THE PERCEPTIONS OF EMPLOYMENT EQUITY AND BLACK ECONOMIC EMPOWERMENT AS PREDICTORS OF UNION COMMITMENT

The purpose of the study was to test whether the perceptions of employment equity (EE) and black economic empowerment (BEE) are related to union commitment and whether the perceptions about the mentor’s role significantly mediate this said relationship. The sampling frame for the study constituted 1200 employees of a division of a large public transport organisation and a convenience sample including all 1200 employees yielded 637 fully completed records (a 53% response rate). The results of the study indicate that the perceptions of EE and BEE are significantly related to union commitment, but that perceptions of the mentor’s role do not mediate this relationship. More detailed findings on the study are reported. OPSOMMING Die doel van die studie was om vas te stel of die persepsies van werkgelykheid (EE) en Swart ekonomiese bemagtiging (BEE) verband hou met vakunieverbondenheid en of die persepsies omtrent die mentorrol hierdie verhouding beduidend medieer. Die steekproefraamwerk van die studie het bestaan uit 1200 werknemers uit ’n divisie van ’n groot vervoeronderneming en ’n gerieflikheidsteekproef wat al 1200 werknemers ingesluit het, het 637 volledige rekords opgelewer (’n 53% responskoers). Die resultate van die studie dui daarop dat die persepsies van EE en BEE beduidend verwant is aan vakunieverbondenheid, maar dat persepsies van die mentorrol nie hierdie verhouding medieer nie. Meer gedetaileerde bevindinge van die studie word gerapporteer. THE PERCEPTIONS OF EMPLOYMENT EQUITY AND BLACK ECONOMIC EMPOWERMENT AS PREDICTORS OF UNION COMMITMENT Requests for copies should be addressed to: G Roodt, Department of Human Resource Management, RAU University, PO Box 524, Auckland Park, 2006 55 SA Journal of Industrial Psychology, 2005, 31 (1), 55-64 SA Tydskrif vir Bedryfsielkunde, 2005, 31 (1), 55-64

that whites are to experience reverse discrimination, because black workers are given preference. Mahanyele (1993) reflects the frustration that organisations only talk about EE, but deny blacks the opportunity.
Gender Müller and Roodt (1998) found women to be more negative than men about affirmative action (EE), because they have fewer opportunities. Contrary to this, Van Zyl and  found women to be more positive than men about EE. This inconsistency is probably attributable to the differences in the country's development stages.
Tenure Tenure is strongly related to age. The reason for the group in their mid-career to be more negative about EE may be related to their career demands and the typical mid-career crisis as pointed out by Walbrugh and Roodt (2003).

Marital status
No research could be found to shed more light on this relationship.

Education levels
No research was found to shed more light on this relationship. Müller and Roodt (1998) found that English-and Afrikaansspeaking respondents were more negative about affirmative action (EE) than the group speaking African languages.

Job level
Several authors have commented on the excessive job mobility of black managers (Matuna, 1996;Primos, 1994;Sibanda, 1995). Innes (2001) and Roberts (1997) found that black managers were poached at high salaries that are higher than those paid by the market. Vallabh and Donald (2001) reported on the following perceptions of black managers: They place a high value on being trusted and recognised for their work, though their values are not recognised. They are placed in positions that do not meet their academic background. They do not have the opportunity to use their knowledge and skills.
They are second-guessed by their colleagues and supervisors.
The above research results indicate that people have different experiences and expectations regarding EE and BEE, which will have a definite impact on their union commitment. For the purpose of this study union commitment is defined as: 'a cognitive predisposition towards a particular focus (the union), insofar this focus has the potential to satisfy needs, realise values and achieve goals' (Roodt, 2004, p. 85).
With reference to the theoretical model ( Figure 1) the next section deals with the relationship between biographical variables and union commitment.

Biographical variables and union commitment
Race Fullagar (1986) found that there are no differences for black and white workers in South Africa with regard to their level of union commitment.
Gender Gordon et al. (1980) found women to be more committed to an union than men. Fukami and Larson (1984) found no relationship between union commitment and age.

Age
Tenure Fukami and Larson (1984) found no relationship between union commitment and tenure. However, Barling, Wade and Fullagar (1990) found that only union tenure was a significant predictor of union commitment an aspect not investigated in this study.

