1 / 61

Cultural Diversity and Workplace Dynamics:

Cultural Diversity and Workplace Dynamics:. A Transmanche Link retrospective. Jim Freeman and Graham Winch E-mail: jim.freeman@mbs.ac.uk. Related publications.

hedva
Download Presentation

Cultural Diversity and Workplace Dynamics:

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Cultural Diversity and Workplace Dynamics: A Transmanche Link retrospective Jim Freeman and Graham Winch E-mail: jim.freeman@mbs.ac.uk

  2. Related publications • Winch, G, Clifton, N. and Millar, C. (1997) ‘Culture and Organization: The Case of Transmanche-Link’, British Journal of Management, Vol. 8, No. 3, pp. 237–249. • Winch, G, Clifton, N. and Millar, C. (2000) ‘Organisation and Management in an Anglo-French consortium: The Case of Transmanche-Link’, Journal of Management Studies, Vol. 37, No. 5, pp. 663-687. OR54 in Edinburgh 3-6 Sept 2012

  3. Cross culturalism • Cross-cultural project teams are increasingly the norm, globally. • Such projects accounted for nearly all gross capital formation - equivalent to 22% of the World GDP - in 2009 (http://www.worldbank.org/). OR54 in Edinburgh 3-6 Sept 2012

  4. Structure of Presentation • Background • Data collection • Modelling (SEM) results • Conclusions 22 October 2014 OR54 in Edinburgh 3-6 Sept 2012 OR54 in Edinburgh 3-6 Sept 2012 4

  5. Background Cultural differences (re. Hofstede (1980)) • Power Distance • Uncertainty Avoidance • Individualism vs. Collectivism • Masculinity vs. Femininity • Long term orientation Cultural differences (re. Trompenaars and Hamopden-Turner(1998)) OR54 in Edinburgh 3-6 Sept 2012

  6. Relationships with people • Neutral versus emotional • Universalism versus particularism • Achievement versus ascription • Specific versus diffuse • Attitude to Time: Sequential versus Synchronic • Attitude to Environment • Context • Convergers and Divergers OR54 in Edinburgh 3-6 Sept 2012

  7. Data Collection • Data for the project was collected through self-completed questionnaires distributed among the staff of the TML consortium (Winch et al., 2000). • Target respondents included first line supervisory level staff and above. • The questionnaire - based on the Van de Ven and Ferry scale (Van de Ven and Ferry, 1980) – aimed to capture measures on organisation and workplace dynamics OR54 in Edinburgh 3-6 Sept 2012

  8. Additionally, Hofstede’s Value Survey Module was incorporated to identify and validate cultural differences between the British and the French working at TML. • There were 153 British and 52 French responses to the survey, representing an overall return of 39%. • Item responses were collected mainly on a 5 point Likert scale; demographic measurements on categorical scales. OR54 in Edinburgh 3-6 Sept 2012

  9. Data hygiene checks • Outliers • Missing values • Multivariate normality • Sample size OR54 in Edinburgh 3-6 Sept 2012

  10. Using AMOS 16, modelling was carried out by methodically building and testing confirmatory factor analysis (sub)models before graduating to higher order structural regression (hybrid) models (Anderson and Gerbing, 1988). OR54 in Edinburgh 3-6 Sept 2012

  11. Proposed model OR54 in Edinburgh 3-6 Sept 2012

  12. Reliability • The research instrument was tested for reliability (internal consistency) using Cronbach’sato confirm the adequacy of measures for testing research hypotheses. According to a Hinton et al (2004) for • excellent reliability α ≥ 0.9 • high reliability 0.7 ≤ α < 0.9 • moderate reliability 0.5 ≤ α < 0.7 and • low reliability α < 0.5. OR54 in Edinburgh 3-6 Sept 2012

  13. Construct validity To assess construct validity, a factor analysis was conducted using PCA as an extraction method with Varimax and Kaiser normalisation as a rotation method. OR54 in Edinburgh 3-6 Sept 2012

  14. In general, variables loaded on each factor as anticipated and satisfied the conditions of construct validity both in terms of • discriminant validity (loadings of at least 0.4 and only one cross-loading slightly above |0.4| in the case of the Unit submodel) and • convergent validity (eigenvalues of at least 1, loadings of at least 0.4, items that load on to posited constructs). • Thus the validity of our data and findings was confirmed. OR54 in Edinburgh 3-6 Sept 2012

  15. SEM Modelling results • Following a parallel approach to Winch et al. (2000) submodels for Workplace Dynamics were first formulated and tested at: • Unit Level • Task Level and • Individual Level • Next, submodels for Ideal Job Perceptions and Culture were developed. OR54 in Edinburgh 3-6 Sept 2012

