driving force for the socio-economic development of many African states (Archibong, 2019). Consequently, understanding and advancing leadership practice in African supermarkets requires following trends and significant developments in their British counterparts. Table 1 below provides more detailed profile of the respondents.

Data collection
We used semi-structured interviews with a set of guided questions, follow-up questions and multiple probes. This approach helps to ascertain individuals' experience in detail (McConville et al., 2018), which allowed us to explore in greater depth how the quality of LMX relationship links to supervisor's feedback delivery tactic. The interviews were structured around the following themes: organisations' strategic priorities on feedback, feedback and organisational performance, feedback as a relational mechanism, feedback and LMX quality and finally the participant's own feedback preference. In line with Liu and Rong (2015), we allowed each participant to elaborate on their views on any feedback received, including how it was delivered. The focus of the conversations was therefore: whether the participants receive feedback; the type received; the frequency; how the feedback influence their subsequent job performance; whether the feedback has influenced the interpersonal relationship between members of the organisation; Leadermember exchange relationships any useful information on the feedback environment; and to provide any information that could help the researchers determine if the supervisor's feedback delivery tactic has influenced subsequent LMX relationships.
The raw data was collected during store visits which took place between the months of September 2014 and June 2015. Interviews lasted for an average of 2 h and 20 min. Table 2 below illustrates the data collection and data analysis process.
3.1.1 Rationale for timeline for the interview. Although the data was collected in 2015, it is still valid. Based on Romaniuk et al.'s (2012) argument that a 45-year-old empirical generalisation can still be useful if the authors draw on a few new data sets to test the validity of such previous findings, we attempted to collect more recent empirical data from the three countries. Unfortunately, our participants in Africa agreed to participate only in face-to-face interviews and which has not been possible this time due to COVID-19 restrictions. However, unlike Romaniuk et al.'s (2012) study, our empirical data was only collected in 2015. So, to enhance both the validity and reliability of our data set, the current paper has been strongly supported by up-to-date literature and current research in the field.

Data analysis: procedures and steps adopted
The data analysis draws from both Alo's (2020) and Braun and Clarke's (2006) suggestions on the six-stage process of the thematic data analysis.
3.2.1 Data familiarisation. We commenced the data analysis process by first listening to the recorded audio interviews several times, until we became familiar with the raw data. This was followed by data transcription from recorded audio files into written files. We crosschecked all the written text against the audio interview data to ensure accuracy to ensure a thorough transcription of data. While no significant differences were found for ease of display (Alo, 2020), a few amendments (of the interview quotes) were made.
3.2.2 Generating initial codes. With all the interview data transcribed at this stage, we began to code the interview transcripts. To enhance their validity, the coding involved segmenting the transcribed data (based on their similarities in meaning) into various units Table 2. Data collection and analysis process Case study steps Activities Step 1 Establish the focus and scope of the research Step 2 Develop the research questions Step 3 Decide the individual supermarkets to include in the "multi-site case studies" Step 4 Decide the appropriate research instruments and protocols, e.g. the appropriate qualitative data gathering techniques, in this case, the semi-structured interviews Step 5 Determine the "suitable" participants: a vertical and horizontal slice of the case studies to establish whether each prospective participant is a store manager and not an owner manager Step 9 Data analysis commences (see below for the steps involved) Step 10 Dissemination: report and article development of more meaningful texts. However, the process of coding the data was an iterative one, continuing until adequate distinguishing categories emerged between the various data groups (Eisenhardt, 1989). This thorough coding process has helped us align the related raw interview data with their corresponding themes during data analysis and has thus helped enhance the validity of the study. Although we coded each group/subgroup of data separately, a small number of mix-ups still emerged. This required that we revisited the raw interview data on a few occasions, to modify and recode the data, before realigning them with their corresponding content themes. 3.2.3 Searching for themes. The third stage of our data analysis process involved creating a more understandable expression of the data sets. Based on corresponding theories, the key issues addressed in this study and the similarities in meanings of our data sets (Alo, 2020), we, therefore, identified four distinctive statements (the main themes) to show how the supervisors' feedback delivery tactic influence the LMX relationship quality. To enhance validity, these themes and their corresponding units of texts required scrutiny (Alo, 2020). At this stage, we, therefore, needed to use the services of a team of experienced qualitative researchers to vet the data analysis process.
3.2.4 Reviewing the themes. Therefore, the fourth stage involved using the expertise of a team of three well-experienced qualitative researchers to perform expert checks. These three well-trained qualitative researchers acted as both critical friends (Kember et al., 1997) and research auditors (Filho and Rettig, 2016) and who provided us constructive feedback following each peer debriefing meetings we had. In a handful of cases, we had to recode and regroup a few data units, until consensus was reached between us and our three critical friends.
3.2.5 Re-definition and re-naming of themes. Following a range of scrutiny (by our three critical friends), they recommended that we regroup, redefine and clarify some of our themes. So, the fifth step in our data analysis process involved redefining and renaming the themes. This stage in the data analysis process has helped us maximise the themes (Alo, 2020), thus enhancing the study's validity. It also helps the readers understand the significant relationships among the various constructs in our study.
3.2.6 Report writing. An exhaustive report writing is key to a thorough interpretation and clarification of the link between the data and the corresponding literature. With our conscientious report writing effort, we further clarified the connections between the interview responses and the feedback and LMX literatures. Liu and Rong (2015) also recommend ensuring an extensive comparison and connection between the findings and the dominant theory. This involved repeatedly moving forward and backwards within the empirical data and the literature to facilitate a thorough analysis of the data and a strong connection between the results and the dominant theories. This has helped to enhance both the validity and reliability of our findings. Our painstaking report writing effort has also helped us produce a relevant discussion section and an insightful contribution of this research study.

