About the Author(s)

Claudia Coustas symbol
Gordon Institute of Business Science, University of Pretoria, Johannesburg, South Africa

Gavin Price Email symbol
Department of Academics, Faculty of Leadership, Gordon Institute of Business Science, University of Pretoria, Johannesburg, South Africa


Coustas, C., & Price, G. (2024). Factors influencing followers’ perceptions of the effectiveness of their leaders’ apologies. SA Journal of Industrial Psychology/SA Tydskrif vir Bedryfsielkunde, 50(0), a2170. https://doi.org/10.4102/sajip.v50i0.2170

Note: Additional supporting information may be found in the online version of this article as Online Appendix 1.

Original Research

Factors influencing followers’ perceptions of the effectiveness of their leaders’ apologies

Claudia Coustas, Gavin Price

Received: 05 Dec. 2023; Accepted: 16 May 2024; Published: 28 June 2024

Copyright: © 2024. The Author(s). Licensee: AOSIS.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


Orientation: Given the prevalence of apologies, it is crucial to consider how followers perceive the effectiveness of a leader’s apology.

Research purpose: This article conducts an empirical study on the factors that influence followers’ perceptions of their leaders’ apologies, following leaders’ wrongdoing.

Motivation for the study: This article maps the elements of an effective leader’s apology, as well as identifies the situational moderators of such apologies, which can help equip and empower leaders when they need to apologise.

Research approach/design and method: A total of 311 questionnaires, completed by followers from South Africa and around the world, were quantitatively analysed for the study. After conducting an exploratory factor analysis, a path model was developed, and partial least squares structural equation modelling was conducted.

Main findings: The quality of leaders’ apology content, the promptness of the apology, the perception of justice it evokes and the delivery channel all have a significant positive relationship with both the degree to which followers perceive the apology as authentic and the quality of the leader–follower relationship (LFR) after the apology. These relationships are moderated by followers’ perceptions of leader transgression preventability. The LFR quality prior to the transgression moderates the relationship between leader apology content, promptness, fairness and delivery channel on post LFR.

Practical/managerial implications: The study provides guidance on what leaders should include when formulating a quality apology, especially when followers perceive the wrongdoing as preventable. The study cautions against overreliance on LFRs prior to the wrongdoing.

Contribution/value-add: This study aims to fill an existing gap in empirical research on leaders’ apologies.

Keywords: leader apology; relationship repair strategy; justice; leader-member exchange; accountability; fairness.



Zoodsma and Schaafsma’s (2022) research revealed an upward trend in the instances of political apologies rendered by countries in the 20 years leading up to their study. On an organisational level, Wang et al. (2023) speak of the contemporary prevalence of ‘apology culture’ (p. 2).

Corporate apologies rendered in the recent past include PricewaterhouseCoopers (PwC) Australia and Volkswagen. PricewaterhouseCoopers Australia communicated confidential information as to upcoming taxation changes to clients (Foley, 2023). Volkswagen’s vehicle emissions figures proved much higher than those advertised by the company (Associated Press, 2016; Sillars, 2022). Bain and Company and the South African Revenue Services published apologies for their involvement in what has been termed ‘State Capture’ in South Africa (SA), as revealed in the findings of the Zondo Commission (Laurence, 2022; SARS, 2022).

Research within the field of corporate crisis communication is growing and considers the impact of the formulation and rendering of corporate apologies on company stakeholders to manage public image and relations (for example, see Bentley & Ma, 2020). Studies such as Shao et al. (2022) have considered factors influencing the perception of a good corporate apology and the impact of a good apology on organisational performance and other indicators.

On an interpersonal level, what has been coined ‘cancel culture’ has regained prominence in contemporary times, with individuals and organisations reportedly being cancelled for offensive actions that have been publicised (Bies et al., 2021). Leaders’ apologies to their followers constitute an instance of an apology on an interpersonal level. Researchers such as Epitropaki et al. (2020), Kahkonen et al. (2021) and Sharma et al. (2023) have conducted comprehensive literature reviews in the field of leaders’ apologies. While their syntheses (and Epitropaki et al.’s [2020] proposed framework) remain untested in their entirety, empirical studies have considered the relationships among variables encompassed by these models. Wang et al. (2023), Grover et al. (2019) and Hetrick et al. (2021) have investigated the relationship between leaders’ apologies and follower trust, transgression severity and intentionality and follower attributions for unintentional transgressions.

Consideration of what constitutes an effective apology and the factors influencing the perception of the same potentially assists leaders in empathising with those receiving their apologies and considering relevant factors in apology formulation, thereby better equipping leaders to render effective apologies.

In sum, the objective of the present study is to inform an empirically tested model to explain the factors affecting followers’ perceptions of the effectiveness of their leaders’ apologies. The boundaries of the study need to be defined. Firstly, the focus of the present study is on one group of stakeholders, namely followers (employees). Secondly, the study focusses on an interpersonal level, on leader (manager) apologies – or apologies rendered by a leader to one or more followers internal to the organisation. Thirdly, the efficacy of leaders’ apologies is considered from the perspective of followers as one particular group of stakeholders.