  16. Workplace Dynamics: 1. Unit Level • Based on an exploratory factor analysis, the model was specified as in Figure 1a). • Relevant Cronbacha values for the factors here were as follows: • FactoraReliability • Conflict resolution 0.550 Moderate • Unit cohesion 0.639 Moderate OR54 in Edinburgh 3-6 Sept 2012

  17. Figure 1 a) OR54 in Edinburgh 3-6 Sept 2012

  18. Though the latter model was found to fit the data well, as a refinement, it was decided to allow the loading of the item ‘members of unit encourage excellence’ on to the Conflict resolution latent construct. (The argument being that unit members cooperating to achieve excellence would also be well-disposed to resolving conflicts.) OR54 in Edinburgh 3-6 Sept 2012

  19. Fit results for the revised model were as follows: ThresholdAcceptability CMIN/DF = 1.818 < 1-2 Acceptable CFI = 0.953 > 0.9 Acceptable RMSEA = 0.063 < 0.08 Acceptable • with standardised estimates for the model summarised in Figure 1b). OR54 in Edinburgh 3-6 Sept 2012

  20. Figure 1 b) OR54 in Edinburgh 3-6 Sept 2012

  21. All loadings here were found to be statistically significant (p<0.05) and all relationships in the expected direction. OR54 in Edinburgh 3-6 Sept 2012

  22. 2. Task Level • The CFA model in Figure 2a) captures the task-related dimensions in terms of the Control and coordination achieved at work and the Autonomy provided to employees in performing their job. • Relevant Cronbach a values for the factors here were as follows: • FactoraReliability • Work control 0.734 High • Work autonomy 0.819 High OR54 in Edinburgh 3-6 Sept 2012

  23. Figure 2 a) OR54 in Edinburgh 3-6 Sept 2012

  24. Following encouraging results from fitting the above model to the data it was decided the error terms related to the indicators ‘authority in establishing procedures’ and ‘authority in establishing work exceptions’ could be allowed to be correlated since both items are concerned with the process flow while the rest relate to the actual task, quantum of work and work speed respectively. OR54 in Edinburgh 3-6 Sept 2012

  25. Allowing for this refinement, fit details were: CMIN/DF = 1.660 CFI = 0.968 RMSEA = 0.057 all of which were judged to be acceptable. • Standardised estimates are as summarised in Figure 2b). OR54 in Edinburgh 3-6 Sept 2012

  26. Figure 2 b) OR54 in Edinburgh 3-6 Sept 2012

  27. All indicator estimates are statistically significant at the 5% level with the arrow directions too along expected lines. OR54 in Edinburgh 3-6 Sept 2012

  28. 3. Individual Level • The indicator variables analysed reflect individual behaviour and feelings at the workplace. The model specified is shown in Figure 3a). • Relevant Cronbacha values for the factors here were as follows: • FactoraReliability • Job satisfaction -0.525 Low • Instrumental motivation 0.814 High • Feedback motivation 0.815 High • Job involvement 0.602 Moderate OR54 in Edinburgh 3-6 Sept 2012

  29. Figure 3 a) OR54 in Edinburgh 3-6 Sept 2012

  30. In the next step, the model was marginally modified by allowing ‘motivation from promise of promotion’ to load on Feedback motivation in addition to Instrumental motivation’ This is a realistic assumption as the specified item has the characteristics of being both related to feedback and the material aspect of promotion. OR54 in Edinburgh 3-6 Sept 2012

  31. Fit details were: • CMIN/DF = 1.603 CFI = 0.967 RMSEA = 0.054 • all of which were judged to be acceptable. • Standardised estimates are as summarised in Figure 3b). OR54 in Edinburgh 3-6 Sept 2012

  32. Model 3b) OR54 in Edinburgh 3-6 Sept 2012

  33. Model 3b) • All regression weights for the indicators are statistically significant, high in magnitude and in the hypothesised direction. OR54 in Edinburgh 3-6 Sept 2012

  34. Ideal Job Perceptions • The model in Figure 4 – which takes the form of a second order CFA - reflects the individual preferences and perceptions of an ideal job. • Relevant Cronbacha values for the factors here were as follows: • FactoraReliability • Work relations 0.477 Low • Job content 0.544 Moderate • External factors 0.607 Moderate OR54 in Edinburgh 3-6 Sept 2012

  35. Figure 4a) OR54 in Edinburgh 3-6 Sept 2012

  36. Note that to aid identification of the model (Byrne, 2001), the residual terms R1, R2 and R3 were assumed to have the same variance • Fit details were as follows: • CMIN/DF = 1.818 CFI = 0.954 RMSEA = 0.052 • all of which were judged to be acceptable. • Standardised estimates are shown in Figure 4b). OR54 in Edinburgh 3-6 Sept 2012

