The entrepreneurial climate in organisations is often seen as an important antecedent to innovation and organisational success. Assessing the nature of the climate in a reliable and valid manner is essential, as this will guide the implementation of appropriate interventions where necessary as well as assessing the effects of such interventions.
The aim of this research was to evaluate the psychometric properties of a measure of entrepreneurial climate. Entrepreneurial climate was measured using a shortened version of the Hornsby, Kuratko and Zahra (
The setting was medium to large South African companies. A random sample of employees was drawn from 53 selected companies across South Africa, with 60 respondents per company (
A cross-sectional survey design was used. Several instruments were administered, including the shortened version of the CEAI. Cronbach’s alpha was used to test for reliability and several methods were used to test for validity. Correlation analysis was used to test for concurrent validity, convergent validity and divergent validity. Principle component factor analysis was used to test for factorial validity and a t-test to test for known-group validity.
The results showed that the reliability for the total score of the shortened version of the CEAI was acceptable at 0.758. The results also showed some evidence of concurrent validity, as well as homogeneity among the items. With regard to factorial validity, all items loaded in accordance with the subscales of the instrument. The measure was able to distinguish, as expected, between government organisations and private business entities, suggesting known-group validity. Convergent validity and divergent validity were also assessed. Interesting to note was that entrepreneurship climate correlates more with general employee attitude (e.g. employee engagement;
This paper not only provided information on the reliability and validity of the shortened version of the CEAI in the South African context but also provides norms to be used when researchers or consultants work with smaller groups. Recommendations on the appropriate use of the instrument are offered and this contributes to the responsible use of the instrument.
Though some confusion exists on the exact meaning of innovation in the workplace (Hind & Steyn
Irrespective of the differences in the exact definition of innovation, it is seen as important and considered to be an essential component for competitiveness and survival, embedded in organisational structures, processes, products and services within the organisation (Gunday et al.
The climate in organisations is appreciated by many as an important antecedent to innovation and organisational success (Choi, Moon & Ko
The primary objective of this research was to evaluate the psychometric properties of a measure of entrepreneurial climate. The Hornsby et al. (
The literature review comprises two parts, namely reliability and validity. Both reliability and validity are essential for effective measurement (American Educational Research Association et al.
Many types of reliability are reported in literature, including test–retest reliability, half-split reliability, parallel-forms reliability and internal consistency (Cohen et al.
What an acceptable coefficient alpha constitutes is a matter of debate. Guilford and Benjamin (
Definitions of validity vary. The latest standards for educational and psychological testing, jointly developed by the American Educational Research Association, American Psychological Association and the National Council on Measurement in Education, emphasise the use of tests in their definition of validity: ‘Validity refers to the degree to which evidence and theory support the interpretations of test scores entailed by proposed uses of tests’ (American Educational Research Association et al.
Three types of validity have traditionally been identified, namely content validity, criterion-related validity and construct. The classic trinitarian view of validity is still common among contemporary authors on psychometrics (Cohen et al.
The
Construct validity can also be derived from
Closely aligned to the aforementioned discussion is what Moerdyk (
A fourth test of construct validity is when ‘test scores correlate with scores on other tests in accordance with what would be predicted from a theory that covers the manifestation of the construct in question’ (Cohen et al.
From the aforementioned discussion it is clear that construct validity is a complex matter and judgement on the construct validity of a test ought to be the result of integrating several sources of information.
In this section, the respondents, the procedure, the measuring instruments, the data analysis and the ethical considerations are discussed.
To be included in the study, respondents needed to be employed at a large South African organisation, with a workforce of at least 60 employees. Several organisations were approached and 53 companies were eventually willing to participate. In total 60 respondents were randomly selected from each of these organisations. This presented a convenient sample of South African organisations, but a random sample of employees. More detail about the respondents is reported in the findings section.
The sampling of the respondents is discussed above. The data were generated from paper-and-pencil tests, completed in organisations where permission was granted by the appropriate authorities, and all respondents gave consent. These data were not primarily collected for this research and are archival data collected by the author as part of a larger research project. The ethical clearance obtained allows him to collect and use the data in further analysis and to publish academic articles based on the data. This use of the data was clearly stated in the permission letters as well as the consent forms. After cleaning up the data appropriate statistics were calculated. Cleaning up the data was limited to removing of limit data and replacing it with missing values. The statistical techniques used are described in the section ‘Data analysis’.
Eight instruments were administered.
A shortened version of the CEAI (Hornsby et al.
Top management support: The extent to which employees perceive that top managers support, facilitate and promote entrepreneurial behaviour. This includes top management’s championing innovative ideas and providing the resources required for entrepreneurial actions.