Research purpose and objectives

The purpose of the study is to investigate factors influencing followers’ perception of the effectiveness of leaders’ apologies. The quality of the leader’s apology, which includes the leader’s (cognitive, affective and conative) apology content, the promptness of the apology and the delivery channel through which it is delivered, serves as an independent variable. Followers’ perception is operationalised in two dependent variables, namely firstly, their perception of an apology’s authenticity and secondly, their relationship with their leader post-wrongdoing and apology. A final dependent variable to be considered is followers’ perception of the fairness of the apology, which is defined as the extent to which the apology is viewed as evidencing distributive, procedural and interactional justice. The moderating variables that are considered include, firstly, the leader-follower relationship (LFR) prior to the wrongdoing, and secondly, the type of transgression. The latter is defined as either intentional or preventable.

Literature review

Sverdrup and Stensaker (2018) highlighted that there are two main strands of apology research. In the first, researchers focus on apologies between individuals. In the second apology research strand, namely corporate apologies, researchers focus on apologies on an organisational level, considering groups of stakeholders. The present article falls within the first of these two strands.

Based on a literature review, Epitropaki et al. (2020) have proposed a leader transgression and subsequent relationship repair framework on an individual level. Their framework is comprehensive, including firstly, pre-transgression aspects such as the characteristics of the leader, of the follower and of the relationship between the two. Secondly, characteristics of the transgression itself are considered, as well as the individual’s cognitive and emotional evaluations of the same and of the impact on the LFR. Finally, post-transgression factors come into play, including relationship repair strategies. Epitropaki et al.’s (2020) study has not been empirically tested and does not clearly specify the elements of a leader’s apology.

In the realm of corporate apologies, Shao et al. (2022) developed a holistic corporate apology framework, by qualitatively analysing 40 corporate apology statements in conjunction with an analysis of existing corporate apology literature. The framework these authors developed is presented (exactly as it is presented in the original article) in Figure 1.

FIGURE 1: Conceptual framework of corporate apologies.

Shao et al.’s (2022) framework has not been empirically tested, and the latter framework focussed on corporate apologies rather than on leaders’ apologies. These authors’ framework included various stakeholders and was not specific to the impact of apologies on followers.

The following aspects of Shao et al.’s (2022) framework featured in the development of the questionnaire used for the current study: leader apology elements (LAE), authenticity, fairness (both seen from the perspective of the follower) and transgression type. The prior firm reputation was replaced by the prior LFR. Apology tradeoff, firm performance and spokesperson components were deemed not to be of relevance to leaders’ apologies.

Given that there is no empirically tested, comprehensive model to explain the impact of leaders’ apologies on followers, the question is posed, within the proposed research: can elements of Shao et al.’s (2022) framework be adapted and applied to leaders’ apologies in organisations?

The initial intention of the research was to test a model including 14 variables within Shao et al.’s (2022) framework, in the context of followers’ perceptions of leaders’ apologies in organisations. However, as will be unpacked with regard to methodology, the resultant measurement model limited the extent to which this was possible and determined which variables could be studied. These include LAEs, fairness, authenticity, prior and post-LFRs and transgression type.

Leader apology elements

Each of the content, timing and delivery channels has been elaborated upon here. Shao et al. (2022) differentiated among four types of apologies based on their content: incomplete, double-speak, full and extended. These four types are differentiated by the extent to which each focusses on one or more of the following kinds of content: cognitive (the expression of understanding of the nature and impact of harm caused, acknowledgement of wrongdoing), affective (the expression of emotions including regret, empathy or remorse) and conative content (reference to preventative actions or remediation).

The timing of an apology refers to the time elapsed between the occurrence of the wrongdoing and the apology being offered (Shao et al., 2022). Based on their literature review, Sharma et al. (2023) found that the passage of time post-wrongdoing in which no repair activity took place could have either a positive impact on trust repair (because the pain of the wrongdoing is dulled by time passing) or a negative impact on trust repair (by reducing the perception of procedural fairness).

With regard to delivery channels, Shao et al. (2022) differentiated between an apology delivered on a public versus a private platform. As applied to leaders’ apologies, a public apology would be one rendered on a publicly observable platform, be that in person in front of an audience or on a written or recorded platform such as social media to which others in addition to the aggrieved had access. A private apology would be given by the leader only to the aggrieved (Shao et al., 2022).


An apology perceived to be fair is one evaluated to communicate the leader’s empathy with the aggrieved in a timely and transparent manner (Shao et al., 2022). The concept of fairness as conceptualised by Shao et al. (2022) draws on the constructs of distributive, procedural and interactional justice.

Interactional justice refers to the extent to which the aggrieved individual feels that justice has been served in the interpersonal interaction during which the wrongdoing was addressed. Procedural justice, on the other hand, refers to the extent to which the aggrieved individual perceives the procedure followed to address the wrongdoing to be fair (Javornik et al., 2020). Distributive justice is the extent to which the aggrieved perceived a resolution to have ‘offset the loss’ or harm experienced (Javornik et al., 2020, p. 103). In Shao et al.’s (2022) model, adjusted as explained in the introduction to the ‘Literature review’ section, LAEs serve as an independent variable, while fairness remains a dependent variable. The following hypothesis has been formulated:

H1: There is a significant positive relationship between followers’ perception of leader apology elements and their perception of fairness.


Authentic leadership is defined by Kleynhans et al. (2021) as evidencing self-awareness (of strengths and developmental areas) and contextual awareness yet also self-regulation across contexts and as making unbiased and integrally moral decisions. Thus, it could be said that authentic leaders consistently evidence who they are (which includes a strong moral character) with no preference for persons or situations.

