Employee turnover remains a concern for employers (Silverstone, 2009). Organisations continue to face the challenge of retaining the key skills and competencies required to achieve organisational objectives, particularly so in a very demanding economic climate and competitive talent market (Moore & Bussin, 2012; Munsamy & Venter, 2009; Nzukuma & Bussin, 2011). Bhattacharya and Mukherjee (2009) highlight the association between employee retention and reward. Employee reward is concerned with the reward of individuals as aligned to the value that they add to the organisation (Armstrong, 2002). Despite the introduction of a total reward approach (WorldatWork, 2008), effective reward remains a challenge for organisations.

Research has considered reward preferences from a range of standpoints in an attempt to better comprehend individual preferences. This has included consideration of the influence of demographic factors such as race, gender, age and marital status (Moore & Bussin, 2012; Nienaber, Bussin & Henn, 2011; Schlechter, Thompson & Bussin, 2015; Snelgar, Renard & Venter, 2013), personality type (Nienaber et al., 2011; Vandenberghe, St-Onge & Robineau, 2008), generational theory (Bunton & Brewer, 2012; Bussin & Van Rooy, 2014; Smit, Stanz & Bussin, 2015; Snelgar et al., 2013), industry (Bussin & Toerien, 2015) and the influence of national culture on reward preferences (Herkenhoff, 2000; Newman & Nollen, 1996; Schuler & Rogocsky, 1998; Westerman, Beekun, Daly & Vanka, 2009), led primarily by Hofstede’s (1980) research. Despite these studies, further clarity is still required on how organisations should devise appropriate reward strategies, especially when dealing with non-financial rewards (Armstrong & Stephens, 2005; Schlechter et al., 2015).

While the impact of national culture on reward preferences has been researched (Herkenhoff, 2000; Hofstede, 1980; Westerman et al., 2009), occupational culture offers a further level of culture, which impacts on employee behaviour informed by membership to an occupational group (Lachman, Nedd & Hinings, 1995). As with national culture, occupational culture too may impact on employee behaviour in terms of reward preferences; however, this has not been empirically tested and explored in the literature as yet.

The primary purpose of this study was to research reward preferences from an occupational culture perspective within the South African context. In line with this purpose, the main objective of the research was to propose and evaluate a theoretical model of the relationship between occupational culture dimensions and employee reward preferences. The core research problem relates to a gap in empirical research of the effect occupational culture has on reward preferences.

By assessing the extent to which occupational groups prefer the different types of reward offered in the workplace, organisations will be more informed and enabled to leverage benefit from the structuring of the different reward elements in accordance with these preferences, thereby aiding retention efforts. In addition, consideration of occupational groups from a time orientation perspective in particular may guide organisations to structure incentive plans in such a manner that the preference for short- or long-term incentives can be optimised to achieve the greatest return for this investment. Organisations too can consider the findings in the application of the principles relating to pay allocation and pay orientation.

The next section is a review of the literature, outlining research on the main constructs and integrating it with this research. This is followed by a description of the research design, findings, results and discussion.

While organisations display their own culture, further cultures (known as subcultures) are evident within the larger organisational system. Subcultures form along different lines, and occupation is proposed as one of the most pronounced differentiators in the organisation context (Trice, 1993). Culture pertains to shared values, beliefs and ideas. Occupational culture theory holds that shared values, beliefs, ideas and orientations are based on membership in a specific profession or occupation group (Ames, Duke, Moore & Cunradi, 2009; Sheer & Chen, 2003; Trice, 1993). From an organisational perspective, values, beliefs and orientations influence the behaviour of employees (Herkenhoff, 2004). By nature, shared values and beliefs may therefore impact the behaviour of occupational groups. This impact may be evident in aspects such as reward preferences. Viitanen (2000) highlights that contributing to the formation of occupational cultures are the rights, privileges and obligations that employees have, based on their occupational group. Characteristics of occupational cultures include factors such as unique training and development, particular duties, control over certain functions and tasks, possessing specific competencies, knowledge and skills and obtaining specific qualifications and membership in professional and occupational bodies and associations (O’Hara-Devereaux & Johansen, 1994; Schein, 2004; Trice, 1993).

