Ethical climate has a significant influence on employees’ work engagement where it acts as an organisational resource to bring about positive change and thereby increase organisational effectiveness (Mitonga-Monga & Cilliers, 2015). The presence of an ethical climate reflects an organisation’s processes, procedures, policies and practices, with moral consequences that may help employees in the execution of their daily work (Mitonga-Monga, 2018). An ethical caring climate in particular is believed to be associated with lower organisational deviance (Pagliaro et al., 2018; Tastan, 2019). Of importance to observe is that ethical climates reflect the climate that is adopted by management as it is leaders who define what ethical behaviour is and how ethical issues should be handled within organisations. The importance of ethical climate in influencing positive organisational outcomes highlights the importance of the measurement thereof. The ECQ was identified as a widely used measures of ethical climate. However, this measure was developed in America. It was therefore deemed necessary to ascertain the reliability and validity of the instrument within the Zimbabwean context. This study adds to the body of knowledge on the psychometric properties of the ECQ.

Ethical climates arise when leaders develop organisational policies and practices that exert significant influence on the ethical decision making of organisational members and their attitudes and behaviours at work (Schminke et al., 2007; Simha & Cullen, 2012). The origins of ethical climate theory are attributed to Victor and Cullen (1988) who defined an ethical climate as the prevailing perceptions of typical organisational practices and procedures that have ethical content. It is characterised by the processes through which organisations implement and enforce ethical rules and policies to stimulate ethical behaviours and punish unethical behaviours (Schwepker, 2013). Olson (1998) suggested that ethical climate creates an environment that enhances ethical decision making and behaviours.

Arnaud (2010) broadens the definition of ethical climate by defining it as a moral concept that reflects the content and strength of the prevalent ethical values, norms, attitudes, feelings and behaviours of the members of a social system. Arnaud (2010) developed the theoretical model of EWC as a critique of Victor and Cullen’s (1987, 1988) framework, questioning whether it was ‘comprehensive enough to capture the true breadth of the ethical climate construct’. While comprising the moral reasoning dimension that formed the basis of Victor and Cullen’s model, Arnaud also incorporated dimensions of Rest’s (1984, 1986) four-component cognitive development model, these being the dimensions of moral sensitivity, moral character, and collective moral motivation. However, Arnaud’s model lacks sufficient empirical support partly because of the popularity of Victor and Cullen’s conceptualisation.

Victor and Cullen (1987) developed a theoretical stratum of ethical climate that is grounded in Kohlberg’s (1984) theory of cognitive moral development that proposed that individual moral development is person- and context-specific. Victor and Cullen (1987) adapted this model to an organisational setting (Buchan, 2009) that comprises two dimensions. According to this model, the first dimension is a classification of the ethical climate based on the moral development of the work group, these being egoism, benevolence and principled. Furthermore, there are three loci of analysis that comprise the second dimension, these being the individual, local and cosmopolitan. When these dimensions are cross-classified together they form nine hypothetical approaches to ethical climate (Victor & Cullen, 1987). This nine-cell typology (Victor & Cullen, 1987) of ethical climates determines the condition of an organisation’s ethical climate and how ethical issues are dealt with.

The individual locus of analysis identifies the sources of analysis as residing in the individual such as self-interest (Grobler, 2016; Victor & Cullen, 1988). At the individual level of the criterion principle, morals are self-chosen (one’s morals), which is the principle loci. The local locus of analysis on the other hand identifies the sources of analysis as pertaining to the organisational collective (e.g. company profit). The basis of morals is founded in the organisation at the local locus (its rules and regulations). (principle loci). The cosmopolitan locus of analysis considers those outside the organisation as social responsibility (Grobler, 2016; Victor & Cullen, 1988). Moreover, the source of morals is found outside the organisation (laws and professional codes of ethics) at the cosmopolitan locus. In the context of the egoism criterion, an individual locus of analysis shows a climate that upholds self-interest while the local level of analysis considers the organisation’s interests (profit). Finally, the cosmopolitan locus of analysis considers society’s best interest (efficiency) (Victor & Cullen, 1988; Wyld & Jones, 1997). Benevolence aligns with the friendship at the individual level, team interest at the local level and the social responsibility at the cosmopolitan level as indicated in Table 1.