Similarly, authenticity can be defined as the ‘degree to which the apology is [perceived to be] sincere and genuine’ (Shao et al., 2022, p. 3434) or true to who the leader is. Bies et al. (2021) included sincerity in their definition of remorse, and the latter also included an apology. Mu and Bobocel (2019) found that transgressors’ fear of sanctions served as a motive for transgressors’ apologies. In the latter study, fear of sanctions referred to either fear of the responses of those who were aggrieved, or of organisational sanctions, in the absence of an apology. Shao et al.’s (2022) authenticity construct could be aligned with Mu and Bobocel’s (2019) fear of sanctions because it is possible that an apology perceived to have been rendered out of fear of sanctions such as disciplinary action or negative publicity would be perceived to be inauthentic, which would in turn impact on the perceived quality of leader apology:

H2: There is a significant positive relationship between followers’ perception of leader apology elements and their perception of authenticity.

Prior and post leader–follower relationship

In the present study, both prior and post LFR variables were operationalised as the quality of leader–member exchange (LMX) relationship. Lower-quality LMX relationships are those perceived by followers to be based on formal grounds such as a contract of employment. A high-quality LMX relationship is one in which the follower perceives the relationship to be based on ‘trust, support, and mutual understanding’ (Wang et al., 2021, p. 353). Leader–member exchange is therefore a measure of the quality of the LFR over time (Troster & Van Quaquebeke, 2021).

In the related field of integrity studies, Van Niekerk and May (2019) qualitatively explored the application of their integrity framework (Van Niekerk & May, 2012) that included the leader–follower exchange relationship as a variable. In their studies, they highlighted leaders’ responsibility to understand the impact they could have on the leader–follower exchange relationship. Specifically, leaders can impact perceptions of differences in integrity between leaders and followers and the resultant investment in time required to align integrity in the relationship on followers’ perceptions of values and norms entailed in integral leadership of others and on their feeling empowered to lead with integrity.

Post LFR (the relationship post the wrongdoing and apology) was selected as a dependent variable. Basford et al. (2014) found effective leaders’ apologies to be associated with higher quality LMX relationships.

H3: There is a significant positive relationship between followers’ perception of Leader apology elements and Post leader–follower relationship.

Prior LFR was selected to replace prior leader reputation as a moderating variable in the study. Collins et al. (2023) recommended that future research into crises that are internal to the organisation and intentional on behalf of leaders considers the impact of the prior LMX relationship quality on a follower’s response to a leader’s transgression.

Epitropaki et al. (2020) listed the quality of the LFR prior to the wrongdoing as a likely boundary condition to the relationship repair process. Radulovic et al. (2019) found that high-quality LMX relationships resulted in greater follower forgiveness and through the latter, to higher levels of follower job satisfaction and well-being. Olekalns et al. (2020) found that while a high-quality relationship (not necessarily limited to LMX) can be a positive factor in the subsequent repair of a damaged relationship, it could also be a hindrance: persons in the relationship feel a greater sense of betrayal following the damage to the relationship precisely because it was not expected in the context of the high-quality relationship. Thus, it is possible that higher quality LFRs could moderate the follower’s perception of the quality of the leader’s apology in either direction:

H4: The relationship between followers’ perception of leader apology elements and fairness is moderated by prior leader–follower relationship.

H5: The relationship between followers’ perception of leader apology elements and authenticity is moderated by prior leader–follower relationship.

H6: The relationship between followers’ perception of leader apology elements and post leader–follower relationship is moderated by prior leader–follower relationship.

Transgression type

Transgression type is determined firstly by whether the cause for the transgression was situated within or outside the leader’s control and secondly by whether the transgression was accidental or preventable (Shao et al., 2022). Shao et al.’s (2022) classification of transgression type overlaps with Coombs and Holladay’s (1996) crisis typology in which crises are classified along two dimensions: firstly, intentional-unintentional (on the part of leadership) and secondly, internal-external (to the organisation). In their interdisciplinary literature review of crisis leadership studies, Collins et al. (2023) found leaders’ apologies to be a common leader behaviour or style for crises that were both internal to the organisation and either intentional or unintentional on the part of the leader or leadership. A classification of transgression type along these dimensions draws on attribution theory – in other words, the type of transgression is categorised according to its cause (Martinko & Mackey, 2019).

Sharma et al. (2023) found that transgression-based factors such as the nature of the wrongdoing influenced the effectiveness of apologies. Grover et al. (2019) found transgression severity moderated the relationship between apology quality and forgiveness. Grover et al.’s (2019) definition of apology quality is aligned with Shao et al.’s (2022) apology elements.

H7: The relationship between followers’ perception of Leader apology elements and fairness is moderated by the Transgression type.

H8: The relationship between followers’ perception of Leader apology elements and authenticity is moderated by the Transgression type.

H9: The relationship between followers’ perception of Leader apology elements and Post leader–follower relationship is moderated by the transgression type.

Leader apology conceptual framework

In summary, LAEs serve as the independent variable in the study and are comprised of the content of the leader’s apology, its timing and delivery method. Three independent variables are utilised: fairness, authenticity and post LFR. The effects of two moderating variables are investigated: Prior LFR and Transgression type. The variables to be studied and hypotheses stated above are presented in a conceptual framework in Figure 2.

FIGURE 2: Conceptual framework for the study.