Hofstede’s (1980) national culture dimensions assist in the study of occupational culture (Ulijn, Nagel & Liang, 2001). Herkenhoff (2009) concurs and has highlighted that by measuring occupational culture against relevant occupational culture dimensions, it is possible for researchers to assess relative differences across occupational groups. Herkenhoff (2009) proposed five occupational culture dimensions, namely Power, Risk, Gender, Time and Team.

The international professional reward association, WorldatWork, advocates a total reward approach to employee reward. Total reward can be defined as ‘the monetary and non-monetary return provided to employees in exchange for their time, talents, efforts, and results’ (WorldatWork, 2007, p. 4). This approach considers the employee holistically in terms of personal and professional needs. In consideration with the range of employee needs, total reward encompasses rewards that are financially and non-financially based. Financial reward, for example, pay- and monetary-based incentives, speaks to the individual’s financial needs. Non-financial reward encompasses aspects such as opportunities for learning and growth, career prospects and management, organisational values, performance and recognition, relationships at work, job security and work–life balance (Armstrong, 2002; Thompson & McHugh, 2002; WorldatWork, 2008).

The research design was exploratory and explanatory. The research was quantitative and a structural equation modelling (SEM) approach was selected in order to test the hypotheses and to explore the proposed relationships between the latent and observed research variables (Nye & Drasgow, 2011).

Participants and sample: A South African Information, Communication, and Technology (ICT) organisation was selected for the study purpose considering its 20 420 employees who are employed on a permanent basis. All employees up to and including first-line management who are South African nationals were considered. The population was divided into occupational groups and sorted according to employee reference numbers. Every tenth employee per occupational group was selected for the sample. Some occupational groups consisted of small representations (e.g. Accountants), and these groups were excluded. The final sample included the following occupational groups: Manager, Information Technology, Technician, Clerical or Admin and Sales.

The output of the sampling was a study sample of 10 581. Data collection resulted in 1362 usable questionnaires, which represents a 12.9% response rate (the study organisation typically obtains a 10% response rate to surveys; the response rate was therefore aligned to the norm). Table 1 provides a profile of the respondents. All race groups participated in the study, with the bulk of respondents being white people (48.3%), male gender (70.1%) and from the 40- to 49-year age group (41.3%). The Technician occupational group had the largest representation (44.4%). The age, gender and occupational group distribution of respondents was representative of the target population. In terms of race, the white group was over-represented, and conversely, the African group was under-represented.

Measuring instrument: A web-based survey was utilised in the study. The first part of the survey related to biographical and demographic data collection. The second part, which made use of a five-point Likert-type rating scale, contained the independent and dependent variable measurement scales. The following descriptors were used: 1 = strongly disagree, 2 = disagree, 3 = neither disagree nor agree, 4 = agree and 5 = strongly agree.

Descriptive statistics for the Occupational culture questionnaire (PC10) are included in Table 4. The highest mean score (4.50) was reported for Item A19 on the Environment subscale (‘I make decisions in my job without clearly defined guidelines’). Item A11 (‘I would rather work less hours and get less pay’) displayed the lowest mean score (2.79). The majority of the items displayed a negative skewness. This represented a positive response pattern (Pallant, 2007). Robust maximum likelihood (RML) was utilised in the further analyses to cater for skewness in the data.

SARM10 descriptive statistics are shown in Table 5. The highest mean score (4.51) was reported for Item D7 on the Performance and recognition subscale (‘Being recognised for my skills and abilities is important to me’). Item B26 (‘I would rather work less hours and get less pay’) reported the lowest mean score (1.93); this item was in the Work hours subscale. The majority of the items displayed a negative skewness representing a positive response pattern (Pallant, 2007). RML was utilised in the further analyses to cater for skewness in the data.