Work discretion: The extent to which employees perceive that the organisation tolerates experimentation (and failure). Furthermore, work discretion relates to decision-making autonomy and freedom from unwarranted oversight and also management, which delegates authority and responsibility to lower-level managers and workers.
Rewards and reinforcement: The extent to which employees perceive that the organisation uses systems which reward entrepreneurial activity and success.
Time availability: The extent to which employees experience their job’s structure in such a way that unstructured or free time is available to allow individual employees or groups to pursue innovations.
Organisational boundaries: The extent to which employees perceive that organisational boundaries are flexible and allow the flow of information within the organisation and beyond the organisation and the external environment. Flexile but clear boundaries are tested for. Boundaries induce, direct and encourage coordinated innovative behaviour.
The shortened version proposed and used by Strydom (
The Utrecht Work Engagement Scale-9 (UWES-9; Schaufeli & Bakker
The Organizational Commitment Scale (Allen & Meyer
The Innovative Work Behaviour (IWB) questionnaire was developed by de Jong and den Hartog (
Kleysen and Street (
The Quality of Performance Appraisal Systems Questionnaire (QPASQ) was used to assess the perceived effectiveness of the (traditionally defined) performance appraisal systems in organisations. The QPASQ was developed by Steyn (
The Human Resource Practices Scale (Nyawose
The Multifactor Leadership Questionnaire (Avolio, Bass & Jung
Demographical information about the sample, as well as descriptive statistics on the instrument of interest, the shortened version of the CEAI, were calculated. The mean, standard error of mean, standard deviation, skewness and kurtosis for the CEAI
Next calculations were done concerning the reliability of the instrument. Cronbach’s alpha coefficient and the strict parallel method for calculating reliability, as generated through the SPSS-23 programme, were calculated. As many authors (Clark & Watson
Several calculations were done with regard to gathering information on validity. The first related to criterion validity. To test for concurrent criterion-related validity, the correlation between CEAI, as independent variable, and IWB and IIB, as dependent variables, was calculated. A statistical significant correlation (
The rest of the analyses concerned construct validity. Firstly, the homogeneity of the items was tested. The common inter-item correlation, for the entire as well as the five subscales of CEAI, was calculated. Guidelines on the size of an acceptable inter-item correlation vary, but for this research the range was set in line with Clark and Watson’s (
Factorial validity was assessed through testing if the different subscales of the CEAI loaded in different factors. This is a simple analysis to test whether the items of the subtest correlate more with the subtest’s own items than with items from another subtest (Nunnally & Bernstein
Tests of known-group differences were conducted next. Firstly, it was tested whether government organisations (including parastatal entities) show less corporate entrepreneurship than private business entities, and secondly if managers show more corporate entrepreneurship than non-managers. One-way analysis of variance was performed in both cases. In the case of the multiple groups, the Scheffe post hoc test was performed to detect which groups differed from each other. Statistical significance of differences (
Information on convergent, and to a lesser extent divergent validity, was created by calculating the correlation between CEAI and several other measures. Firstly, it was hypothesised that corporate entrepreneurship would correlate significantly with innovative behaviour, more than generic organisational attitudes (employee engagement and organisational commitment). It was hypothesised that CEAI would correlate more with active forms of leadership (transformational and transactional) than with more passive forms of leadership (laissez-faire). Also, it was hypothesised that CEAI is related to performance, and as such, it would correlate more with a measure of effective performance management, rather than with a general measure of human resources management. Lastly, it was hypothesised that CEAI would correlate more with attitudes towards the job (employee engagement) than with attitudes towards the organisation (organisational commitment). A correlation of 0.5 would be seen as a clear sign of convergence, following Gregory’s (
Factor analysis would also be performed to test for divergent validity. This was done through forcing CEAI and each of the other measures used in this study into a two-factor solution, and then performing a Varimax method rotation with Kaiser normalisation. After suppressing values smaller than 0.4, to make interpretation easier, the percentage of items that load (correctly) on the appropriate factor was calculated. Should 70% of the items with loadings of 0.4 load ‘correctly’, it would be interpreted as signalling divergent validity. Should this percentage not be achieved, it would point to a lack of divergent validity. The tolerance for items loading ‘incorrectly’, indicative of poor divergent validity, was set at 10%. Thus, should more than 10% of items load ‘incorrectly’, the instrument would be seen as invalid from a factorial point of view.
Ethical clearance for the collection of the data was obtained from the ethics committee of the University of South Africa’s Graduate School of Business Leadership (2014_SBL_018_CA dated 27 February 2014). All standard requirements for collecting data from human subjects were followed and no breaches of procedures were reported during the collection process.
