The effective delivery of human resource management (HRM) services is often associated with positive organisational outcomes, including innovation. Within the context of HRM service delivery, as well as within the scope of innovative behaviour, gender differences are often researched.
To effectively research the role of HRM services in organisations, including the effects thereof on innovation, instruments that yield valid and invariant measures for men and women are required.
To date no measurement invariance research on the Human Resource Practices Scale (HRPS), with reference to gender, could be located. Researchers and practitioners alike should be hesitant to use the HRPS without such information.
A cross-sectional quantitative survey design was used. The present study addressed the measurement invariance for the HRPS across men and women, applying pairwise multigroup confirmatory factor analyses with robust maximum likelihood estimation to examine four levels of measurement invariance across the groups. Data from 2936 employees, representing 52 South African organisations, were used.
Results support the construct validity of the HRPS and demonstrate strict measurement invariance for the HRPS across gender, which implies that the HRPS yields scores with equivalent meaning, measurement units and measurement precision for men and women.
It will therefore be possible to test hypotheses regarding mean differences between men and women as well the relationship between the effective delivery of HRM services and positive organisational outcomes without fear that the HRPS will yield gender-biased results.
The research demonstrates that the items of the HRPS are valid for both men and women, suggesting that men and women have similar experience of the workplace. This finding should advance debate and research regarding the segregated delivery of HRM services and gender matters in general.
Gender (used in this text to refer to men and women) is a prominent variable within the workplace and life in general. Several journals are dedicated to the topic (see
At a societal level, the
Despite the aforementioned concerns, Dickens (
Focusing on human capital models, Lips (
This research aims to analyse the validity of a measurement of HRM practices across men and women, testing if respondents interpret the measure in a conceptually similar manner. Stated more operationally, the research aims to test whether the relationships between manifest indicator variables (scale items, subscales) and the underlying construct are the same across groups (Bialosiewicz, Murphy, & Berry,
Two matters are reviewed. Firstly, the contention that HRM practices constitute an antecedent to organisational outcomes is considered, and secondly the focus will be on ways in which HRM practices are measured. This review grounds the present research within the context of the present body of knowledge.
HRM practices can positively influence employees’ attitudes and lift workplace performance, which will most likely affect organisational outcomes (Kehoe & Wright,
The outcomes associated with effective HRM practices are not limited to organisational performance as a singular concept. Links have also been found with flow (Kasa & Hassan,
When considering quantitative methodologies, the measurement of constructs is important. Focusing specifically on the measurement of high-performance or effective HRM practices, some authors develop their own measures (e.g. Madmoli,
Some researchers prefer to present measurement of high-performance or effective HRM practices as a single construct (e.g. Makongoso, Gichira, & Orwa,
It may be important to note that the evaluation of HRM practices depends on the degree to which employees experience HRM practices as effective (Kehoe & Wright,
This research did not attempt to explain differences between men and women through identifying the most potent explanatory variables. Rather, it focused on the validity of the explanatory variables themselves, as Lips (
This study examines the HRPS structure across 1652 men and 1284 women employees of 52 companies in South Africa. Full data were available across all of the companies concerned. All applicants completed the HRPS in English (which is the lingua franca of high school and post-school education, as well as of business, in South Africa). The objectives of the study were (1) to examine if the HRPS structure could be replicated across gender groups, (2) to examine the level of measurement invariance attained across the groups and (3) to report on the psychometric properties of the HRPS when used in South African organisations.
The matter of measurement invariance is central to this research and to this article. Measurement invariance relates to an observed score being reflective of an individual’s standing on a construct, independent of his or her group membership (Mellenbergh,
The target population consisted of employees, at different levels of responsibility, who are exposed to various HRM practices. Organisations with more than 50 employees were targeted as it was presumed that the HRM services would be formalised in these organisations and that a broad range of services would be available.
The HRPS (Nyawose,
Participants responded to the items on a five-point Likert scale, ranging from ‘disagree strongly’ (1) to ‘agree strongly’ (5). For each of the seven HRM practices, the scores ranged from 3 to 15. A high score would be reflective of an individual who perceived the HRM practice as effective, whereas a low score would reflect that the participant was dissatisfied with the particular HRM practice. Nyawose (
The participants were 2936 employees (44.7% women), representing several public and private organisations based in South Africa. The distribution of participants with respect to race and ethnicity was approximately as follows: 8% Asian, 58% black people, 8% mixed ethnicity and 24% white people. The participants’ ages ranged between 20 and 72 years, with a mean of 37.8 years and with a standard deviation of 9.1. Participants’ tenure at their present companies ranged from 1 month to 42 years, with an average of just more than 9 years and a standard deviation of 7.5 years.