In line with the study purpose, analyses focused on testing the proposed SEM theoretical model. The first step consisted of assessing the model fit with specific consideration to the standardised residuals and the ‘goodness of fit’ statistical analysis. In terms of the proposed theoretical model, 1301 statistically significant standardised residuals with an absolute value greater than 2.58 (p < 0.01) were detected in the reproduced residual matrix. The output of the goodness of fit analysis is reflected in Table 6. As reflected, the root mean square error of approximation = 0.040, p = 1.000, 90% confidence interval [0.039, 0.040] and was indicative of good fit. In addition, the standardised root-mean-square residual value of 0.069 provided support for good model fit. However, lower-than-desired comparative fit index and Tucker–Lewis index values of 0.777 and 0.768 were found; therefore, only partial empirical support was reported for the fit of the proposed theoretical model.

Focus was given to the Gamma parameters or regression lines in considering the model parameter estimates of the proposed theoretical model. Of particular interest were the size, direction and statistical significance of the relationships. Table 7 presents the standardised and unstandardised Gamma values. Nine of the hypothesised regressions showed statistically significant relationships, with the majority of these values displaying statistical significance at the p < 0.01 level. All the statistically significant relationships were positive, with the exception of H₁₃ (Service and Family-related benefits) and H₁₅ (Time and Short-term incentives), which reported standardised Gamma values of -0.168 (p < 0.01) and -0.089 (p < 0.05), respectively. While a negative relationship was hypothesised between Time and Short-term incentives (H₁₅), a positive relationship was hypothesised between Service and Family-related benefits (H₁₃). The direction of the relationship is therefore unexpected.

Model R-square values or commonalities were considered, given that in some cases it was found that dependent variables in the proposed theoretical model had more than one independent variable contributing to the reported variance. Cohen’s (1988) guidelines assisted in interpreting the effect size, whereby r = 0.10 reflected a small effect size, r = 0.30 a medium effect size and r = 0.50 a large effect size. The Model R-square values for the proposed theoretical model are reflected in Table 8. As can be seen, results revealed statistically significant model R-square values in that Power and Environment together explain 11.2% variance in Performance and recognition. Power and Environment combined explain 7.9% variance in Career and development. The Team occupational culture subscale explains 6.3% of the variance in Team pay, and Environment and Service jointly explain 5.8% variance in Family-related benefits. Work Hours, Environment and Service collectively explain 4.3% of the variance therein. Relatively small effect sizes were found.

Model R-square values were not statistically significant for the Guaranteed package, Short-term incentives, Pension, Job-level–based benefits, Job security, and Team equal reward subscales. Measurement issues were reported for the dependent variable measurement model relating to the Guaranteed package, Short-term incentives, Pension, Job-level–based benefits and Team equal subscales. Issues around model fit were also cited for these particular subscales. The lack of statistical significance of these R-square values could therefore be measurement related. With regards to job security, Chiang (2005) suggests that a preference in this regard is typically influenced by broader external or contextual factors rather than culture values, and this may assist in understanding the lack of statistical significance with regard to the Job security reward element at a model level. To this end, environmental elements such as the economic position as well as values and practices in the organisation, rather than a sole focus on cultural issues, could have an influence on the job security reward element.

Hypotheses H₁–H₁₇ were formulated relating to the proposed relationships between the occupational culture dimensions and reward preferences. As proposed by Pallant (2007), in reviewing the hypothesised relationships, the practical and statistical significance of relationships was considered. In terms of practical significance, of particular concern is the extent of the correlation and direction of the relationship. Guidelines proposed by Cohen (1988) assisted in guiding the interpretation of the size of the correlation (r), whereby r = 0.10 indicated a small effect size, r = 0.30 indicated a medium effect size and r = 0.50 and greater indicated a large effect size. Statistical significance considers the probability that the relationship exists and a cut-off of p < 0.05 was applied. R-square values provide insight into effect sizes. R-square values and commonalities were calculated and aided in the hypotheses testing.

Based on the output of the statistical analyses, 7 of the 17 formulated hypotheses were empirically supported. This pertained to the relationships between the Environment occupational culture dimension and Performance and recognition (H₈), Career and development (H₉), Work hours (H₁₀) and Family-related benefits (H₁₁), as well as the occupational culture dimension Time and Short-term incentives (H₁₅) and the Team occupational culture dimension and Team pay (H₁₆) and Team equal (H₁₇).