Research design

Research approach

A cross-sectional survey design was utilised as a change in followers’ perceptions of their leaders’ apologies over time was not being assessed. The research design was quantitative, consistent with the objective of examining causal relationships among variables.

Population and sampling strategy

The research population was comprised of all individuals who are working in an organisation. Non-probability sampling was used. The questionnaire was distributed via LinkedIn and WhatsApp, via SurveyCircle™ (2023) in particular and also SurveySwap™ (Ringle et al., 2022). SurveyCircleTM is a moderated platform with built-in checks to ensure participants do not complete too quickly. The questionnaire was also distributed to the researchers’ network, and via fellow students and colleagues to request volunteers to be part of the study. Volunteers were also requested to forward the same to their contacts.

Research participants

A total of 486 responses, the final sample comprised 311 respondents who responded ‘yes’ to the filtering question, ‘My leader (past or present) committed a wrongdoing or caused harm (could be major, or relatively minor), towards me’. South Africans made up 50.20% of the sample, with the remainder hailing from over 50 other nationalities. Women comprised most of the sample (61.10%). The sample consisted of 49.50% white people, 22.50% black people, 19.60% Asian people, 4.80% mixed race people and 3.50% Other. Most of the respondents were between 35 years and 44 years of age (34.40%), while 28.60% were aged 25 years–34, 17.00% between 45 years and 54 years of age, 15.10% between 18 years and 24 years of age and 4.80% were 55 years of age or older. Most (37.00%) were in middle or senior management, 21.90% were Specialists, 14.80% in Executive positions, 13.20% in Administrative positions and 6.40% in Other positions. Out of the total number, 73.60% held postgraduate qualifications, while 1.30% did not have a matriculation or school-leaving qualification. For at least half of the sample that is South African, this qualification demographic is very high and therefore potentially a skewed sub-sample.

Measuring instrument

To assess the independent variable, LAEs, as well as two of the dependent variables (fairness and authenticity), new items were constructed by the authors. The same five items from Radulovic et al.’s (2019) seven-item LMX scale were used to assess the remaining dependent variable, post LFR and the moderating variable, prior LFR, Radulovic et al.’s (2019) original scale spoke to a hypothetical scenario; therefore, items were edited such that they were no longer scenario specific. Items assessing prior LFRs were prefaced, ‘[before] the wrongdoing or harm I experienced’, while post LFRs were prefaced, ‘[following] my leader’s apology (or, in the absence of one, following the process post-wrongdoing)’. Again, for the second moderating variable, Transgression type, new items were constructed by the authors. Questions were edited to take into account feedback received from pilot study participants. The final questionnaire is provided in Table 6 at the end of this article. In addition to the filtering question mentioned under the ‘Research participants’ section, two more filtering questions were added. The first (Q2) checked whether or not participants were fulltime employees. However, this requirement was removed early on in the data collection phase to broaden the sample base, and while the question was included, a ‘yes’ was not required to continue through the questionnaire. The second (Q3) filtering question ascertained how long ago the wrongdoing occurred in order to account for possible effects of the passage of time on memory. Six questions (Questions a–f) were included to collect demographic information concerning the sample gathered (sex, nationality, qualification level, job level, age and ethnicity). This was done in order to provide a detailed description of the eventual sample while still maintaining respondent anonymity. The questionnaire was pretested by three people, followed by slight edits to enhance user-friendliness and logical flow.

Statistical analysis

Responses were exported to Microsoft® Excel® Version 16.77.1 and imported into both IBM® SPSS® Version and SmartPLS™ 4 (Ringle et al., 2022). Descriptive statistical techniques and exploratory factor analysis (EFA) were conducted, along with partial least squares structural equation modelling (PLS-SEM).

Distributional assumptions

While PLS-SEM is non-parametric and therefore does not assume normally distributed data, it is not unaffected by outliers (Hair et al., 2021). For all Likert-scale items, skewness ranged between -2 and 2, within the acceptable range defined by Hair et al. (2022). For most items, kurtosis was also within the acceptable range between -2 and 2 (Hair et al., 2021), except Q2 (5.94), a binary item not included in the PLS-SEM and omitted from the analysis, Q10, (3.06) and Nationality (-2.01), the latter a categorical variable also not included in the PLS-SEM.

Exploratory factor analysis

Following a principal component analysis and Oblimin rotation, a five-factor solution emerged. The suitability of the data was assessed using the Kaiser–Meyer–Olkin measure of sampling adequacy, which yielded a result of 0.922 (> 0.6) and Bartlett’s test of sphericity, yielding a chi-square of 5838.089 (p < 0.001). Factor 1 included most LAEs (no differential loadings among content, timing and delivery method), fairness (F) and one question from each of accountability and authenticity. These elements were retained as part of Factor 1. Prior LFR loaded on Factor 2, Post LFR on Factor 3, Transgression type on Factor 4 and the remaining authenticity items on Factor 5, these factors were therefore each named as per the elements that loaded on each. The five-factor solution explained 67.182% of the common variance. Albeit the eigenvalue of the first factor proved much higher than that of successive factors’, all eigenvalues exceeded 1 (see Table 1).

TABLE 1: Exploratory factor analysis results.