Marital status
No evidence could be found in the relevant literature on the relationship between different marital status categories and their levels of union commitment.
Educational level Fukami and Larson (1984) found no relationship between union commitment and educational level.

Home language
No evidence could be found in the relevant literature on the relationship between different home language groups and their levels of union commitment.

Job level
No evidence could be found in the relevant literature on the relationship between job level categories and their levels of union commitment.
In the theoretical model ( Figure 1) it is suggested that the relationship between EE, BEE and union commitment is mediated by perceptions of the mentor's role.
One context in which mentorship can work is within BEE. If one departs from the vantage point that mentoring is an equal relationship between the mentor and protégé, then entrepreneurial and managerial skills can be transferred while allowing the new manager or entrepreneur to develop the right competencies (Clutterbuck & Abbott, 2003).
According to Storm and Roodt (2002) and Wingrove (2002), white, middle-aged mentors may be cynical and negative about EE and this negativity may be projected to the protégé, which can again lead to lower commitment levels. Little empirical evidence could be found to support this rationale, except for the exploratory research by Storm and Roodt (2002) that focused on organisationrelated commitment.
Objectives of the study Against this background, the primary objective of this research will be to determine whether the perceptions of EE and BEE are related to union commitment.
The secondary objectives of the study will be: 1. to determine whether this relationship is mediated by the perceptions of the mentor's role; and 2. to determine which biographical variables are significant predictors of the perceptions of EE and BEE.
The objectives of the study can be inferred from the theoretical model (Figure1) compiled from the research literature.

Research approach
The research approach followed in the study is from the quantitative paradigm and the application of measuring instruments in a cross-sectional, non-random field survey generated the primary data for the study. An ex post facto approach to data analysis was used in exploring the interrelationships between variables in the data set.

Research methodology
The research methodology followed, is described according to the following three headings:

Sample of participants
The largest division of this transport organisation was targeted for the research. A convenience sample including all employees in a sampling frame of 1200 employees yielded 637 completed questionnaires. A response rate of 53% was obtained. Only fully completed records were used for the data analyses.
As indicated in Table 1, the majority of the respondents were Whites (62%) followed by Africans (24%) and Coloureds/Indians/Asians (13,5%). Most of the respondents were male (80,5%) and a smaller group (19,5%) female. The majority of the respondents belong to an union (70,5%). To no extent 1 2 3 4 5 To a very large extent The questionnaire consisted of seven sections: Section A and B -the background information of the respondent; Section C -18 item Organisation-related Commitment Questionnaire (Roodt, 1997) (not discussed for the purposes of this article); Section D -the 16 item Union Commitment questionnaire (Roodt, 1997); Section E -the 25 item Perceptions of Employment Equity Scale [a radically adapted version from Martins (1999) Employment Equity Questionnaire]; Section F -an 18 item Perceptions of Black Economic Empowerment scale; and Section G -The Perceptions on the Mentorship Role Scale -a 26 item questionnaire. The perceptions on EE, BEE and the role of the mentor scales were designed by the authors.
The questionnaires will be discussed in more detail next focussing on the rationale for inclusion in the study, the composition of the questionnaire as well as the reliability and validity of the instruments.
Union Commitment Questionnaire (Roodt, 1997) The Union Commitment Questionnaire of Roodt (1997) was used and adapted. The instrument was used to measure the degree of commitment to the union. This particular union commitment scale was applied successfully in a South African setting before. A common theoretical base will enable the comparison of organisation and union commitment scores. There were 16 items, which were put into a question format. Roodt's questionnaire has an acceptable Cronbach Alpha coefficient of 0,909 (Roodt, 1997). Factorial validity (a subdimension of construct validity) can be inferred from the single factor structure. Content and face validity can be inferred from the item wording.
The Perceptions of Employment Equity Questionnaire (adapted from Martins, 1999) A radically, adapted version of the Employment Equity Questionnaire based on Martins' (1999) theorisation was used for this st udy. Statements were converted into questions and most of the limitations in itemconstruction as identified by Smith and Roodt (2003); Van Zyl and  and Walbrugh and Roodt (2003) were addressed. The 41 statements that were used in the initial instrument were adapted into 25 items in question format. Martins' (1999) questionnaire has a Cronbach Alpha of 0,959 and it can be concluded that this questionnaire measures employment equity practices consistently and reliably. Construct validity is defined as the extent to which the instrument measures the theoretical contract that is designed to measure (Allen & Yen, 1979;Schepers, 1992). According to Smith and Roodt (2003), the Employment Equit y Questionnaire meets this requirement. According to Smith and Roodt (2003), the Employment Equity Questionnaire has face validity, because it seems that the instrument covers the practices as indicated in the Employment Equity Act. Walbrugh and Roodt (2003) also support the face validity of the instrument. There is also evidence of limited discriminant validit y (Smith & Roodt, 2003), but the discriminant validity is questioned by Van Zyl and  and Walbrugh and Roodt (2003).