Findings
Before considering, what the data shows about the relationship between supervisor's feedback delivery tactic and LMX, we will briefly highlight some of the observable patterns in when feedback is given. In terms of feedback frequency, there was a pronounced difference between countries in how often store managers received feedbackdaily in Nigeria, weekly in South Africa and quarterly in the UK (the latter often linked to the quarterly review). It is interesting to consider which frequency is preferable and whether cultural differences influence this. Although some studies suggest feedback should be Leadermember exchange relationships provided as soon as possible following any outcome that warrants feedback (Jawahar, 2010), other studies suggest professionals achieve the best outcomes when they receive detailed but infrequent feedback (Casa-Arce et al., 2017;Casal et al., 2017).
In coding the interview transcripts, we observed that many quotations could be coded against several themes, reflecting the extent to which participants discussed their supervisor's feedback delivery tactic in rather holistic ways. This is perhaps unsurprising, as feedback is a source of information that can be rich and nuanced. In making sense of the data, we identified four themes relating to the different information which participants obtained from their supervisors' feedback, which we termed confirmation, development, reciprocity and relationship. In this section, we use these four constructs to link supervisor's feedback delivery tactic, subordinate's perception of feedback received and subsequent LMX relationships.

Confirmation
While the decision to draw the supermarkets from three different countries was to help the researchers establish whether any variance in supervisors' feedback delivery tactic and/or the participants' perception of feedback received can be attributed to differences in objectives, staffing compositions, feedback processes and operational strategies, in contrast, there is a typical pattern of response. The responses show store managers commonly view feedback as offering information on whether their behaviour and actions were acceptable to their line manager. In general, several reviews of the interview data set from the three countries could not reveal any obvious differences in the participant's responses; as such, a comparative evaluation of the data from these three countries would not make any sense. Thus, the authors believe that a combination of responses (from the three countries) would make the most sense: Although some of the above quotes could represent superficially simple information, they are important for all staff but especially for managers, who must exercise judgement and discretion and, thus, would need constant reassurance they are performing their role acceptably (i.e. positive feedback). Such reassurance and clarity on performance fosters a positive experience and minimises emotional exhaustion for individuals (Grant, 2007). Participants (e.g. UK participant 10 and Nigerian participant 4 above) were also happy to be told politely how and why they need to improve (i.e. through constructive criticism). However, more than 60% of the participants (including those who rarely receive negative feedback) also have a common view on negative feedback. The quotes below illustrate perceptions on negative feedback as a confirmation that their efforts are not appreciated: [. . .]  A strict adherence to formal rules (by a supervisor) could arouse feelings of a mere economic exchange relationship (with a subordinate) (Blau, 1964)hence a low-quality LMX relationship. Participants also think that negative feedback influence the way they think about their supervisor and/or the organisation and thus a damage in their LMX relationship: [. . .] Not exactly that we don't get on very well, but one thing is that everyone likes to be appreciated for a job well done, not only to be scolded when things go wrong [. . .] (UK participant 8).
Negative feedback has been linked to a strained relationship with the supervisor, as follower attitudes and behaviours are directly influenced by leader characteristics and behaviours and vice versa (Dulebohn et al., 2012;Holmvall et al., 2019) and thus a low-quality LMX prevails.