The following variables (and their corresponding items) were dropped from the measurement model as they were not reflected in the EFA (and were therefore not included in the ‘Literature review’ section), thereby enhancing the construct validity of the measurement model: Transgression criticality (Q12–Q15) and Format richness. Q21 was not included owing to low anti-image correlations. Q29 and Q30 were excluded as their bivariate correlations were not significant, and they did not load on the five-factor model. Q17 also did not load on any resultant factors. Q39 could not be included as it resulted in qualitative data.

All remaining items were loaded on one factor each (there was no evidence of item cross-loadings), with factor loadings in excess of 0.630, with the exception of Q32 with a loading of 0.503 on Factor 1. Item communalities exceeded 0.504, with the exception of Q22 with a commonality of 0.391.

Thus, while the initial measurement model included 14 variables and 15 hypotheses and sought to confirm an adapted version of Shao et al.’s (2022) theoretical model, the model was adapted to reflect the five-factor model revealed in the EFA, with the nine hypotheses outlined in the ‘Literature review’ section. Among the remaining hypotheses, H1, H4 and H7 were eliminated as Leader apology elements and fairness had been conflated into a single factor. The pattern matrix and Cronbach alpha coefficients are included in Online Appendix 1 and Online Appendix 2, respectively. As some hypotheses fell away, and LAEs and fairness were conflated as a result of the EFA, the hypotheses to be tested have been restated in Table 2.

TABLE 2: List of hypotheses to be tested.
Partial least squares structural equation modelling measurement model

All latent variables were reflectively measured. The latent variable correlation matrix is shown in Table 3. Measures of reliability and validity were compared with Hair et al.’s (2019) rules of thumb to assess the acceptability of the measurement model.

TABLE 3: Latent variable correlation matrix.

Most outer loadings exceeded 0.500, except Q22’s loading of 0.492. Q22 was retained, because loadings exceeded 0.700 per latent variable overall. Reliability was acceptable as Cronbach alpha coefficients (α) all exceeded 0.700, and composite reliability (rho_c) exceeded 0.700. It should, however, be noted that the Cronbach alpha coefficient of leader apology effectiveness and fairness exceeded 0.900, as did its composite reliability, indicating possible indicator redundancy, which could impact content validity. The average variance extracted (AVE) exceeded 0.550 except for Transgression type: 0.453, because of Q11 (0.539). Q11 was not removed from the model as doing so would have lowered the AVE. Overall, therefore convergent validity was deemed acceptable. Full reliability and descriptive statistics are provided in Table 4.

TABLE 4: Latent variable reliability and validity.

Overall, the discriminant validity of the measurement model was acceptable. The Fornell–Larcker criterion was good, except for the cross-correlations on post LFR and LAEs and fairness: ~ 0.300 among Qs 35–38 and 0.000 between Qs 32–33 and Qs 22–25. These items were retained, because the heterotrait-monotrait (HTMT) ratio values were < 0.850, below the 0.900 threshold value.

Partial least squares structural equation modelling structural model assessment

Collinearity was not evident, as all variance inflation factor (VIF) values were < 3. The endogenous variable, post LFR, had moderate variance explanatory power (0.585 unadjusted), while Authenticity’s explanatory power was weak (0.343 unadjusted) (Kock, 2015).

The PLS Predict analysis revealed positive Q2 predict scores; thus at an indicator level, the model had good predictive power overall. Q26 (RMSE 1.625 < MAE 1.320) and Q27 (RMSE 1.489 < MAE 1.187) were not strong predictors of authenticity, however. At a latent variable level, the Q2 predicted scores of authenticity (0.244) and post LFR (0.475) were positive; thus the inner model had good predictive power (Gaskin, 2022).

Chi-square values exceeded 5. The normed fit index (NFI) value was 0.802, lower than the requisite 0.900. Thus, model fit was achieved (SmartPLS GmbH, 2023). The final PLS-SEM structural model is shown in Figure 3.

FIGURE 3: Final partial least squares structural equation modelling showing R2 and b values.

Ethical considerations

Written ethical approval was obtained from the Ethics Committee of the University of Pretoria’s Gordon Institute of Business Science (GIBS), prior to questionnaire distribution. Questionnaire responses were gathered anonymously. The first page of the questionnaire consisted of an informed consent clause, and respondents were informed that their completion of the questionnaire would be taken to indicate that they granted their informed consent for their responses to be utilised.


Bootstrapping was conducted to establish levels of significance (Hair et al., 2019). Leader apology elements and fairness were positively related to authenticity (t = 7.787; p = 0.000), lending support to H2. H3 was also supported, as a positive relationship was found between LAEs and fairness and post LFR (t = 9.859; p = 0.000).

The moderating effect of prior LFR on the relationship between LAEs and fairness, on the one hand, and authenticity, on the other, was not significant at the p = 0.05 level (t = 1.843) and H5 is not supported. Prior LFR had a significant positive moderating effect on the relationship between LAEs and fairness and post LFR (t = 3.075; p = 0.002). H6 was thus supported. Transgression type was also found to positively moderate the relationships between LAEs and fairness and each of authenticity (t = 2.982; p = 0.003) and post LFR (t = 2.801; p = 0.005); thus H8 and H9 are supported.

The f-square values reveal large effect sizes for H2 (0.317) and H3 (0.554) and small effect sizes for H6, H8 and H9. Results are summarised against the hypotheses in Table 5.

TABLE 5: Latent variable reliability and validity.