The total sample consisted of 3180 employees, representative of 53 companies. In total 57.1% reported that they were male employees, compared to 42.5% reporting that they were female employees (missing data = 0.4%). As far as race is concerned, 8.3% marked Asian, 58.4% black people, 8.4% mixed race and 24.6% white people (missing data = 0.3%). Their ages ranged between 20 and 72, with an average of 37.80 (standard deviation = 9.11). As far as tenure at their present company is concerned, it varied between 1 month and 42 years, with an average of 8.39 (standard deviation = 7.47).
As far as functions are concerned, the findings showed that 46.6% indicated that they were involved in the core business of the company, with 52.8% reporting that they fulfil supportive roles (missing data = 0.5%). With regard to position, 36.5% indicated that they hold some kind of managerial position, while 62.9% reported that they did not form part of management (missing data = 0.7%). In
Description of post levels of respondents.
Semantic scale | Patterson | % |
---|---|---|
Senior management and top management | E and F | 5.0 |
Professionally qualified, experienced specialists and middle management | D | 27.7 |
Skilled technical and academically qualified workers, junior management, supervisors, foremen and superintendents | C | 43.1 |
Semi-skilled and discretionary decision-making | B | 20.1 |
Unskilled and defined decision-making | A | 2.4 |
Concerning formal schooling, 5.0% reported that they received less than 12 years of formal schooling and 25.5% said that they had completed 12 years of formal schooling. A further 40.2% reported that they had completed a degree or diploma, while 28.9% indicated that they had a higher degree or higher diploma (missing data = 0.4%).
The descriptive statistics for the total instrument, as well as for the five subscales, are presented in
Descriptive statistics for subscales and shortened Corporate Entrepreneurship Assessment Instrument total score.
Subscales or total score | Mean | Standard error of mean | Standard deviation | Skewness |
Kurtosis |
---|---|---|---|---|---|
Management support | 12.77 | 0.053 | 3.01 | −0.467 | −0.011 |
Work discretion | 13.57 | 0.060 | 3.42 | −0.386 | −0.299 |
Rewards and reinforcement | 13.39 | 0.054 | 3.10 | −0.347 | 0.008 |
Time availability | 11.15 | 0.053 | 3.03 | −0.025 | −0.429 |
Organisational boundaries |
14.99 |
0.046 |
2.63 |
−0.585 |
0.656 |
The standard error for skewness was 0.043 and the standard error for kurtosis was 0.087 for the total score.
Almost all subscales were skewed to the left, with the exception of time availability, which was within the boundaries of normality (-0.281 to 0.281). The total score was also skewed to the left, with a value of -0.291. With regard to kurtosis, the subscale time availability had a heavier tail and the subscale organisational boundaries had a sharper peak. However, the total CEAI score of 0.257 fell well within the boundaries of -0.47 to 0.62.
Reliability was reported as per the Cronbach’s alpha coefficient and the strict parallel method, as generated through the SPSS-23 programme. The coefficient for the total instrument was 0.758 (20 items). The unbiased reliability was 0.723 and the common inter-item correlation was 0.115. The coefficients for the subscales are presented in
Reliability coefficients and common inter-item correlation for subscales and Corporate Entrepreneurship Assessment Instrument total score.
Subscales or total score | Cronbach’s alpha | Strict parallel unbiased reliability | Common inter-item correlation |
---|---|---|---|
Management support | 0.676 | 0.662 | 0.328 |
Work discretion | 0.746 | 0.714 | 0.384 |
Rewards and reinforcement | 0.635 | 0.623 | 0.292 |
Time availability | 0.586 | 0.562 | 0.242 |
Organisational boundaries |
0.560 |
0.502 |
0.201 |
Interesting to note from the above is that common inter-item correlation related positively to reliability for the subscales (each with four items), though this parallel is not found for the total instrument (with 20 items).
The results pertaining to the correlation and CEAI and innovative behaviour are presented in
The correlation between Corporate Entrepreneurship Assessment Instrument scores and innovative behaviour in the workplace.
Corporate entrepreneurship | Innovative work behaviour | Individual innovative behaviour |
---|---|---|
Management support | 0.234 | 0.216 |
Work discretion | 0.206 | 0.201 |
Rewards and reinforcement | 0.223 | 0.188 |
Time availability | 0.014 | 0.023 |
Organisational boundaries |
0.162 |
0.183 |
Note: In all cases,
The correlation coefficients presented in
Evidence on construct validity is presented under four subheadings.
