The South African apartheid regime denied black people their basic rights and freedom prior to 1994 (Clark & Worger, 2022; Ramphele, 2008). After 1994, the newly elected South African government, the African National Congress (ANC), set out efforts to redress the inequalities presented under the then apartheid system (Jackson et al., 2005). This involved devising and implementing the BBBEE strategy, policy (Iheduru, 2003) and Codes of Good Practice (Republic of South Africa, 2007). Broad-Based Black Economic Empowerment is the ‘economic empowerment of all black people including women, workers, youth, people with disabilities and people living in rural areas through diverse but integrated socioeconomic strategies’ (Republic of South Africa, 2014, p. 4).

Despite its intended purpose, BBBEE has raised some concerns. For instance, BBBEE has resulted in tender corruption, where unqualified companies with an unsatisfactory BBBEE score are awarded tenders (Pike et al., 2018). BBBEE may lead to unfair discrimination against non-beneficiaries on the grounds of ethnicity (Krüger, 2014a). Wehmhoerner (2015) expressed that BBBEE has not assisted in eliminating historical inequities. This assertation is supported by statistics extracted from the 20th Commission for Employment Equity Annual report 2019 and 2020 (Department of Labour, 2019 and 2020), which shows the persistent underrepresentation of black employees across gender groups in managerial roles.

For example, in 2017, African men accounted for 9.6% of top management roles (10% in 2018 and 9.9% in 2019). Similarly, African women accounted for 4.7% in 2017 (5.1% in 2018) and 5.4% in 2019. The minimum compliance target for black employees occupying top management roles (including board participation) is 50%, and for black female employees, the minimum compliance target is 25%. However, the statistics show that the target has yet to be achieved. Oosthuizen and Naidoo (2010) found that African women are perceived to have been afforded preferential treatment in the form of appointments, promotions, training and performance appraisals, far greater than any previously disadvantaged group, which is considered to be unfair by African men who participated in the study. The Organisational Justice Theory maintains that employees judge the extent to which outcomes (Colquitt et al., 2001; Moorman, 1991), procedures (Leventhal, 1980) and interpersonal interactions are fair in nature (Colquitt et al., 2001). Considering this theory, the awarding of tenders to companies that lack the competencies and that do not satisfy the criteria specified in the BBBEE scorecard would relate to unfair outcomes. Not adhering to standard tender policy processes in awarding tenders would relate to the practice of unfair procedures, while affording preferential treatment to African women over African men in the form of appointments, promotions, training, and performance appraisals (Oosthuizen & Naidoo, 2010) would relate to unfair interpersonal interactions. Based on the theory and these examples, it can be assumed that employees would not perceive the implementation of BBBEE as fair in nature. The above-mentioned literature demonstrates a prevalence of adverse experiences, which lead the authors to hypothesise that (H1) BBBEE is perceived as unfair in the workplace.

BBBEE compliance is measured against the following indicators: ownership, management control, skills development, enterprise and supplier development, and socioeconomic development (B-BBEE Commission, 2016). While this study acknowledges these objectives, the focus remains on comprehending how individuals perceive and interpret the BBBEE policy. Moreover, while recognising the importance of BBBEE in advancing economic participation of black South Africans, the literature reveals the absence of a single measure that adequately captures the fairness perceptions of BBBEE by South African workers. The majority of the past few studies have focused on the development of measures that assess the positive and negative impacts of BBBEE on SMEs (Van Niekerk, 2019), as well as the positive and negative effects of BBBEE on business operations and competitiveness (Krüger, 2014b).

More so, Van Niekerk’s (2019) measure focuses on the following dimensions: advantages of BEE for SMEs, advantages of BEE for employees of SMEs, disadvantages of efficiency and reliability for SMEs, economic disadvantages for SMEs and perceived fundamental disadvantages of BEE (Van Niekerk, 2019). No specific dimension is assessed in Krüger’s (2014b) measure. A major limitation of both these measures is the failure to assess the fairness perceptions of BBBEE. To this end, there is the absence of a single instrument that evaluates the fairness perceptions of BBBEE within the workplace. It is important to observe that none of these measures were used to inform the development of items of the current BBBEE measure.

The process involved in scale development usually includes the following:

Given the absence of a singular measure to assess workplace perceptions of BBBEE fairness and drawing from the literature that relates to scale development, the following objective was framed: To develop and validate the BBBEE Fairness Perception Questionnaire, consequently, the following hypothesis was formulated:

H2: The BBBEE Fairness Perception Questionnaire is valid and reliable.