A closer look at Table 4 reveals that the mean of post LFR is lower than that of Prior LFR. Using a dependent samples t-test, this difference proved significant (t = 16.611, p < 0.001). Thus, although there is a relationship between LAEs and fairness and post LFR, respondents’ perceptions of their relationship with their leader declined post wrongdoing and apology.


The purpose of the study was to empirically assess factors influencing followers’ perceptions of the effectiveness of their leaders’ apologies, following their leaders’ wrongdoings or transgressions.

Outline of the results

Because H2 was accepted, the quality of the content of the leader’s apology, its timing in relation to the wrongdoing, the channel via which it is communicated (public or private) and the degree to which the apology enhances the follower’s perceptions of distributive, procedural and interactional justice is likely to have a positive effect on to the extent to which the follower perceives the leader’s apology to be sincere and genuine in nature.

TABLE 6: Survey questions.

The quality of the contents, timing, channel of the leader’s apology and the extent to which it enhances perceptions of fairness are together likely to enhance the quality of the LFR after the apology has taken place, given that support was found for H3 in the current study. Because no support was found for H5, the quality of the LFR prior to both the leader’s wrongdoing and apology is unlikely to either dampen or enhance the relationship between LAEs including fairness, and the extent to which the apology is perceived to be authentic. Similarly, as H6 was not supported, the relationship prior to both the leader’s wrongdoing and apology is not likely to impact on the relationship between followers’ perception of LAEs and their relationship with their leader post-wrongdoing and apology.

The extent to which the follower perceives the wrongdoing to have been accidental and not preventable on the part of the leader enhances the relationship between the quality of the apology elements and the follower’s perception of the authenticity of the apology (H8). Also, the more leaders’ transgressions are viewed as accidental and not preventable by followers, the more positive the relationship between elements of leaders’ apologies and post LFR (H9).

Practical implications

The findings contribute both to leader apology andto LMX research. The results confirm Sharma et al.’s (2023) finding that the nature of leaders’ wrongdoing can influence the extent to which apologies are effective. Basford et al.’s (2014) finding that effective leaders’ apologies to be associated with higher quality LMX relationships are also confirmed by the present study.

Epitropaki et al. (2020) asserted that the quality of the LMX relationship could serve as a boundary condition for relationship repair, and Radulovic et al. (2019) found that high-quality LMX relationships could enhance relationship repair post-wrongdoing. In support of these authors’ studies, the current study found that the quality of the LMX relationship prior to the wrongdoing moderated the impact of the quality of the LAEs on the LMX relationship post-wrongdoing (and, if relevant, apology). Epitropaki et al.’s (2020) and Radulovic et al.’s (2019) findings are not fully supported by the current study; however, the quality of the LMX relationship prior to the wrongdoing was not found to moderate the impact of the apology elements on its perceived authenticity. While high-quality LMX relationships may serve as a boundary condition to relationship repair, they may not serve as one to followers’ perceptions of leader apology authenticity.

Managers and supervisors, in their capacity as leaders, are therefore cautioned to work on the quality of their apology in terms of its content, its promptness and the channel via which it is delivered. Managers should not assume that the strength of their relationship with their followers will enhance the extent to which followers are convinced of leaders’ apologies’ authenticity. Extra care should be taken should their followers perceive the leaders’ wrongdoing to have been preventable.

Leaders’ proactively working on the quality of their relationships with their followers is likely to protect (and possibly enhance) those relationships, if a leader commits a wrongdoing and has to apologise, provided the quality of the elements of the apology is perceived to be comprehensive, prompt and fair in nature. Again, should followers perceive a transgression to have been accidental or as not preventable on the part of the leader, this is likely to enhance the impact of the quality of the elements of the leader’s apology on the LFR post-apology.

Leaders should take the time to prepare their apologies to ensure that the wrongdoing is acknowledged, defined and explained at length revealing a deep understanding of its magnitude and harm. The apology should include an acceptance of full responsibility for the wrongdoing and detail as to remediation as well as a plan of action to build and maintain relationships with those affected by the wrongdoing. Leaders’ expression of empathy and remorse during the apology will assist in its effectiveness (Shao et al., 2022).

Limitations and recommendations

The present study opens avenues for future research. Firstly, while a large and diverse sample was sourced, South Africans were best represented, as were women, white people, those 35 years–44 years of age, middle to senior managers and those with post-graduate degrees. Future studies should assess the generalisability of results. Secondly, the discriminant validity of the measurement model compromised the extent to which the model could confirm the adaptation of Shao et al.’s (2022) model as initially intended. Future studies could refine the model.

Thirdly, while the research assumed Shao et al.’s (2022) definition of a leader apology as that which included the contents stipulated in these authors’ adapted model, the extent to which respondents agreed with this definition was not assessed. Future research should establish what elements are essential for a leader’s apology to be perceived as such by followers.

Fourthly, the variables such as LAE and fairness were conflated as a result of the EFA in the present study. It is recommended that the relationship between these two variables be explored in more depth in future studies.

Finally, 176 of the original 486 respondents stated their leader had not committed a transgression against them; thus respondents may assume a threshold above which an act is considered a wrongdoing. Future research should consider possible demographic and contextual differences that could contribute to respondents falling above or below such a threshold. Also, the significant decline found in post LFRs suggests followers’ perceptions of the nature of transgression could play a role in the post-wrongdoing outcome. Future studies, establishing the nature of the transgression and the factors that influence it, will inform situations in which apologies are necessary and impactful.