As a result of insufficient reliable empirical support uniformly across all the nine theoretical ethical climate types, Victor and Cullen (1988) refined the original nine types into five climates for which empirical support was found (Victor & Cullen, 1988). The five types of ethical climates are caring, law and code, rules, instrumental and independence (Martin & Cullen, 2006; Victor & Cullen, 1988).

A caring climate is characterised by an environment in which employees are sincerely interested in the well-being of each other (Nafei, 2015). This genuine and sincere attitude is directed towards individuals within and outside the organisation who are likely to be affected by their decisions. The caring dimension is associated with the benevolence ethical criteria at the individual level (Nafei, 2015). A law and code climate entails a context where employees are expected to abide by the laws, codes and professional standards of the organisation and government (Nafei, 2015). The law and code dimension is associated with the principle criteria at the cosmopolitan level (Grobler, 2016). A rules climate is associated with employees strictly obeying the rules and procedures of the organisation or their subunit (Victor & Cullen, 1988). This dimension is associated with the principled ethical criteria at a local level (Victor & Cullen, 1988). An instrumental climate focuses on the fulfilment of individual interests. It pertains to the extent to which employees look out for their self-interest in relation to company interest and that results in it being regarded as a negative type of climate (Nafei, 2015). An independence climate allows employees to follow and be guided by their own moral beliefs when making decisions (Nafei, 2015). Attention is given to the individual’s convictions and personal morality. An independence climate is associated with principal criteria at the individual level (Nafei, 2015).

While the aforementioned suggests distinctions between typologies, organisations generally encompass multiple climate types because of the fact that they comprise diverse facets. Furthermore, Victor and Cullen (1987, 1988) stress that the climate type reflected by an organisation is linked to the ethical dilemmas typically faced in that organisation.

The ECQ was developed by Victor and Cullen (1987) as a measure of ethical organisational climate and based on the typology presented here as established by theory. Barnett and Vaicys (2000) describe the ECQ ‘as a typology based on ethical philosophy as well as the sociological theory of reference groups’. The ECQ assesses the perceptions of organisational members on how they deal with ethical choices (Appelbaum et al., 2005). As indicated, the original ECQ incorporated nine ethical climate types established by theory. This version required respondents to specify how precisely how each of 26 items describes their work climate on a six-point Likert-type response scale (Victor & Cullen, 1987). The ECQ thus regards respondents as observers, reporting on but not evaluating the perceived organisational climate. Ten additional items were added to the scale when Cullen et al. (1993) performed a review, which resulted in very strong reliability and validity. Cullen et al. (1993, p. 673) reported that the 36-item version of the questionnaire yielded seven climate types as the benevolence, climates of friendship and team interest loaded on the same factor thereby prompting the authors to conclude that ‘the questionnaire lacks the precision necessary to discriminate between these two climates’. In addition, social responsibility was found to be a discrete climate in the same study. The 26-item version was also chosen because it yielded more ethical climates that are easier to interpret than the 36-item version as argued by Grobler (2016). While limitations have been identified in Victor and Cullen’s (1987, 1988) conceptualisation and framework of measurement (Salamon & Mesko, 2016), the ECQ remains a preferred measure of ethical climate because it has been widely validated (Dark & Rix, 2015). This provides a sound basis for the comparison of findings between studies.

The 26-item version of the ECQ evaluating five climate types was used in this study. The scale assesses organisational climate against the five dimensions of caring, law and code, rules, instrumental and independence. Scale items are accompanied by six-point Likert-type response scale ranging from disagree strongly (1) to agree strongly (6). Victor and Cullen (1988) found evidence of acceptable reliability for the 26-item ECQ. The reliability coefficients ranged from 0.73 to 0.81, which are generally acceptable except for the independence dimension whose Cronbach’s alpha was 0.65 (Tabachnick & Fidel, 2007). Wimbush et al. (1997) confirmed these findings pertaining to the reliability for the dimensions, which ranged from α = 0.69 to α = 0.92. The independence dimension also had a relatively lower reliability (α = 0.69) that was, however, sufficient to establish internal consistency. Victor and Cullen (1988) reported evidence of convergence validity in the parameter estimates and t-values of the ECQ. The measure was developed in the USA and typically utilised in Japan, Singapore, China and Australia. While the psychometric properties have been established for these countries, evidence of use in African countries such as Zimbabwe is not available.

This study, therefore, assessed the reliability and construct validity of the ECQ on a sample of participants from selected organisations in Zimbabwe.