Development
Feedback also provided information on areas in which they could improve their performances. Participants tended to link this to confirmation, suggesting that feedback on specific areas for improvement was received best when building upon overall feedback, indicating acceptable performance. The identification of potential for development was also linked by some participants to support for their career progression, thus, a perceived obligation to return such a favour: Participants linked feedback to subsequent development interventions, perceiving this as evidence of an organisational environment that facilitates leadership support for employee development and career progression: [. . .] following a successful appraisal, for deserving employees who wish to further their education, the company gives grants and bursaries to further their education in related areas to the company's operations (such as retailing, business management, etc). in Higher Institutions of learning. . .this is very appealing to every beneficiary [. As can be perceived from the above quotes, perceptions of greater support for development (Dulebohn et al., 2012) provoke feelings of social exchange, mutual obligation and reciprocity (Chang and Cheng, 2018;Wayne et al., 1997). On the contrary, when such an expected support for development and career progression is missing, feedback is perceived as negative, but also as an indication of lack of mutual trust, lack of respect, lack of liking and lack of attention from the leader/organisation and thus a high turnover intent prevails: . . .if you ask my colleagues here, I am sure many will also tell you that this job is only good for someone who is still searching for their dream job, as there are limited opportunities to learn and grow (Nigerian participant 8).
Another participant from South Africa added: [. . .] when we say we would like to leave this job if we can find a greener pasture elsewhere that does not mean we do not like retail job, we do, but you see, it is not good for us to remain here and keep watching the top management hire senior managers from outside and only to be told we lack the experience to become a senior manager. . .imagine if you are in our shoes, will you be happy? (South African participant 13).
When feedback reveals a lack of mutual trust, lack of respect, lack of liking, lack of attention and lack of support from the leader, a low-quality LMX relationship prevails, as follower attitudes and behaviours are directly influenced by leader characteristics and behaviours (Dulebohn et al., 2012;Holmvall et al., 2019).

Reciprocity
A common theme across the data was the reciprocity in relationships, where each member of a dyad is obliged to return any benefit (in this case valuable feedback) received from the other partner. As the participants were managers themselves and knew how time consuming and potentially difficult it is to provide feedback, they viewed their manager's efforts in providing constructive feedback as a reciprocation of their efforts. Accordingly, participants feel an obligation to return such a favour through improved performance and commitment, which contributes to developing high-quality LMX relationships with their supervisors: Earlier LMX studies  had emphasised such a reciprocity, and which our study has found to be characteristic of a high-quality LMX relationship. On the contrary, perception of negative feedback can stimulate intent to leave the organisation: [. . .] if you ask my colleagues here, I am sure many will also tell you that this job is only good for someone who is still searching for their dream job, as there are limited opportunities to learn and grow (Nigerian participant 8).
Some participants also seem to show some form of recalcitrant behaviour due to negative feedback received, e.g.: [. . .]  Both members of a dyad evaluate the feedback received from each other (Snodgrass et al., 1998;Maslyn and Uhl-Bien, 2001;Sin et al., 2009) and which is the basis for subsequent opinion formed about their dyadic partner (Engle and Lord, 1997;Lord and Maher, 1999) and thus the LMX relationship quality.