The Perceptions of Black Economic Empowerment Questionnaire
The questionnaire developed by the authors consists of 18 items and was based on existing literature on black economic empowerment. No information regarding the reliability and validity (except face and content validity) of the questionnaire is available yet, but findings of this study will be reported under the results section.
The Perceptions of the Mentor's Role Questionnaire (adapted from Dreher & Ash, 1990) The questionnaire of Dreher and Ash (1990) was adapted for this study, because it supported the theory of Kram (1988) -nine roles of a mentor and it was a shorter version of Noe's (1988) questionnaire based on the same theory. Twenty-nine items in question format were included in this questionnaire. No information on the reliability is available on this instrument, but findings of this study will be reported under the results section. This instrument seems to have content and face validity based on the item content.

Research procedure
The measuring instrument was distributed via the organisation's intranet to all individuals in the division, excluding those on the lowest levels who were not able to read or write. A hard copy was also sent to employees via Human Resources Practitioners. All the ethical codes, e.g. control procedures, were adhered to. A letter of reminder was sent at a later stage to urge individuals to participate in the study. Individuals could respond anonymously and all the returned responses were treated with the utmost confidentiality. The identity of all the individuals was thus protected to ensure reliable responses.

RESULTS
The analysis of the data was conducted in two stages. The objective of the first stage was to establish whether all the variables were robust and reliable. In order to achieve this objective, all the measuring instruments were factor analysed separately on two levels. This procedure suggested by Schepers (1992) was followed to avoid the creation of artificial factors, resulting from differential item skewness.
In the first level factor analyses, all the items of each instrument were intercorrelated and anti-image correlations were executed to test the suitability of the matrices for factor analyses by referring to the Kaiser-Meyer-Olkin Measure of Sampling Adequacy and the Bartlett's Test of Sphericity. Eigenvalues were calculated on the unreduced item intercorrelation matrices and a number of factors were postulated according to the Kaiser's (1970) criterion of eigenvalues larger than unity.
In the second level factor analyses, sub-scores for each postulated factor were calculated and the same procedure as explained above was conducted on the sub-score intercorrelation matrices. Items of the obtained second level factors (scales) were subjected to an iterative item analysis to establish the internal consistency of each scale. Based on this procedure, the following reliabilities were obtained for each scale: The Union Commitment (UC) Scale of Roodt (1997)  The second stage of the analyses was focused on the coefficients of association (Cramer's V) between background variables and the intercorrelations of the different scales' total scores (respectively displayed in Table 2 and 3). From Table 2 it can be seen that race and home language are significantly associated as well as job level and level of education. Tenure and race and tenure and age are also significantly associated. It is highly unlikely that significantly related variables will display any interactive effects. The intercorrelation matrix of scale scores (Table 3) indicates that UC is significantly correlated with the perceptions of EE (0,364), BEE (0,409) and the mentor's role (0,210). EE and BEE are positively correlated (0,509). All the correlations are significant on the 0,01 level.   For the next stage in the analyses, a multi-factor analysis of variance (ANOVA) was applied to predict a particular criterion variable. A model modification process was followed whereby the most parsimonious model was extracted. Residuals were systematically investigated to establish any possible interaction effects with the predictor variables.
The following biographical variables were included in the final summary model of the multi-factor analysis of variances (ANOVA) for predicting the perceptions of EE, namely gender, race and mentor Y/N. No significant interaction effects were detected. These three variables explain 7,4% of EE's variance. See Figures 3-5 for the significant predictors.  As can be seen from Figure 5, employees who have a mentor are more positive in terms of their perceptions of employment equity than those without a mentor.
The following biographical variables were included in the final summary model of the multi-factor analysis of variances (ANOVA) for predicting the perceptions of BEE, namely race, home language, gender x race, age x race and age x home language. These variables explain 40% of BEE's variance. See Figures 6-10 for the significant predictors.    In applying univariate General Linear Modelling (GLM) (ANCOVA) with all biographical variables, the perceptions of employment equity (EE) and the perceptions of black economic empowerment (BEE) as well as the perceptions of the mentor's role (MR) as predictors of union commitment (UC), the following emerged from the model modification process: 30% of the variance was primarily explained by the perceptions of employment equity (EE), while secondary predictors were educational level, race, having a mentor or not and perceptions of employment equity (EE) in an interactive relationship with the perceptions of black economic empowerment (BEE). Educational level also has a direct relationship with union commitment.
The results of the final summary model are depicted in Table 4.
The statistical parameters of the final summary model are depicted in Table 5.  The results of the above model can be graphically illustrated in Figure 11.