The data were initially scanned for normality, after which measurement invariance was tested for. Following the recommendations of Vandenberg and Lance (
The analysis only focused on measurement differences between self-identified men and women. This divide (mainly) represents the biological sex and more traditional gender role identification prevalent in the South African society. It is acknowledged that in the present era gender identification is more fluid and that identification as a lesbian, gay, bisexual or transgender (LGBT) individual may have more negative consequences (Badgett, Lau, Sears, & Ho,
The analyses were performed with the lavaan package (Rosseel,
Although highly desirable, it was expected that the hypotheses of perfect fit for the measurement models would be rejected, given that the χ2 statistic is very sensitive to sample size (in this case more than 3000) and is no longer relied upon as a basis for acceptance or rejection of a model fit (Schermelleh-Engel, Moosbrugger, & Müller,
A CFI > 0.95 is used as indicative of a good model fit (Vandenberg & Lance,
Vandenberg and Lance (
The BIC was used as a measure of comparative fit. Models that generate lower BIC values are generally preferred, and the absolute value was not interpreted. BIC was therefore used to assess model deterioration, which was visible when BIC values increase.
These parameters were used when interpreting the measurement invariance results. Once measurement invariance is established, more descriptive statistics on the HRPS will be provided. These will include the factor loadings, descriptive statistics, including reliability information, as well as the correlations between the observed scores as well as the latent factors. Last-mentioned will provide insight into the uni- or multidimensionality of the measurement of HRM practices.
Permission (2014_SBL_018_CA dated 27 February 2014) to conduct the research was obtained from the Research Ethics Review Committee of the Graduate School of Business Leadership at the University of South Africa before commencing with sampling. Once approval had been obtained, a list of staff members was requested from the organisation’s HRM department. Respondents were selected randomly from this list. The selected respondents were invited to a meeting at which the purpose of the research was explained. They were informed as to the nature of their participation, including that participation was completely voluntary. Those who agreed to participate then completed a consent form specifying ethical issues, including confirmation regarding the anonymity of participation, confidentiality, the right to withdraw from participation at any time without any explanation or any adverse effects, and the fact that the data would be used for research purposes only. Then only did they complete a hard copy of the questionnaire.
Preliminary analysis showed that the skewness and kurtosis of the HRPS items ranged from -0.08 to -0.97 and -0.99 to 0.79, respectively. None of the items demonstrated excessive deviation from normality and they appeared appropriate for factor analysis with robust maximum likelihood estimation (cf. Loehlin & Beaujean,
In each group, a baseline independent cluster confirmatory factor analysis model was specified in accordance with the structure given in
Chi-square test and change in chi-square statistics.
Invariance level | BIC | ∆ |
∆ |
∆ |
||
---|---|---|---|---|---|---|
Configural | 336 | 160 436 | 1341 | - | - | - |
Weak (loadings) | 350 | 160 343 | 1359 | 18.5 | 14 | 0.1867 |
Strong (intercepts) | 364 | 160 243 | 1371 | 12.0 | 14 | 0.6095 |
Strict (residuals) | 385 | 160 115 | 1411 | 39.5 | 21 | 0.0085 |
Equal latent means | 392 | 160 060 | 1411 | 0.7 | 7 | 0.9984 |
,
BIC, Bayesian information criterion.
As expected, the hypothesis of perfect fit for the configural invariance model was rejected (χ2(326) = 1341,
Fit measures and changes in fit measures.
Invariance level | CFI | RMSEA | ∆CFI | ∆RMSEA |
---|---|---|---|---|
Configural | 0.97 | 0.045 | - | - |
Weak (loadings) | 0.97 | 0.044 | 0.000 | 0.001 |
Strong (intercepts) | 0.97 | 0.043 | 0.000 | 0.001 |
Strict (residuals) | 0.97 | 0.043 | 0.001 | 0.001 |
Equal latent means | 0.97 | 0.042 | 0.000 | 0.001 |
CFI, comparative fit index; RMSEA, root-mean-square error of approximation.