  37. Figure 4b) OR54 in Edinburgh 3-6 Sept 2012

  38. The latter model makes it possible to infer the strength of relationship between the first-order and second-order factor. The latent constructs – job content (0.89) and external factors (0.92) are strong measures of ideal job preferences followed by work relations (0.65) with all three being statistically significant. Additionally, all indicators load well on to the respective latent variables and in expected directions. OR54 in Edinburgh 3-6 Sept 2012

  39. Culture • This model - based on three of Hofstede’s cultural dimensions – Collectivism, Uncertainty Avoidance and Power Distance - is displayed in Figure 5a). • Relevant Cronbacha values for the factors here were as follows: • FactoraReliability • Collectivism 0.305 Low • Uncertainty avoidance 0.624 Moderate • Power distance 0.459 Low OR54 in Edinburgh 3-6 Sept 2012

  40. Figure 5a) OR54 in Edinburgh 3-6 Sept 2012

  41. As with the previous model, the residual terms R1, R2 and R3 were assumed to have the same variance. • Model fit results were as follows:   • CMIN/DF = 1.600 CFI = 0.876 RMSEA = 0.054 • which were judged to be reasonably acceptable. • Standardised estimates for the model are shown in Figure 5b). OR54 in Edinburgh 3-6 Sept 2012

  42. Figure 5b) OR54 in Edinburgh 3-6 Sept 2012

  43. All indicators show sufficient loadings on their respective first order latent variables. The directions are also as hypothesised earlier. The latent constructs uncertainty avoidance (0.97) and collectivism (0.89) proved to be statistically significant (p<0.05) and hence, represent strong measures of culture. However, power distance (-0.16) is a borderline case in terms of its statistical significance (p<0.1). OR54 in Edinburgh 3-6 Sept 2012

  44. Relationships between Culture, Ideal Job Perceptions and Workplace Dynamics • Summarising hypothesis test relationships for the above, we have results as follows: • Standardised • Expected regression • sign weight Conclusion • ___________________________________________________________________ • H1: Culture → Ideal Job Perceptions + 0.158* Supported • Unit level • H2: Culture → Conflict resolution - -0.103 Not supported • H3: Culture → Unit cohesion + 0.071 Not supported • H4: Ideal Job Perceptions→ Conflict resolution - -0.123 Not supported • H5: Ideal Job Perceptions→ Unit cohesion + 0.000 Not supported OR54 in Edinburgh 3-6 Sept 2012

  45. Standardised • Expected regression • sign weight Conclusion • ___________________________________________________________________ • Task level • H6: Culture → Work control + 0.155* Supported • H7: Culture → Work autonomy - -0.074 Not supported • H8: Ideal Job Perceptions→ Work control + 0.174* Supported • H9: Ideal Job Perceptions→ Work autonomy - -0.079 Not supported OR54 in Edinburgh 3-6 Sept 2012

  46. Standardised • Expected regression • sign weight Conclusion • ___________________________________________________________________ • Individual level • H10: Culture → Job satisfaction - -0.162* Supported • H11: Culture → Instrumental motivation + 0.205** Supported • H12: Culture → Feedback motivation + 0.130 Not supported • H13: Culture → Job involvement + 0.174* Not supported • H14: Ideal Job Perceptions→ Job satisfaction + 0.249** Supported • H15: Ideal Job Perceptions→ Instrumental motivation + 0.075 Not supported H16: Ideal Job Perceptions→ Feedback motivation + 0.182* Supported • H17: Ideal Job Perceptions→ Job involvement - -0.361** Supported OR54 in Edinburgh 3-6 Sept 2012

  47. MULTIPLE GROUP ANALYSIS • The final stage of the analysis involved tested the preceding (sub)models for invariance between the British and French groups. This was done by simultaneously fitting and estimating the (sub)model for each of the two groups and comparing the results. • A sequential, increasingly restrictive procedure of applying cross-group equality was employed. Of particular interest were: • 1) factor loading paths, • 2) factor variances / covariances and • 3) structural regression paths. OR54 in Edinburgh 3-6 Sept 2012

  48. For the testing, equal (forced) unstandardised parameter estimates were first derived for the two groups. • Next, the fit of the constrained model was compared to that of the unconstrained, baseline model. • Where the fit as signified by the chi-square statistic was found to be significantly worse, group variance between the British and the French was concluded. OR54 in Edinburgh 3-6 Sept 2012

  49. Table 1: Multiple Group Invariance Analysis (Unit level) OR54 in Edinburgh 3-6 Sept 2012

  50. As can be seen, the factor variance of conflict resolution differs significantly across the British and French groups and is the cause of overall non-invariance at the unit-level. OR54 in Edinburgh 3-6 Sept 2012

More Related