Relationships
Feedback was one of the main points of interaction with their managers and, therefore, a key source of information about their relationship. Performance appraisal provides an opportunity to review the quality of relationship between the dyadic partners: [. . .] people who learn so fast would be made to go on the FastTrack so they can do the training within half the normal time (i.e. within 18 weeks or 12 weeks instead of a year) [. . .] making you to feel that your potentials are recognised (UK participant 9).
As the above quotes illustrate, perceptions of positive feedback boosts subordinate's conviction that they possess the requisite competence to "organise and execute the courses of action required to produce a given attainment" (Bandura, 1997, p. 3), thus, resulting in higher self-confidence (Burns, 1978) at work. This, therefore, improves the relationship with colleagues and supervisor. Due to mutual evaluation of dyadic partners (Snodgrass et al., 1998;Maslyn and Uhl-Bien, 2001;Sin et al., 2009), perceptions of negative feedback also indicate the degree of a strained relationship with the other partner (Lam, et al., 2017;Nishii and Mayer, 2009) and is thus an antecedent of low-quality LMX relationship: I honestly think that if he (i.e. the supervisor) doesn't like you, nothing else can change that [. . .] it is not a matter of trying to do things differently, [. . .] after all I know a number of colleagues who I am better than in terms of our job experiences but who have been promoted to senior management posts due to their positive reviews [. . .] (South African participant 16).

Leadermember exchange relationships
The problem I have been having with him (i.e., the supervisor) is that he always wants me to do things his own way, even when I prove to him that there are other better ways of doing it [. The quotes above echoes Dulebohn et al.'s (2012) and Holmvall et al.'s (2019) findings that follower attitudes and behaviours are directly influenced by leader characteristics and behaviours and vice versa. Lapierre et al. (2006) and Martin et al. (2005) also found attitudes of both members of a dyad as equally central in determining LMX relationship quality.