Figure 11: A graphical representation of the predictors of Union Commitment
It should be noted that educational level acts as definite differentiator for union commitment levels. Within the three cohorts there are also differences with regard to racial groups and those who have a mentor or not.

DISCUSSION
The first and second level factor analyses of the different measuring instruments, followed by iterative item analyses, yielded scales with acceptable coefficient Alphas (ranging from 0,91 to 0,97), indicating that the variables were suitable for inclusion in the second phase of the data analysis.
The second phase of the data analyses yielded the following associations or effect sizes between background variables (displayed in Table 2): Race and home language -The Africans speak African languages, the Coloureds/Indians/Asians speak English and the majority Whites speak Afrikaans, which were expected from the culture of this organisation. Job level and education -The majority of managers have a higher level of education, which is expected. Tenure and race -The African employees have less years of service compared to the White employees with longer years of service. Tenure and age -The younger employees have less years of service, which is self-explanatory.
As a result of these strong associations, one would expect that there would be no interaction effects between these mentioned variables.
From the intercorrelations of the scales' total scores, the mentioned variables were all significantly correlated, ranging from 0,210 to 0,509 (displayed in Table 3). BEE and EE were positively correlated.
If EE was predicted with all biographical variables by using a multi-factor analysis of variance (ANOVA), the following background variables emerged as predictors in the final summary model, namely gender, race and mentor Y/N. These variables explain 7,4% of the variance in EE. If BEE was predicted with all biographical variables by using a multi-factor analysis of variance (ANOVA), the following background variables emerged as predictors in the final summary model, namely race, home language, gender x race, age x race and age x home language where 40% of the variance is explained. The findings of the above two multi-factor ANOVA's indicated the biographical variables which are predictors of EE and BEE. Hereby, the second secondary objective of the study is achieved.
If union commitment (UC) was predicted by using a univariate GLM (ANCOVA), the following variables emerged as predictors in the summary model, namely educational level (direct relationship with union commitment) and also educational level, race, having a mentor or not and perceptions of employment equity (EE) in an interactive relationship with perceptions of black economic empowerment (BEE) that explain 30% of the variance. No other research could shed light on this finding. It is also clear that the perceptions of the mentor's role do not mediate the relationship between EE, BEE and UC. This result does not correspond with the findings of Storm and Roodt (2002) To illustrate the above, see Figure 14. The findings on the prediction of union commitment suggest that the perceptions of EE and BEE are significant predictors of the criterion variable and that perceptions of the mentor's role do not mediate this relationship. These findings support the primary objective of the study as well as the first secondary objective.
This situation creates a unique challenge for the top management of unions to create a context that will address this dichotomy between lower and higher education levels, those who have a mentor or not in the organisation as well as the polarising perceptions of EE and BEE.
If the different models as displayed in Figures 12-14 are combined, the following empirical model as displayed in Figure  15 emerges from the data: Figure 15: An Integrated predictive model of union commitment The study has little nomothetic value, as it was limited to a single organisation. Further research needs to be conducted across different organisations in similar contexts to establish the external validity of the findings. The findings have some theoretical value as the perceptions of employment equity and the perceptions of black economic empowerment were for the first time included as predictors of union commitment. Furthermore, a more sophisticated, multi-variate General Linear Model (GLM) was used for the prediction of union commitment. Besides providing a parsimonious prediction model by isolating only predictors that were significantly explaining variance in the criterion, this model also illustrates possible significant interaction effects between the different predictor variables. In doing so the model serves as a good point of departure for understanding and explaining union commitment in a diverse workforce setting, where the perceptions of employment equity (EE) and the perceptions of black economic empowerment (BEE) are applied.