Very significant positive relationships with a small effect size were evident for Environment and Performance and recognition (H₈), Environment and Career and development (H₉), Environment and Work hours (H₁₀) and Environment and Family-related benefits (H₁₁). The variances explained in Performance and Recognition, Career and development, Work hours and Family-related benefits were 11.2%, 7.9%, 4.3% and 5.8%, respectively. A significant negative relationship between Time and Short-term incentives (hypothesis H₁₅) was evident with a small effect size. A 1.2% variance was explained in Short-term incentives. A very significant positive relationship was identified for hypothesis H₁₆, namely the relationship between Team and Team pay. A small effect size was evident, and 6.3% variance was explained in Team pay. A significant positive relationship between Team and Team equal (hypothesis H₁₇) was found. A very small effect size was evident and 0.4% variance was explained in Team equal.

Although statistically significant relationships were found for hypotheses H₆ (Risk and Short-term incentives) and H₁₃ (Service and Family-related benefits), the direction of these relationships was not as predicted, and as such, these hypotheses could not be empirically proven. The remaining proposed hypotheses did not achieve the required level of statistical significance and could therefore not be empirically validated.

Findings of the research suggest that evaluating the extent to which an occupational group places the needs of the team above those of the individual may assist organisations in considering a team-based approach to reward. It is probable that those groups that are more concerned about team needs will demonstrate a greater preference for a team-based reward approach. This may assist the organisation in decision making regarding pay practices, that is, in opting for individual versus team-based rewards. Furthermore, where risk-based incentives are incorporated as part of the overall reward strategy, the consideration should be given to whether employees are short- or long-term focused. This will assist in aiding maximum impact of adopting an incentive-based reward strategy. In this way, the study provides practical input to organisations and reward professionals in the design of reward strategies and programmes.

Adopting a completely individualised approach to reward is problematic for organisations administratively (Nienaber & Bussin, 2011). A group approach to reward may be more feasible. In practically implementing such an approach, organisations could consider a fixed basic pay and benefits structure with provision for flexible benefits, which are occupational group and reward preference-based (e.g. this may include the option for short- and/or long-term incentives for the sales occupational group). Based on the findings, this approach may be of particular value to organisations in the implementation of non-financial rewards elements such as Career growth and development, work-life balance and Performance and recognition. This would ease the administrative burden for the organisation while leveraging reward benefit from a retention perspective. Given that turnover levels tend to be higher for ICT staff versus general staff (P-E Corporate Survey, September 2010) and that attracting and retaining knowledge workers remains a challenge (Bussin & Toerien, 2015; Schlechter et al., 2015), the findings may be of particular relevance for organisations operating in the ICT space.

As previously highlighted, the influence of occupational culture extends further than its influence on reward preferences and resultant reward practices. A consideration of the influence of occupational culture may assist organisations in better understanding not only individual behaviour but also group and organisation behaviour. From a human resource management perspective, an enhanced understanding of behaviour at these three levels may assist organisations in optimising efforts, enhancing performer–manager relationships and creating a conducive and motivating work environment.

The research was conducted in one country and one company in order to contain the effects of national and organisational culture. By nature of this research approach, generalisation of the findings has to be arrived at with caution. Despite a sizable sample and adopting a stratified strategy on sampling, the over-representation of the white race group needs to be noted in the interpretation of the final results. Future research should include replication studies as well as broader populations in order to aid the generalisability of the findings. Concerns around the psychometric properties of certain subscales of the measurement tool were acknowledged, and this may have impacted the results incorporating the Job-level–based benefits and Work hours as well as the Power, Service, and Risk dimensions. Follow-up studies should focus on a further refinement of these measures.

With workforces becoming more diverse and globalised, the status quo of one-size-for-all cannot possibly last. Reward preferences exist, and much more research is required to better understand these preferences. We know already that reward preferences exist by gender, race, Myers Briggs profile, life cycle, and now we know that it also exists within different occupational groups. Organisations now need to find a way to implement reward choices to ensure optimal retention of staff. This will make them more productive, efficient and profitable.