The population was made up of 5000 private security employees drawn from 10 companies. The overall sample of the study consisted of 304 participants drawn from selected security organisations in Zimbabwe. The response rate was 76%. Females constituted 28.9% while males constituted 69.9%. The majority (42.4%) fell in the 31–40 age category and 28.9% fell within the 41–50 age category. All of the respondents were black Africans. A total of 89.1% of the respondents had an ordinary level certificate (equivalent of grade 11) as their highest level of education. A total of 64.5% of the participants were permanent employees while 24.7% were on contract.

Ethical climate was measured using the 26-item version of the ECQ (Victor & Cullen, 1988). The questionnaire has five subscales for measuring caring; 7 items, law and code; 4 items, rules; 4 items; 7 items for instrumental and independence with 4 items. While evidence could not be found that the ECQ had been utilised in studies within Zimbabwe, the internal reliability of the subscales had been determined within the South African context. These ranged from α = 0.79–0.87 as reported by Wolmarans (2014). Example items are as follows: Caring: ‘What is best for everyone in the company is a major consideration here’; Law and code: ‘People are expected to comply with the law and professional standards over and above other considerations’; Rules: ‘It is very important to follow the company’s rules and procedures here’; and on Independence: ‘In this company, people are expected to follow their own personal and moral beliefs’. An example item for the instrumental subscale is: ‘In this organisation, people protect their own interests above all else’.

The questionnaires were distributed face-to-face through the human resources management (HRM) and operations units of the organisations. These two departments (HRM and operations) closely assisted in the study because they had easy access to all the employees in the organisations. Convenience sampling was used and only willing participants took part. The participants were given 2 weeks to complete the questionnaires. The questionnaire was in English, and most of the employees were proficient in English as the minimum recruitment criteria are five ordinary-level subjects, which include English. The researcher personally collected the questionnaires directly from the participants to ensure confidentiality. Participants were assured that their responses would be treated with anonymity and that no names would be revealed in the study. Informed consent was obtained from the respondents before they completed the questionnaires and voluntary participation in the study was emphasised.

The following analyses were conducted in this study: (1) Item analysis, to determine the reliability of the scales using the reliability procedure in the SPSS program (corrected-item statistics, inter-item correlations and scale-if-item is deleted); (2) Exploratory factor analysis (EFA) (using the principal axis factoring method and direct oblimin rotation), to ascertain scale dimensionality; and (3) structural equation modelling (SEM) to determine the construct validity of the ECQ.

Structural equation modelling is a collection of statistical techniques that allow a set of relationships representing causal hypotheses between one or more independent and dependent variables to be evaluated (Tabachnick & Fidell, 2013). It is a large sample technique that helps in explaining the patterns of covariances found among variables in terms of hypothesised relationships suggested by the measurement thereof and structural models (Diamantopoulos & Siguaw, 2000). Structural equation modelling is conducted to, firstly, determine if the measure reflects the intended constructs through CFA and to evaluate the measurement properties of psychological measures. Secondly, SEM allows for the specification and testing of path models. Lastly, SEM simultaneously assesses the quality of measurement and examines the predictive relationships among constructs by performing confirmatory factor analysis and path analysis at the same time. Structural equation modelling was used to perform a confirmatory factor analysis on the observed inter-item covariance matrix using robust maximum likelihood estimation.

The CFA is a technique by which hypotheses or theories relating to the structure underlying a set of variables are tested (Pallant, 2016). The CFA serves to confirm whether a set of measures (the observed data) are related to specific latent variables according to the form described in the measurement model (Blaikie, 2003) by producing a series of fit indices. These indices allow the researcher to establish how well the measurement model with its parameter estimates fits the observed data. The measurement model describes how the indicator variables are meant to reflect the specific underlying latent variables that they were earmarked to represent. The goodness-of-fit of the measurement model was tested with the CFA technique available in LISREL 8.80 (Jöreskog & Sörbom, 2006). In CFA, the number of factors or latent variables and the pattern of indicator-factor loadings are specified in advance. The pre-specified factor solution is evaluated in terms of how well it reproduces the sample covariance matrix of the measured variables (Brown, 2006).

The evaluation of the ECQ CFA models was based on the following indices: the root mean square error of approximation (RMSEA); root mean squared residual (RMR); the goodness-of-fit index (GFI); normed fit index (NFI); non-normed fit index (NNFI); comparative fit index (CFI); incremental fit index (IFI) and the relative fit index (RFI).