Discussion
This study sought to extend the focus of the literature linking feedback with LMX relationship beyond the confines of follower perception of feedback quality to incorporate the importance of the supervisor's feedback delivery tactic. Drawing on the theorising that subordinate cognitions on supervisory feedback arises from aggregated personal observations of leader behaviour (Zohar and Polachek, 2017), this study has found two distinct subordinate perceptions of supervisory feedback and their dissimilar influences on LMX relationship quality. Firstly, participants perceive positive feedback and constructive criticisms as indicating that their actions are acceptable to their supervisors, thus, expressing the level of prosocial influence they feel they have on their supervisors (Grant, 2007(Grant, , 2008. This helps in confirming a sense of connection to their supervisors (Grant and Parker, 2009;Kahn and Heaphy, 2014), thus leading them to feel an obligation to reciprocate, which results in high-quality LMX relationships. Valuing and appreciating the other's contributions and telling them politely how and why they need to improve are also a measure of relational affection between the two members of a dyad and is thus evidence of a high-quality LMX relationship (Conway et al., 2020).
Previous studies have not investigated the psychological processes by which the supervisor's feedback delivery tactic contributes to developing high-quality LMX relationships. From participants' responses, we found a logical sequence of how positive feedback and constructive criticisms link to a high-quality LMX relationship. Positive feedback and constructive criticisms lead to: a perception that their actions were acceptable to the supervisor; a perception of a supportive and trusting managerial environment; a felt need to reciprocate through their behaviour in role; and facilitated development of a high-quality interpersonal relationship with the supervisor. Figure 1 above shows that employees' perception of a supportive and trusting managerial environment is the foundation for high-quality LMX relationships, and constructive criticism and positive feedback play a key role in shaping these perceptions. However, the reciprocal interdependence theory indicates that both members of a dyad continuously shape their working relationships through exchange of information, which shapes their subsequent LMX relationships (Blau, 1964;Graen and Uhl-Bien, 1995;Maslyn and Uhl-Bien, 2001;Dulebohn et al., 2012;Lam et al., 2017). Our second finding (in Figure 2 below) shows that negative feedback and unconstructive criticisms also shape such subsequent LMX relationships between the dyadic partners. Negative feedback and unconstructive criticisms lead to: a perception that subordinate's efforts were unacceptable to the supervisor; a perception of an unsupportive and untrusting managerial environment; a provoked feeling to quit or to retaliate through recalcitrant behaviour; and facilitated development of a low-quality interpersonal relationship with the supervisor.
This study has found that employees perceive negative feedback and unconstructive criticisms as confirmation that their efforts are not appreciated, and thus, a low-quality LMX relationship ensues. Subordinates expect supervisors to support their development and career progression (Steelman et al., 2004), but when feedback is perceived as negative, it indicates lack of mutual trust, lack of respect, lack of liking, lack of attention and lack of support from the leader (Sommer and Kulkarni, 2012). As negative feedback indicates a strained relationship with their supervisors, a low-quality LMX relationship prevails, as  Leadermember exchange relationships follower attitudes and behaviours are directly influenced by leader characteristics and behaviours (Dulebohn et al., 2012;Holmvall et al., 2019). Managers should, therefore, be trained to deliver constructive feedback. Previous feedback studies (Leung et al., 2001(Leung et al., , p. 1155 found that when criticism is delivered with greater interpersonal fairness, it results in more favourable dispositional attributions about the supervisor, more acceptance of the feedback and more favourable reactions towards the superior and the organisation. This results in a higher LMX relationship because "just interpersonal treatment reduced negative dispositional attribution, which in turn increased feedback acceptance and improved attitudes towards the supervisor". Furthermore, despite burgeoning evidence of the importance of feedback in follower commitment and LMX quality (Audenaert et al., 2021;Fenwick et al., 2019), there has been relatively little focus on supervisors' tactics of feedback delivery. Our study provides a starting point for understanding how supervisors' various feedback delivery tactics may influence follower attitudes towards feedback and subsequent LMX quality. Consistent with the original theorising on LMX, which focuses on the dyadic relationship between a leader and the follower and the role of social exchanges (between the dyadic partners) in creating and sustaining the quality of such a relationship (Graen and Uhl-Bien, 1995), we found that participants who received positive and constructive feedback felt an obligation to reciprocate such needed support from their supervisors through improved performance and commitment. With this, our study makes complementary but distinctive contributions to the extant literature on the importance of supervisory support and appropriate feedback in advancing evidence-based practices (Forman-Hoffman et al., 2017) and in influencing subsequent LMX quality (Fenwick et al., 2019). Supervisory support and appropriate feedback are also argued as key to subordinates' adaptation to changing workplace and to solving client problems (Bearman et al., 2017). Yet, a supervisor's leadership behaviour may play a key role in influencing subordinates' attitudes towards receiving supervisory feedback (Aarons, 2006;Aarons et al., 2012). Research has consistently established a strong correlation between LMX quality and subordinates' perceptions about supervision, reception to feedback and FSB (Ashford et al., 2016). Yet, there are paucity of empirical data that examines, if and how, a supervisor's feedback delivery tactics can influence subordinates' perception about the quality of supervision, reception to feedback and subsequent LMX quality. This study offers original insights needed in filling these gaps. This study also has the potential to inform leadership interventions aimed at increasing staff openness to feedback and FSB.