To address the research objectives and hypotheses, an exploratory sequential mixed-method design was used, where a qualitative, inductive methodology was adopted. Semi-structured interviews were held with a convenience sample of eight participants, across South African organisations and sectors to gauge the perceived fairness of BBBEE. More than half of the sample were male (five), were African (six) and had obtained a bachelor’s degree (four). Almost half of the participants were employed in middle and junior management positions. The collection of data ceased with eight participants, as no new themes emerged from the data.

The qualitative quotes were converted into measurement items. If an item reflected experience, a perception related to the (un)fairness of BBBEE or understanding of BBBEE, it was considered for inclusion in the item pool. Similarities identified in the items were clustered together into themes. Braun and Clarke’s (2006) thematic analysis process was used to identify the following perceptions of BBBEE: empowerment (and lack thereof), window dressing, ineffective policy and unethical behaviour. These themes thus constituted the factors of the measurement tool (Creswell & Clark, 2018). Relatedly, a minimum of three items per factor was considered necessary for the measure to be regarded as valid and reliable (Fabrigar & Wegener, 2012).

Having a clear understanding of the construct guided the researchers in item writing. The researchers made certain that all items of the measure made reference to BBBEE and some degree of (un)fairness towards BBBEE. The initial item pool comprised 38 items to enhance reliability compared to a measure with fewer items after item deletion. Alongside this, a five-point Likert scale response format was chosen by which respondents could judge each item. This allowed for finer gradations and reduced the likelihood of variance error.

The items were subjected to content and face validity, where methodologists and subject matter experts, as well as laypersons, assessed and identified the items that should be accepted, modified or rejected. More specifically, the convenience sample (n = 7) consisted of two methodologists, two subject matter experts and three adults from the general population with knowledge of BBBEE. Among those participants, 71% were women and 29% were men, between the ages of 30–40 (57%) and 40–50 (43%) years, respectively. The sample consisted of 71% white participants and 29% African participants. Among the participants, 29% spoke Afrikaans, 29% spoke Sepedi and 42% spoke Setswana. Majority of the participants held a bachelor’s degree (42%), while others held a master’s degree (29%) and a doctorate qualification (29%). The participants were academically qualified (57%), occupied senior management (29%), and junior management roles (14%), in various industries (finance and insurance 14%, scientific and technical services 57% and public administration 29%).

The panel were informed about the overall purpose of the study and were requested by email and telephone to provide written feedback on item quality and the degree to which the items captured the overarching construct, item clarity and the necessity for item rewording, deletion or retention. More so, each panel member was requested to provide written feedback on each individual item and its relatedness to BBBEE, and to assess which items should be grouped together. This was carried out between the months of October and November 2020.

From the consultations with the panel members, it was recommended that 28 items of the 38 newly developed items be modified to facilitate better understanding on the part of the potential research participants. For example, ‘To what extent are you willing to listen attentively to the ideas developed by your black colleagues who are from designated groups’ was modified to ‘I am in support of listening attentively to the ideas developed by my colleagues who are black’. ‘To what extent are you willing to be empathetic about the personal challenges faced by your colleagues who are black’ was revised to ‘I am in support of being empathetic about the personal challenges faced by my colleagues who are black’. The panel members were then asked to review the modified items that were included in the measure, in a second round of evaluations. There were no deletions or changes made to the items, which constituted the final measurement that was administered to a sample of 300 participants.

Administering the measure to a sample of 300 participants would help to determine the dimensionality of the set of items, which can be achieved through factor analysis. Various recommendations exist regarding the suitable sample size required when conducting factor analysis. Both Gorsuch (1983) and Kline (1979) recommended that the adequate sample size for factor analysis be a minimum of 100. Comrey and Lee (1992) suggested a rough rating scale for acceptable sample sizes: 100 = poor, 200 = fair, 300 = good, 500 = very good, ≥1000 = excellent. Guadagnoli and Velicer (1988) supported this recommendation and proposed that the sample size should be at least 300–450.

Based on the recommendation made by Guadagnoli and Velicer (1988), a convenience sample of 300 participants took part in this study for the purposes of performing the factor analysis, more specifically, an exploratory factor analysis (EFA). This sampling method was chosen because it afforded the researcher the opportunity to gain access to research participants who were conveniently accessible (Wienclaw, 2019). Individuals needed to have worked in a South African organisation for at least 1 year in order to participate in the study, be proficient in the English language and voluntarily consent to partake in the research.

