RESEARCH METHODOLOGY
3.9 CROSS-VALIDATION OF THE ERM VALUES CONSTRUCTS .1 Introduction
As indicated in earlier sections of this methodology chapter, in the process of developing a measurement scale and resulting constructs, an independent variable is required to test and cross-validate the dependent variables – in the case of this study the ERMVs constructs. According to MacKenzie et al. (2011), it is important to consider the interlocking system of laws which constitute theory – these are referred to as the “nomological network”. One of the best ways of testing the nomological network is to determine how indicators of the focal construct relate to measures of other similar constructs and what the lawful relationships between the focal construct and related established constructs are. Hinkin (1998) proposed to administer the new scale (in the case of this study the ERMVS) along with similar established measures to examine the nomological network, i.e. the relationship between existing measures from the literature and the newly-developed scales.
In their seminal work, Bagozzi and Edwards (1998:129) in fact insisted that “one must consider the relationship of the concept under investigation to other concepts in an overall context of a theoretical structure”. MacKenzie et al. (2011) elaborated on this point by outlining that the other constructs to be investigated in the context of the focal construct could be antecedents, correlates or consequences of the identified construct. According to Hinkin (1998), in order to test the external construct validity, the researcher can utilise convergent validity tests (assessing correlation with similar measures) or criterion-related validity, namely variables hypothesised to correlate with the constructs or variables being developed. As proposed by Morgado et al. (2017), incorporating other objective or independent measures supplements the subjective evaluation of variables (i.e. bias) that is possible in self-reporting quantitative studies, and furthermore improves interpretation of the findings.
In the case of this study, as was discussed throughout Chapter 2, and specifically in Section 2.8,, the culture values instruments of Hofstede (2003) and GLOBE (Javidan, Dorfman, De Luque &
House, 2006) are ideally aligned to test convergent and criterion-related validity. Hofstede (2003) and GLOBE (Javidan et al., 2006) promulgated similar items and measurement criteria to those of
the ERM values scale and construct development. Most specifically, they requested respondents to rate values on a Likert scale. Furthermore, there is a clear nomological (theoretical) link between the proposed ERMVS and the culture dimensions selected for cross-validation, as provided by the hypotheses (exhibited in Appendix I) presented for each culture dimension’s possible relationship with ERM values constructs, for example, the culture dimension of Uncertainty Avoidance (UAI) being associated with rules-based ERM values. These Hofstede (2003) and GLOBE (Javidan et al., 2006) culture dimensions therefore presented themselves as ideal candidates for cross-validation, i.e. correlation testing with the ERM values constructs.
Within this study, culture values dimensions were thus defined by NC scales both at an aggregated level of analysis, and also broken down into sub-groups for cross-validation with the ERMVs constructs. GLOBE (Javidan et al., 2006) was the first of the major culture dimensions research projects to define sub-groups, for example, the South African sample is split into ‘Black’ and
‘White’. Investigating culture values at sub-group level is an important, recent research direction in the study of culture. This study aims to contribute to this area of research, by having split the sub-groups further and reporting on the findings of validating the culture values dimensions in the selected samples.
Building from this starting point of analysis, culture was broken down into various cultural constructs or dimensions derived from the literature and hypothesised to exhibit a relationship with ERM values constructs. Examples of culture dimensions tested in this study are Power Distance (PDI), Uncertainty Avoidance (UAI) and Future Orientation (FO). Individual responses were aggregated to a group level by means of indices to reach a score for each cultural construct.
Replicating the Hofstede (2003) and GLOBE (Javidan et al., 2006) instruments directly from the literature, each individual survey questionnaire respondent addressed a number of questions related to the various cultural dimensions, demographics and ERM actual and desired behaviour.
This enabled a number of levels of analysis, such as organisational and functional.
The hypotheses developed around the effect of each of the 11 culture dimensions from GLOBE (Javidan et al., 2006) and Hofstede (2003) selected for the study were discussed previously in Chapter 2, and are exhibited in Appendix I. The analysis featured simple regression tests to determine whether there were statistically-significant relationships between the constructs/
variables (correlation). The findings are presented in Chapter 4.
3.9.2 Hofstede dimensions replication
As outlined in Chapter 2, with regards to NC values, Hofstede, Hofstede and Minkov (2010:18) elaborated on the link between culture and nationality as a grouping:
Within nations that have existed for some time there are strong forces toward further integration: (usually) one dominant national language, common mass media, a national education system, a national army, a national political system, national representation in sports events with a strong symbolic and emotional appeal, a national market for
certain skills, products and services. Today’s nations do not attain the degree of internal homogeneity of the isolated, usually non-literate societies studied by field anthropologists, but they are the source of a considerable amount of common mental programming of their citizens.
The Hofstede Values Survey Module (VSM-13) has been found to be an acceptable measure for NC values in numerous replications in a management sciences context, including Mearns and Yule (2009:783) who had utilised VSM-94 and indicated that “the measure was sensitive enough to differentiate between the national groups on all cultural dimensions”.
The study utilised the VSM-13 and the following Hofstede (2003) culture dimensions were replicated in the main study:
Power Distance (H-PDI);
Uncertainty Avoidance (H-UAI);
Individualism (IDV);
Masculinity (MAS);
Long-term Orientation (LTO); and
Indulgence vs. Restraint (IVR).
Each of the dimensions was tested for reliability and validity within each of the sub-group samples.
Those dimensions found to be reliable and valid for a group were utilised for cross-validation (correlation analysis) with the ERM values constructs.
3.9.3 GLOBE dimensions replication
As discussed in the previous chapter, the GLOBE (Javidan et al., 2006) study is an extension of Hofstede’s (2003) efforts in NC dimension research. GLOBE (Javidan et al., 2006) provided additional culture dimensions that were hypothesis-tested in this study and utilised for cross-validation of the ERM values constructs.
The following GLOBE culture dimensions were replicated in the main study, utilising the GLOBE methodology and survey instrument (Javidan et al., 2006):
Power Distance (G-PDI);
Uncertainty Avoidance (G-UAI);
Performance Orientation (PO);
Future Orientation (FO); and
Collectivism (COL).
Each of the dimensions was tested for reliability and validity within each of the sub-group samples.
Those dimensions found to be reliable and valid for a group were utilised for cross-validation (correlation analysis) with the ERM values constructs.
3.10 THE PRE-STUDY (PILOT) AND THE MAIN STUDY