Conclusion and implications
This study has shown how the supervisor's feedback delivery tactic can influence the subordinates' perception of the feedback received and thus influencing the ensuing LMX relationships qualities. The perception of positive feedback and constructive criticisms can lead to high-quality LMX relationships. Although the perception of negative feedback can lead to a low-quality LMX relationship, criticisms can achieve greater acceptance and positive reactions (from subordinates) towards the superior and the organisation if delivered in a constructive way (Leung, et al., 2001). This also enriches the leader's performance review and enhances their reputation within the organisation (Bass, 2008;Herold and Fields, 2004).
These findings have practical implications for leadership behaviour, as well as leadership development and employee development more generally. Leadership, in its simplest form, "is grounded in a relationship" (Bennis, 2007, p. 3), so it is important for leadership development to support managers to develop behaviours that promote good interpersonal relationships with subordinates, such that high LMX relationships can develop (Dulebohn et al., 2012). The relational framework literature (Conway et al., 2020) suggests training managers on how to provide feedback that meets three outcomesrecognising those prosocial differences that individuals make through their work, offering more significant support for employees to cope with challenging work demands and providing information on developments within the organisation. Integrating the feedback environment into management development programmes and developing counselling and communication skills in managers who deliver feedback have also been suggested (Peng and Lin, 2016). Previous research (Van der Rijt et al., 2012) has also suggested fostering a supportive and mentally safe work environment and encouraging employees and supervisors to create enough time and opportunities to look for and give informal, constructive feedback.
This research also has some theoretical implications. Firstly, this research fits well into the increased interest in the links between feedback, survival, adaptation, performance improvement and LMX relationship, which is at the heart of the mainstream feedback and LMX literatures. At the heart of the current theorising is how HR can build organisational climate for supervisory support, feedback delivery and FSB and strongly encourage subordinates to actively seek, instead of waiting for the feedback (Qian et al., 2017). Although feedback seekers' personal characteristics and other contextual factors are argued as antecedents of FSB (Anseel et al., 2015), given the crucial role of a leader in individual followers' lives (Chen et al., 2015), recent studies have been examining the role of leadership in employees' feedback seeking (Qian et al., 2012(Qian et al., , 2016a(Qian et al., , 2016bAnseel et al., 2015;Dahling et al., 2015). By making complementary but distinctive contributions to these streams of management literature, this research also provides the much-needed insights to further the ongoing debate on the antecedents of FSB.

Limitations and future research directions
Our paper does have several limitations, like any academic study. Firstly, this study found that positive feedback and constructive criticisms will lead to a favourable LMX relationship. It is possible that the opposite causal relationship may also be true. It is possible that high performance and resultant high LMX relationship may already have been present when a supervisor delivers positive feedback or constructive criticism. Therefore, empirical research is needed to establish such a causality further. Such further studies should examine whether high performing subordinates can develop a favourable LMX relationship and receive favourable feedback because they are high performers and have a favourable LMX. An interesting research question is whether the causal order between positive supervisory feedback and high-quality LMX can be reversed.
Secondly, positive feedback may not be the only feature of feedback that is important in a relationship and, arguably, may not be among the least important. For instance, just to make a subordinate feel good and have a good relationship with them, any supervisor can say "good job". Therefore, an empirical data is needed to establish whether lasting LMX is more likely based on concrete, constructive, ongoing feedback processes. An interesting research question is to establish any causality between feedback frequency, feedback quality and lasting LMX.
The future research can focus on managers to ask questions about how they deliver feedback. Further research is also needed to inform our understanding of how supervisors can be trained to deliver positive feedback and constructive criticisms to maintain highquality LMX relationships. Further research can also focus on how employees can be trained to actively seek and apply supervisory feedback to help them continuously adjust their behaviours according to organisational expectations (London, 2003;Peng and Lin, 2016), which will create a feedback-oriented culture and enhance a closer manager-subordinate Leadermember exchange relationships relationship (Peng and Lin, 2016). Finally, our research focusses on primarily African context and data from the supermarket managers, which can result in the limitations concerning the applicability of our findings. The future studies can undertake similar research endeavours in other context to see whether their findings are similar or different from ours.