As stated above, the homogeneity of the items, expressed as inter-item correlations, could be indicative of content validity. The inter-item correlation of the subscales varies between 0.201 and 0.384, and for the total score it is 0.115 (see
As stated before, Cronbach’s alpha could also be indicative of homogeneity. For the total CEAI score as well as for work discretion, a margin of 0.7 was reached. Unlike in the previous paragraph, where inter-item correlations were used, the CEAI met the requirement of homogeneity using the internal consistency measure.
The Kaiser–Meyer–Olkin measure of sampling adequacy was performed and the value was 0.805. The Bartlett’s test of sphericity was also conducted and the approximate chi-square value of 11753.89 (degrees of freedom = 190) was significant at a level smaller than 0.001. Given that these values were acceptable, the principle component method was use for factor extraction and this was based on eigenvalues greater than 1. Five factors met the eigenvalue criteria, and these five factors explained 50.5% of the variance in the data. The factors were rotated using the Varimax method with Kaiser normalisation, and this is presented in
Rotated component matrix of Corporate Entrepreneurship Assessment Instrument items.
Item | Subscale | Component |
||||
---|---|---|---|---|---|---|
1 | 2 | 3 | 4 | 5 | ||
1 | Management support | 0.063 | 0.718 |
0.088 | 0.069 | −0.043 |
2 | Management support | 0.104 | 0.734 |
0.117 | 0.130 | −0.005 |
3 | Management support | 0.061 | 0.644 |
0.247 | 0.098 | 0.067 |
4 | Management support | 0.132 | 0.628 |
0.056 | −0.002 | 0.074 |
5 | Work discretion | 0.653 |
0.202 | 0.041 | 0.097 | 0.035 |
6 | Work discretion | 0.828 |
0.100 | 0.078 | 0.027 | 0.021 |
7 | Work discretion | 0.822 |
0.073 | 0.123 | 0.006 | 0.052 |
8 | Work discretion | 0.623 |
0.025 | 0.105 | 0.179 | −0.038 |
9 | Rewards and reinforcement | 0.051 | 0.060 | 0.686 |
0.060 | 0.076 |
10 | Rewards and reinforcement | 0.080 | 0.017 | 0.659 |
0.081 | −0.121 |
11 | Rewards and reinforcement | 0.103 | 0.232 | 0.727 |
0.076 | 0.093 |
12 | Rewards and reinforcement | 0.123 | 0.299 | 0.559 |
0.109 | 0.030 |
13 | Time availability | 0.155 | 0.206 | 0.171 | 0.209 | 0.674 |
14 | Time availability | −0.068 | −0.158 | −0.067 | −0.070 | 0.650 |
15 | Time availability | 0.103 | 0.177 | 0.076 | 0.050 | 0.675 |
16 | Time availability | −0.062 | −0.041 | −0.050 | −0.049 | 0.639 |
17 | Organisational boundaries | 0.100 | 0.133 | 0.163 | 0.733 |
0.023 |
18 | Organisational boundaries | 0.119 | 0.097 | 0.087 | 0.761 |
0.045 |
19 | Organisational boundaries | 0.139 | −0.007 | −0.011 | 0.492 |
−0.068 |
20 | Organisational boundaries | −0.070 | 0.065 | 0.079 | 0.595 |
0.066 |
Loadings higher than 0.4.
The above rotation converged in five iterations. The results presented in
Validity was also assessed by considering whether the measure could distinguish between groups where differences were expected. In this case it was foreseen that government organisations (including parastatal entities) would show less corporate entrepreneurship than private business entities and that non-managers would similarly show less corporate entrepreneurship than managers. The results revealed that the total scores of private business (
It was hypothesised that corporate entrepreneurship would correlate significantly with certain constructs, and not with others. From
Correlation between Corporate Entrepreneurship Assessment Instrument scores and related measures.
Corporate entrepreneurship | Transformational leadership | Transactional leadership | Laissez-faire leadership | Human resources practices scale | Quality of performance appraisal | Organisational commitment scale | Utrecht work engagement scale-9 |
---|---|---|---|---|---|---|---|
Management support | 0.365 | 0.366 | 0.174 | 0.437 | 0.296 | 0.284 | 0.287 |
Work discretion | 0.209 | 0.185 | 0.151 | 0.234 | 0.139 | 0.197 | 0.255 |
Rewards and reinforcement | 0.411 | 0.464 | 0.144 | 0.495 | 0.334 | 0.287 | 0.357 |
Time availability | 0.151 | 0.150 | 0.127 | 0.126 | 0.041 | 0.033 | 0.078 |
Organisational boundaries | 0.255 | 0.256 | 0.088 | 0.322 | 0.220 | 0.209 | 0.308 |
Note: In all cases,
As hypothesised, the CEAI correlates more with active forms of leadership (transformational and transactional) than with passive forms of leadership (laissez-faire). CEAI did not correlate more with a measure of effective performance management than with a general measure of human resources practices, as hypothesised. However, CEAI did correlate more, as expected, with employee engagement than with organisational commitment. Only the correlation with human resources practices surpassed the margin of 0.5, indicating a ‘hefty’ correlation (Gregory
As explained above, factor analysis was also performed as a measure of convergent and divergent validity. Presenting the actual results is extensive, and a summary of that is presented in
Distinctiveness of Corporate Entrepreneurship Assessment Instrument items and other items exposed to factor analysis.