As a final step, the constraint of equal latent means across men and women was added, producing a statistically non-significant Δχ2 (
Against the background of the support yielded by the ΔCFI and ΔRMSEA for strict measurement invariance,
Standardised factor loadings of the Human Resource Practices Scale items for men and women jointly.
Item | Factor |
||||||
---|---|---|---|---|---|---|---|
T&D | Rem | PM | SS | Sta | Div | Comm | |
1 | 0.85 | - | - | - | - | - | - |
2 | 0.84 | - | - | - | - | - | - |
3 | 0.73 | - | - | - | - | - | - |
4 | - | 0.64 | - | - | - | - | - |
5 | - | 0.89 | - | - | - | - | - |
6 | - | 0.84 | - | - | - | - | - |
7 | - | - | 0.80 | - | - | - | - |
8 | - | - | 0.79 | - | - | - | - |
9 | - | - | 0.61 | - | - | - | - |
10 | - | - | - | 0.70 | - | - | - |
11 | - | - | - | 0.88 | - | - | - |
12 | - | - | - | 0.82 | - | - | - |
13 | - | - | - | - | 0.77 | ||
14 | - | - | - | - | 0.54 | - | - |
15 | - | - | - | - | 0.70 | - | - |
16 | - | - | - | - | - | 0.75 | - |
17 | - | - | - | - | - | 0.77 | - |
18 | - | - | - | - | - | 0.57 | - |
19 | - | - | - | - | - | - | 0.74 |
20 | - | - | - | - | - | - | 0.85 |
21 | - | - | - | - | - | - | 0.80 |
Note: All factor loadings are statistically significant (
T&D, training and development; Rem, remuneration; PM, performance management; SS, supervisor support; Sta, staffing; Div, diversity management; Com, communication.
Noting that latent means were assessed to be invariant, descriptive statistics on the observed HRPS construct scores for men and women and reliability coefficients are presented in
Scale means, standard deviations and reliability coefficients on the Human Resource Practices Scale per gender.
Variable | Men |
Women |
||||
---|---|---|---|---|---|---|
Mean | SD | Cronbach |
Mean | SD | Cronbach |
|
T&D | 11.35 | 2.99 | 0.845 | 11.31 | 2.95 | 0.853 |
Rem | 9.03 | 3.09 | 0.834 | 8.96 | 3.15 | 0.852 |
PM | 9.99 | 2.79 | 0.787 | 9.96 | 2.76 | 0.784 |
SS | 10.58 | 2.89 | 0.835 | 10.52 | 2.97 | 0.853 |
Sta | 10.12 | 2.70 | 0.735 | 10.07 | 2.62 | 0.710 |
Div | 10.18 | 2.67 | 0.742 | 10.18 | 2.64 | 0.763 |
Comm | 11.35 | 2.99 | 0.841 | 11.31 | 2.95 | 0.844 |
T&D, training and development; Rem, remuneration; PM, performance management; SS, supervisor support; Sta, staffing; Div, diversity management; Com, communication.
The range of the Cronbach’s alpha reliability coefficients of the HRPS scales varied from 0.735 and 0.845 for men and 0.710 and 0.853 for women. The reliabilities of the seven scales were uniformly satisfactory and similar across men and women. Given the evidence in support of strict measurement invariance these reliabilities can be assumed to be invariant across the groups. As a last step the correlations between the latent constructs as well as the scale scores were calculated and are presented in
Factor and scale correlations of the Human Resource Practices Scale.
Variable | T&D | Rem | PM | SS | Sta | Div | Comm |
---|---|---|---|---|---|---|---|
T&D | (0.85) | 0.46 | 0.49 | 0.41 | 0.45 | 0.45 | 0.49 |
Rem | 0.51 | (0.79) | 0.65 | 0.44 | 0.51 | 0.48 | 0.52 |
PM | 0.56 | 0.78 | (0.75) | 0.55 | 0.54 | 0.54 | 0.63 |
SS | 0.45 | 0.47 | 0.61 | (0.84) | 0.46 | 0.44 | 0.51 |
Sta | 0.55 | 0.62 | 0.68 | 0.55 | (0.84) | 0.58 | 0.54 |
Div | 0.53 | 0.56 | 0.65 | 0.52 | 0.72 | (0.73) | 0.60 |
Comm | 0.56 | 0.56 | 0.73 | 0.56 | 0.65 | 0.71 | (0.84) |
Note: Factor correlations are below the diagonal. Scale correlations are above the diagonal. Coefficient alphas are on the diagonal, in parentheses. All correlations are statistically significant (
T&D, training and development; Rem, remuneration; PM, performance management; SS, supervisor support; Sta, staffing; Div, diversity management; Com, communication.
