Chapter III: Methodology and Study Design
Step 1. Item Development Process
Moreover, Abell, Springer, and Kamata (2009) considered the careful attention required from different lines of evidence in examining construct validity, stating: “To the extent that these tests are well anticipated by scale developers, the relevant analyses will build toward conclusions supporting or refuting the accuracy of the measure” (p. 130).
A thorough literature review was initially conducted searching for gaps in the literature to determine how other scholars approached measurement of network ambidexterity. Furthermore, the literature review included broader paradigms to extend the boundaries of the construct of interest (Clark & Watson, 1995). While the broad swath of literature examined cast an initial wide net, thinking along abstract levels of the construct of interest helped crystalize the theoretical model’s development.
The rationale advanced by Abell et al. (2009) was of significant value in making this decision to conduct an exploratory factor analysis; they argue:
The type of factor analysis is referred to as exploratory, because (a) the model formation ( i.e., the number of latent variables) are explored, rather than specified by the intention or theory in the test/scale construction process and (b) factor structure is explored by
modeling each item as a function of all common factors, rather than as a function of only a subset of the factors, to see which factor has a strong relationship with the item and which factor does not. (pp. 133–134)
This methodological approach was of importance to the development of this scale for two reasons. First, it would uncover scale items that would not correlate strongly to the proposed indicator and second, construct validity evidence would be further supported through the unpredictable way test items can reveal an unexpected factor. Another specification in the scale development process was the consideration of how many factors/indicators I would include in my analysis. Edmundson and McManus (2007) suggest that, “methodological fit is created through an iterative learning process that requires a mindset in which feedback, rethinking and revising are embraced as valued activities” (p. 156). The overarching goal of this research was to examine the survey so that it served its intended purpose prior to it being administered. Abell, et al. (2009) cite two reasons why EFA is useful in scale construction: “[F]irst, it may identify test items that are not correlated to an intended common factor. Second it may uncover an
unexpected factor structure of test items” (p. 34). Figure 3.2 illustrates the multidimensional model used in this research.
Figure 3.2. Multidimensional model for survey item specification.
In developing an item pool for this scale, Clark and Watson (1995) argue, “the
fundamental goal at this stage is to sample systematically all content that is potentially relevant to the target construct” (p. 5). Generating the scale items required iteratively weaving in and out of the literature and the item development process in a process that is representative of a funnel’s shape. The initial entry point in the item development process began broadly with a large group of items that were reduced as they were found to be less applicable to the primary construct of interest—RNA. Abell et al. (2009) argue that when developing a large item pool, “either way the goal is to achieve an item pool that is larger than ultimately desired but small enough to be reasonable for administration in large-sample validation” (p. 43). This also meant reaching a level of theoretical saturation where the myriad ways of expressing a construct have been exhausted.
Following this process, four tentative indicators emerged: knowledge sharing,
collaboration, network cohesion, and creativity. This process considered the implications cited by Clark and Watson (1995) regarding initial item pool generation: “[Items] should be broader and more comprehensive than one’s own theoretical view of the target construct and should include content that will be ultimately shown to be tangential or even unrelated to the core construct” (p. 311). Items were based on the notion that while they were understood to be independent, collectively they related broadly to resilient network ambidexterity. Ultimately, 10 to 15 items were created for each of the tentative indicators representing the construct of interest for a total of 52 initial items.
An important concern in deciding what to measure for scale developers is in defining terms that can be considered outcomes, characteristics, or processes (Abell et al., 2009, p. 16). I understood that, theoretically, that the features of RNA stemming from the interplay among the paradigms of complex adaptive systems, were adaptive capacity, resilience, and social networks.
The constructs measured were defined as the following:
• Knowledge sharing: the process of imbuing the organization with information and knowledge for optimization and exploration strategies;
• Collaboration: the process of connecting tasks to performance regardless of where they are situated in the network (micro level, meso level, macro level);
• Creativity: the process behaviors spurred by ongoing novelty, regardless of where they occur along the parallel continuum of optimization and exploration;
• Network cohesion: the process characterized by shared values and vision within the internal boundaries of the network structure.
Upon closer examination, it became clear that knowledge sharing, and collaboration were overlapping notions of the same process I was attempting to measure. Isolating the resources of knowledge sharing, independent of collaboration, and vice versa, brought the realization that they are both inextricably linked. As such, I removed collaboration as an independent indicator and kept knowledge sharing as a theoretical indicator with sufficient explanatory power to define the process of connecting tasks to performance and imbuing the organization with meaningful information. Table 3.1 offers insights into the revised definitions of each of the RNA indicators, along with a brief explanation of what each is intended to measure in an organization.
Table 3.1 think and foster novelty through interactions.
The level of centralized or decentralized novelty that exists in an organization.
An organization
Knowledge sharing: Capacity of the organization to share information across organizational business units for learning and ongoing viability.
The level to which the organization shares useful commitment to values and mission.
The degree to which the organization encourages members to share values and mission.
An organization
Step 2. Validity assessments. This study followed Clark and Watson’s (1995) logic