The RMSEA is the most informative fit indices that tests for the closeness of fit in the null hypothesis (Diamantopoulos & Siguaw, 2000; Schumacker & Lomax, 2016). The RMSEA values below 0.05 are suggestive of a good fit; those between 0.05 and under 0.08 suggest reasonable fit; values between 0.08 and 0.10 denote mediocre fit while values > 0.10 indicate poor fit (Diamantopoulos & Siguaw, 2000).

The RMR presents the average value of the difference between the sample covariance (variance) and a fitted (model-implied) covariance (variance). In other words, it is a summary measure of fitted residuals (Diamantopoulos & Siguaw, 2000; Hair et al., 2010).

The GFI depicts how closely the model comes to perfectly reproducing the observed covariance matrix. The GFI is usually recommended as the most reliable measure of the model fit (Diamantopoulos & Siguaw, 2000). The values of the GFI greater than 0.90 are usually considered as indicating an acceptable fit (Diamantopoulos & Siguaw, 2000).

For model comparative purposes, the RFI; IFI; the Bentler-Bonett NNFI; the Bentler–Bonett NFI and the CFI (Diamantopoulos & Siguaw, 2000; Schumacker & Lomax, 2016) are recommended. Values greater than 0.90 are usually regarded as acceptable (Diamantopoulos & Siguaw, 2000).

The researcher received ethical clearance from the Human and Social Sciences Ethics Committee of the University of the Western Cape to conduct the study. The researcher was granted permission from the organisations from which participants were drawn.

Missing values were addressed through the use of multiple imputations (Jöreskog & Sörbom, 2006). This technique allowed the missing values to be substituted with values derived from averages with the aid of simulation (Jöreskog & Sörbom, 2006; Rubin, 1987). The final sample size was 304; no cases were deleted.

Using the SPSS version 28 (SPSS Inc., 2021), item and dimensional analyses were performed on the items of the ECQ. All of the internal consistency coefficients of the ECQ were above the acceptable cut-off level (α > 0.70) (Nunnally & Bernstein, 1994) (see Table 2). The five ECQ subscales were found to be uni-dimensional and the variance explained by each of the factors was generally above 50% (see Table 2).

The CFA was carried out on the items of the ECQ through LISREL 8.80 (Du Toit et al., 2008) to evaluate the construct validity of the measurement models. The ECQ measurement model was considered as an exogenous variable and list-wise deletion and the robust maximum likelihood estimation method were utilised to produce the required estimates. The normalised data set was used for subsequent analyses.

The two measurement models for ECQ tested in this study show similar levels of fit regarding the CFI, NFI, NNFI, IFI and RFI. Both first order and second-order measurement models indicate reasonable model fit. In terms of the RMSEA, the first order shows a value of 0.0771, while the second-order measurement models show a value of 0.0768. In terms of the SRMR the first-order measurement model shows a value of 0.0707 while the second-order model shows a value of 0.0720. The GFI values for all the models missed the 0.90 cut-off (see Table 3).

The unstandardised Gamma matrix shows the strength of the relationship and influence of the exogenous latent variable (ethical climate) on its manifest variables. The parameters are significant (p < 0.05) if t-values are ≥│1.96│ (Diamantopoulos & Siguaw, 2000). The t-values show that the dimensions are significant indicators of the ethical climate higher-order factor, as the t-values are greater than 1.96 except independence with 1.70. The results are shown in Table 4.

The inter-correlations among the five latent ECQ dimensions are presented in Table 3. The intercorrelations were within reasonable limits and not above 0.90 as this could indicate multicollinearity (Pallant, 2016; Tabachnick & Fidell, 2013), except for the independence dimension, which correlated below 0.30 with the other dimensions.

Power assessment was conducted using the Rweb (1.03) translation of the SAS syntax (Preacher & Coffman, 2006). This syntax is used to derive the power estimates for the tests of exact and close fit. In the this study, the inputs for the analysis comprised a significance level (a) of 0.05, a sample size of 304 and 146 degrees of freedom (see Table 7). A reasonably large power value of 0.9995581 for the test of exact fit was obtained. The power of the test of close fit obtained was 0.99732, which was high implying that, in the conditions that characterised this specific study, one would reject approximately 99.7% of incorrect models thereby instilling some confidence in the model.