The aim of the study was to empirically assess factors influencing employees’ perception of leaders’ apology effectiveness. The elements of leaders’ apologies were found to have significant positive relationships with both LFRs post-apology and on perceived apology authenticity. The perceived preventability of wrongdoings moderated these relationships. The quality of the LFR was moderated by pre-apology, affecting only the relationship between LAEs and the post-apology LFRs. The findings contribute to the mapping of the research space of transgressions and apologies by organisational leaders, through empirical research. It is hoped that the results will assist leaders in effectively formulating and delivering apologies to followers.


This article was compiled in partial fulfilment of the requirements of the degree of Master of Business Administration. This article was compiled in partial fulfilment of the requirements of the degree of Master of Business Administration in the Gordon Institute of Business Science of the University of Pretoria, South Africa, awarded on 16 April 2024, with Prof. Gavin Price supervising. The URL was not available at the time of publication.

Competing interests

The authors declare that they have no financial or personal relationships that may have inappropriately influenced them in writing this article.

Authors’ contributions

C.C. conceptualised and designed the article, conducted and disseminated the results of the statistical analyses.

Funding information

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Data availability

The data that support the findings of this study are available from the corresponding author, G.P., upon reasonable request.


The views and opinions expressed in this article are those of the authors and are the product of professional research. It does not necessarily reflect the official policy or position of any affiliated institution, funder, agency or that of the publisher. The authors are responsible for this article’s results, findings and content.


Associated Press. (2016, January 10). Volkswagen CEO apologizes for emissions scandal [Video]. YouTube. Retrieved from https://www.youtube.com/watch?v=t7ne3cwqI4A

Basford, T.E., Offermann, L.R., & Behrend, T.S. (2014). Please accept my sincerest apologies: Examining follower reactions to leader apology. Journal of Business Ethics, 119(1), 99–117. https://doi.org/10.1007/s10551-012-1613-y

Bentley, J.M., & Ma, L. (2020). Testing perceptions of organizational apologies after a data breach crisis. Public Relations Review, 46, 101975. https://doi.org/10.1016/j.pubrev.2020.101975

Bies, R.J., Tripp, T.M., & Barclay, L.J. (2021). Second acts and second chances: The bumpy road to redemption. Journal of Management Inquiry, 30(4), 371–384. https://doi.org/10.1177/1056492620986858

Collins, M.D., Dasborough, M.T., Gregg, H.R., Xu, C., Deen, C.M., He, Y., & Restubog, S.L.D. (2023). Traversing the storm: An interdisciplinary review of crisis leadership. The Leadership Quarterly, 34, 101661. https://doi.org/10.1016/j.leaqua.2022.101661

Coombs, W.T., & Holladay, S.J. (1996). Communication and attributions in a crisis: An experimental study in crisis communication. Journal of Public Relations Research, 8(4), 279–295. https://doi.org/10.1207/s1532754xjprr0804_04

Epitropaki, O., Radulovic, A.B., Ete, Z., Thomas, G., & Martin, R. (2020). Leader-follower transgressions, relationship repair strategies and outcomes: A state-of-the-science review and a way forward. The Leadership Quarterly, 31, 101376. https://doi.org/10.1016/j.leaqua.2019.101376

Foley, S. (2023). PWC Australia’s culture attacked in tax leak scandal report. Financial Times. Retrieved from https://www.ft.com/content/95ea1144-f792-491e-a20c-12af9d9417b6

Gaskin, J. (2022, May 18). SmartPLS 4: PLS Predict [Video]. YouTube. Retrieved from https://www.youtube.com/watch?v=pd-2FboTiw4

Grover, S.L., Abid-Dupont, M.-A., Manville, C., & Hasel, M.C. (2019). Repairing broken trust between leaders and followers: How violation characteristics temper apologies. Journal of Business Ethics, 155, 853–870. https://doi.org/10.1007/s10551-017-3509-3

Hair, J.F., Hult, G.T.M., Ringle, C.M., & Sarstedt, M. (2022). A primer on partial least squares structural equation modeling (PLS-SEM) (3rd ed.). Sage.

Hair, J.F., Hult, G.T.M., Ringle, C.M., Sarstedt, M., Danks, N.P., & Ray, S. (2021). Partial least squares structural equation modeling (PLS-SEM) using R. A workbook. Springer.

Hair, J.F., Risher, J.J., Sarstedt, M., & Ringle, C.M. (2019). When to use and how to report the results of PLS-SEM. European Business Review, 31(1), 2–24. https://doi.org/10.1108/EBR-11-2018-0203

Hetrick, A.L., Blocker, L.D., Fairchild, J., & Hunter, S.T. (2021). To apologize or justify: Leader responses to task and relational mistakes. Basic and Applied Social Psychology, 43(1), 30–45. https://doi.org/10.1080/01973533.2020.1828083

Javornik, A., Filieri, R., & Gunmann, R. (2020). ‘Don’t forget that others are watching, too!’ The effect of conversational human voice and reply length on observers’ perceptions of complaint handling in social media. Journal of Interactive Marketing, 50, 100–119. https://doi.org/10.1016/j.intmar.2020.02.002

Kahkonen, K., Blomqvist, K., Gillespie, N., & Vanhala, M. (2021). Employee trust repair: A systematic review of 20 years of empirical research and future research directions. Journal of Business Research, 130, 98–109. https://doi.org/10.1016/j.jbusres.2021.03.019