An online self-administered survey using Google Forms was created. The survey comprised an introduction, the aim of the study and the consent form. Individuals whom the researcher had a professional relationship with were informed of the research via email and were requested to partake in the study. The link to the survey was distributed on social networking sites (WhatsApp®, Instagram®, Facebook®). Several departmental executives from various organisations in South Africa were approached via email. The participant organisations were informed of the study by the researcher and permission for the study was granted. The researcher requested that the departmental executives disseminate the survey link among their colleagues and to their professional networks, whom they thought might be interested in partaking in the study.

The participants comprised 69% female and 30% male. Most of the participants were between 26 and 35 years. Most were African (80%), were English speaking (20%), held a master’s degree (24%), were academically qualified (44%) and were employed in the education industry (46%).

Participants completed a biographical questionnaire and the BBBEE Fairness Perception Questionnaire. The main purpose of administering the questionnaire to a sample of 300 participants was to validate the items of the measure. Each of the measures listed here is described in the following sections.

The questionnaire comprised two parts, namely parts 1 and 2. Part 1 of the questionnaire comprised sections A and B. Section A of the survey comprised a biographical information section. The biographical questionnaire was used to gather information pertaining to gender, age, race, home language, education, the job level of the participants and the industry of employment.

Section B comprised the BBBEE Fairness Perception Questionnaire. The instrument was developed to measure the respondents’ fairness perceptions of BBBEE. The measure consisted of 38 items rated on a five-point frequency measure ranging from 1 (strongly disagree) to 5 (strongly agree).

Statistical analyses, namely, descriptive statistics, reliability analysis and EFA, were carried out using the Statistical Package for the Social Sciences (SPSS) software version 25 (SPSS Inc, 2018). The first step involved examining the data for missing and unexpected values. Next, the researcher obtained the descriptive statistics for all factors of the BBBEE measure, where the mean, standard deviations, skewness and kurtosis were inspected to describe their plausibility. A cut-off point of >-2.00 and <2.00 was set for skewness to determine the normal distribution of the data, and in terms of kurtosis, Pyzdek and Keller (2003) suggested a cut-off score of >-4.00 and <4.00, while Curran et al. (1996) suggested a cut-off score of >±7, as data consisting of kurtosis values of >±7 have shown to be problematic and are considered to have elevated levels of kurtosis. Thus, the guideline proposed by Curran et al. (1996) was used to determine the normal peak of the data.

The next step in the analysis was to examine sampling adequacy. The Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy (MSA) and Bartlett’s test of sphericity were analysed to determine whether the sample size was appropriate for conducting an EFA (Pallant, 2011). A KMO value of 0.6 and above is recommended as a suitable indicator (Kaiser, 1960) and Bartlett’s test of sphericity should generate a significant statistic (p < 0.01) (Bartlett, 1954). Factor analysis is concerned with identifying the shared variance among variables and grouping variables that are highly correlated with one another into factors, while removing items that are not correlated with other items (Tabachnick & Fidell, 1989). Exploratory factor analysis is a multivariate technique that attempts to determine latent variables that parsimoniously explain the structure among observed manifest variables (Preacher et al., 2013). It has been instrumental in the design, development and validation of psychological measures (Williams et al., 2010).

To aid in the number of factors to keep, the eigenvalues-greater-than-one technique (Kaiser, 1960), the scree plot (Cattell, 1966) and parallel analysis scree plot (see Figure 2), as well as the parallel analysis outcomes (Horn, 1965), communality values and corrected item-total correlations were studied and used as criteria to retain items or for possible further item analysis. The objective of Kaiser’s criterion is that factors are retained when the eigenvalue is greater than the threshold of one (Kaiser, 1960). The parallel analysis technique is a Monte Carlo simulation method (Ledesma & Valero-Mora, 2007). The basic premise of parallel analysis is to compare the eigenvalues obtained from the actual data with the eigenvalues generated from the random data (Horn, 1965). The principle is that, when the ith eigenvalue from the real dataset is larger than the ith eigenvalue obtained from the random data, the factor is retained (Horn, 1965).

Reliability was determined by using the Cronbach’s alpha coefficient. Netemeyer et al. (2003) recommended a cut-off of α < 0.70. As an additional test for reliability, the inter-item correlations were calculated. Positive inter-item correlations that significantly (p < 0.05) correlate moderately or strongly with all other items in the measure are desired. This indicates that the items measure the same underlying characteristics. An additional measure that was employed to provide further information about the reliability of the measure was the corrected item-total correlations. In terms of the corrected item correlation, a coefficient value of 0.30 and above indicates that each individual item score correlates with the total score (Field, 2013).