Across the groups, medium-sized correlations between factors were observed, which points to some, but not excessive, overlap of the seven factors. This affirms the interrelatedness of the HRM functions (see Becker et al.,
The objectives of the study were (1) to examine if the HRPS structure could be replicated across gender groups, (2) to examine the level of measurement invariance attained across men and women and (3) to report on the psychometric properties of the HRPS when used in South African organisations.
The results of the maximum likelihood χ² suggest that the hypothesis of perfect fit for all the measurement models had to be rejected (see
The ΔCFI values in
In conducting this research the seldom-answered call for questioning the assumption of measurement invariance (Tsaousis & Kazi,
The research also affirms the multidimensional conceptualisation of HRM practices, as presented by Nyawose (
This study contributes to addressing limitations in the existing literature and practice through validating the factorial structure of the HRPS and its invariance across the gender spectrum. The results empower industrial psychologists in South Africa to use the HRPS to assess the level at which employees are satisfied with the delivery of HRM services across gender. The HRPS is now in compliance with the specifications of the
The distribution of men and women in the sample presents an over-representation of women when considering the demographics of the South African workforce (Statistics South Africa,
The results provide ample evidence of measurement invariance of the HRPS across gender in the workplace context in South Africa and also support the veracity and stability of the elements among job incumbents in South Africa. After establishing measurement invariance, it will be appropriate for researchers to proceed with testing substantial hypotheses about the means and interrelations between these latent constructs across groups (Hirschfeld & von Brachel,
We gratefully acknowledge the assistance of the 2014 Master of Business Leadership class at the Graduate School of Business Leadership at the University of South Africa for collecting the data.
The authors declare that they have no financial or personal relationship(s) that may have inappropriately influenced them in writing this article.
R.S. constructed the research as part of his doctoral degree and therefore wrote the concept paper. G.d.B was the supervisor of the study and assisted in doing the statistical analysis and writing the final manuscript.
Constructs and items of the Human Resource Practices Scale.
Construct | # | Item |
---|---|---|
Training and development | 1 | My company is committed to the training and development needs of its employees. |
2 | Employees are encouraged to accept education and training within the company. | |
3 | This organisation has provided me with training opportunities enabling me to extend my range of skills and abilities. | |
Remuneration | 4 | My salary and benefits have been an adequate return for the time and energy demanded of me. |
5 | I am satisfied with my company reward system to compensate good performance. | |
6 | The company’s compensation and reward system encourages team and individual contributions. | |
Performance management | 7 | My company’s performance management system is fair and based on clear objectives at the beginning of the term/year. |
8 | The company has provided enough information regarding specific methods of the performance evaluation system. | |
9 | Employees are allowed to formally communicate with supervisors/managers regarding the appraisal results. | |
Supervisor support | 10 | My supervisor would personally use his/her power to help me solve my work problems. |
11 | My supervisor always gives credit and encourages an employee for a job well done. | |
12 | My supervisor often lets me know how well he/she thinks I am performing the job. | |
Staffing | 13 | Proper company procedures and processes are always followed when staffing/recruitment decisions are made. |
14 | Interview panels are used during the staffing process in this organisation. | |
15 | All appointments in this organisation are based on merit (i.e. the best person for the job is selected, regardless of their personal characteristics). | |
Diversity management | 16 | The company spends enough time and effort on diversity awareness related to race, gender and religion. |
17 | Management is supportive of cultural difference in this organisation. | |
18 | People living with disabilities have employment opportunities in this organisation. | |
Communication | 19 | My company regularly provides information sharing sessions to all employees. |
20 | Continuous improved communications between management and staff is stated as an important company objective and is being practiced. | |
21 | My company’s communication channels are open and effective in dealing with matters that are relevant to employees. |