Kleynhans, D.J., Heyns, M.M., & Stander, M.W. (2021). Authentic leadership and follower trust in the leader: The effect of precariousness. SA Journal of Industrial Psychology, 47, a1904. https://doi.org/10.4102/sajip.v47i0.1904

Kock, N. (2015). Common method bias in PLS-SEM: A full collinearity assessment approach. International Journal of e-Collaboration, 11(4), 1–10. https://doi.org/10.4018/ijec.2015100101

Laurence, S. (2022, August 16). ‘We know we have to say we’re sorry, but…’ Bain’s open letter of apology. BizNews. Retrieved from https://www.biznews.com/global-citizen/2022/08/16/bain-open-letter-apology

Martinko, M.J., & Mackey, J.D. (2019). Attribution theory: An introduction to the special edition. Journal of Organizational Behavior, 40, 523–527. https://doi.org/10.1002/job.2397

Mu, F., & Bobocel, D.R. (2019). Why did I say sorry? Apology motives and transgressor perceptions of reconciliation. Journal of Organizational Behavior, 40, 912–930. https://doi.org/10.1002/job.2376

Olekalns, M., Caza, B.B., & Vogus, T.J. (2020). Gradual drifts, abrupt shocks: From relational fractures to relational resilience. Academy of Management Annals, 14(1), 1–28. https://doi.org/10.5465/annals.2017.0111

Radulovic, A.B., Thomas, G., Epitropaki, O., & Legood, A. (2019). Forgiveness in leader-member exchange relationships: Mediating and moderating mechanisms. Journal of Occupational and Organizational Psychology, 92, 498–534. https://doi.org/10.1111/joop.12274

Ringle, C.M., Wende, S., & Becker, J.M. (2022). SmartPLS 4. Oststeinbek: SmartPLS. Retrieved from http://www.smartpls.com

Sharma, K., Schoorman, F.D., & Ballinger, G.A. (2023). How can it be made right again? A review of trust repair research. Journal of Management, 49(1), 363–399. https://doi.org/10.1177/01492063221089897

Shao, W., Moffett, J.W., Quach, S., Surachartkumtonkun, J., Thaicon, P., Weaven, S.K., & Palmatier, R.W. (2022). Toward a theory of corporate apology: Mechanisms, contingencies, and strategies. European Journal of Marketing, 56(12), 3418–3452. https://doi.org/10.1108/EJM-02-2021-0069

Sillars, J. (2022, May 25). Dieselgate scandal: More than 90,000VW drivers in England and Wales will get payouts. SkyNews. Retrieved from https://news.sky.com/story/dieselgate-scandal-vw-drivers-in-uk-class-action-win-193m-settlement-12621224

SmartPLS GmbH. (2023). Resources: Fit measures in SmartPLS. SmartPLS GmbH. Retrieved from https://www.smartpls.com/documentation/algorithms-and-techniques/model-fit/

South African Revenue Service. (2022, November 10). SARS’ public apology to its employees adversely affected by its action. SARS. Retrieved from https://www.sars.gov.za/media-release/sars-public-apology-to-employees-adversely-affected-by-its-action/

SurveyCircle. (2023). Research website SurveyCircle. Retrieved from https://www.surveycircle.com

SurveySwap. (2023). Research website SurveySwap. Retrieved from https://www.surveyswap.io

Sverdrup, T.E., & Stensaker, I.G. (2018). Restoring trust in the context of strategic change. Strategic Organization, 16(4), 401–428. https://doi.org/10.1177/1476127017739843

Troster, C., & Van Quaquebeke, N. (2021). When victims help their abusive supervisors: The role of LMX, self-blame, and guilt. Academy of Management Journal, 64(6), 1793–1815. https://doi.org/10.5465/amj.2019.0559

Van Niekerk, A., & May, M.S. (2012). Exploring how middle managers experience the impact of senior management on their integrity. South African Journal of Labour Relations, 36(2), 42–61.

Van Niekerk, A., & May, M.S. (2019). Co-constructing integrity: A conceptual framework. SA Journal of Industrial Psychology, 45, a1498. https://doi.org/10.4102/sajip.v45i0.1498

Wang, P., Dust, S., & Wang, Z. (2021). Leader sex and employee power distance orientation as boundary conditions of the relationship between lead humility and leader-member exchange. Human Performance, 34(5), 351–367. https://doi.org/10.1080/08959285.2021.1966632

Wang, X., Wen, X., Liu, Z., Jiang, Y., & Huai, M. (2023). Leader apology in the employee-organization relationship: The roles of subordinate power distance belief and leader competence. Tourism Management, 96, 104694. https://doi.org/10.1016/j.tourman.2022.104694

Zoodsma, M., & Schaafsma, J. (2022). Examining the ‘age of apology’: Insights from the political apology database. Journal of Peace Research, 59(3), 436–448. https://doi.org/10.1177/00223433211024696


Crossref Citations

1. Unveiling the toxic leadership culture in south African universities: authoritarian behaviour, cronyism and self-serving practices
Olaniyi J. Olabiyi, Marieta Du Plessis, Carel Jansen Van Vuuren
Frontiers in Education  vol: 9  year: 2024  
doi: 10.3389/feduc.2024.1446935