Items from ‘other measures’ included with CEAI items in a factor analysis | Number of items loading on the CEAI factor | Number of items loading on the ‘other measure’ factor | Total items loading correctly with a load >0.4 |
---|---|---|---|
Innovative Work Behaviour | 14/20 | 9/10 | 23/30 = 76.6% |
Individual Innovative Behaviour | 14/20 | 14/14 | 28/38 = 73.6% |
Multifactor Leadership Questionnaire | 13/20 | 18/21 | 31/41 = 75.6% |
Organisational Commitment Scale | 14/20 | 7/8 | 21/28 = 75.0% |
Utrecht Work Engagement Scale-9 | 12/20 | 9/9 | 21/29 = 72.4% |
Human Resources Practices Scale | 10/20 |
21/21 | 31/41 = 75.6% |
Quality of Performance Appraisal | 14/20 | 17/18 | 31/38 = 81.5% |
One item loaded on the ‘other measure’ factor; ‡, One item loaded on the ‘other measure’ factor, thus 1/41 = 2.4%.
In all the (factor) analyses presented above, the requirements were met which could define CEAI as a distinct measure. This is evidence of divergent validity.
Data were sampled from 53 companies. In total, 3 180 employees were respondents in this study. From a demographic point of view, most respondents were men, reported that they were members from the black group and that they were employed in non-core roles. Furthermore, most respondents reported that they did not form part of management. The majority categorised themselves as technically skilled and as being part of junior management or working on a supervisory level. Though their demographic characteristics varied widely, the respondents’ profiles overall mirrored those of the current South African workforce profile. This meets the call for using a non-Western data to verify the psychometric properties of the CEAI.
The descriptive statistics for the CEAI are presented in
Reliability for the total CEAI, as reported as per the Cronbach’s alpha coefficient and the strict parallel unbiased reliability, was acceptable at 0.758 and 0.723. As coefficients for some of the subscales were below the set margin, it would be desirable to rather use the total score.
Test of criterion-related validity revealed that the CEAI subscale of management support related most with innovation at work, while the contribution of the subscale time availability was small. When combining the subscales, the correlation between CEAI and innovation at work remained small (
Evidence on construct validity is presented below. Several measures were used to assess construct validity, as well as some of the results already presented above. The
The findings pertaining to
In this article the psychometric properties of a shortened version of the CEAI are presented. The discussion of the psychometric properties was informed by data from a relatively large sample representative of numerous companies. This sample size compares well to other studies investigating psychometric properties. The reliability scores of the total instrument were acceptable. Also, though the validity evidence was mixed, a multitude of evidence found supported validity. It may be expected that when you collect evidence from a large number of sources, some may yield the consistent confirmation. Given all the evidence provided, particularly the evidence obtained from both applications of factor analysis, it is judged that the CEAI has acceptable validity.
Researchers and practitioners are urged to use the shortened version of the CEAI, using the 20 instead of 48 items. The shortened version of the CEAI showed acceptable reliability and validity and the use of the central statistics provided in
Though the sample size was relatively large, and sampling within organisations was done randomly, organisations were selected on the basis of convenience. The generalisation of the results is thus limited. Furthermore, a judgement had to be made regarding the overall validity of the instrument, as not all the indicators of validity were positive. Although subjectivity was uncomfortable, most authors refer to validity assessment as a judgement call. Further research on this is needed to demonstrate evidence of validity. Also, additional statistical techniques, such as structural equation modelling, could be used in future studies. This may provide additional insights about the topic. The research made use of only a single method of data collection, self-reporting. This limitation may be mediated by adding additional methods of reporting, and this is recommended for future studies.
The authors declare that they have no financial or personal relationships which may have inappropriately influenced them in writing this article.
Discriminant validity is often used as a synonym to divergent validity. In this paragraph the term ‘discriminant validity’ is preferred as to align the content of the text to the sources consulted. In the rest of the text the term divergent validity will be used.
In most of the discussions that follow, reference will be made to the Strydom (