![[ Visit Client Website ]](images/banner.gif)
Methods: The measurement model for the new questionnaire was developed based on Nahapiet and Ghoshal’s (1998) theoretical framework. A hierarchical factor structure was proposed with a first-order social capital factor consisting of 3 second-order factors (structural, relational, and cognitive social capital), each with 3 third-order factors (for a total of 9 subscales). Structural social capital consisted of three subscales: network size, network functional diversity, and perceived social status. Relational social capital was measured using three subscales: trust, affective energy, and the norm of positive reciprocity. Cognitive social capital was measured using three subscales; cognitive common ground, shared language, and shared narratives. The questionnaire was developed using relevant items from previously validated scales and new items generated by the author. Content validity of the proposed instrument was assessed using the procedures outlined by Polit and Beck (2012). In March, 2015, a panel of 16 registered nurses with a wide range of clinical expertise were given the definition of each subscale and asked to rate each the relevance of each item in the questionnaire on a scale from 1 = not relevant to 4 = very relevant. A cross-sectional survey design using the tailored design method (Dillman, Smyth, & Christian, 2009) was used to evaluate the psychometric properties and factor structure of the new questionnaire. A sample of 1000 Registered Nurses working in hospitals across Ontario, Canada were invited to participate in the study (July-Sept 2015). A total of 249 useable surveys were returned, while 16 were returned undeliverable (26.83% response rate, excluding undelivered surveys). Descriptive statistics were conducted using SPSS (IBM, 2014, version 23.0). The measurement model, item factor loadings, and scale reliability was assessed using confirmatory factor analysis with robust maximum likelihood estimation in Mplus (Muthén & Muthén, 2012).
Results: Content validity results showed that the items in the new questionnaire had acceptable face validity, with CVI scores ranging from .75 to 1.0. The nurses also stated that the questionnaire was easy to complete and an appropriate length. Suggestions about question wording were made for some of the items such as removing qualifiers from items. Changes were made to simplify the questions. The first model was not a good fit for the data: χ²(769) = 1849.311, p = .000; CFI = .783; TLI = .769; RMSEA = .077; SRMR = .083. The item factor loadings revealed that the following items were not strongly related to their respective latent factors: STAT5 (.217), LANG4 (.296), LANG5 (.113), LANG6 (.098). In addition, at the second-order factor level, network functional diversity and network size had weak factor loadings on structural social capital (.164 and .137, respectively). Reassessing the items and the nature of the questions, it was decided to remove these items from the measurement model. Confirmatory factor analysis results showed that the final measurement model for the questionnaire was an adequate fit to the data: χ²(544) = 1043.237, p = .000; CFI = .882; TLI = .871; RMSEA = .063; SRMR = .066. Item factor loadings were generally high (>.70) but ranged from .36 to .94, depending on the subscale. Total social capital had a Cronbach’s α of .94, composite reliability (CR) of .97, and average variance explained (AVE) was .53. These values are above the recommended cut-off values of .70, .50, and .70, respectively (West, Finch, & Curran, 1995). Second-order factors: Structural social capital, represented by status, had a Cronbach’s α of .73, CR of .69, and AVE was .38. Low CR and AVE likely reflects the way the items were worded because of the four items retained in the subscale, two items referred to co-workers and two items referred to physicians. Relational social capital demonstrated strong internal consistency with a Cronbach’s α of .93, CR of .96, and AVE was .63. Cognitive social capital demonstrated strong internal consistency with a Cronbach’s α of .89, CR of .93, and AVE was .46. First-order factors: Trust demonstrated strong internal consistency with a Cronbach’s α of .88, CR of .88, and AVE was .60. Similar results were found for the norm of positive reciprocity (Cronbach’s α = .88; CR = .88; AVE = .55) and shared energy (Cronbach’s α = .94; CR = .94; AVE = .75). Cognitive social capital first-order factors also demonstrated good reliability. Cognitive common ground had a Cronbach’s α of .86, CR of .86, and AVE was .50. Shared language had a Cronbach’s α of .79, CR of .74, and AVE was .49. Finally, shared narratives had a Cronbach’s α of .82, CR of .81, and AVE was .41.
Conclusion: Social capital represents a set of valuable social resources that have been identified as an important asset for healthcare organizations by healthcare leaders (DiCicco-Bloom et al., 2007) and requires more empirical support. The development of the WSCQ-N was needed to provide researchers with a valid and reliable method to measure nurses’ social capital in the workplace. Overall, the findings provide initial support for the final version of the WSCQ-N as a valid and reliable self-report measure to assess nurses’ perceptions of social capital in Canadian hospital settings. Reliability estimates suggest that while the overall scale is reliable, some revisions to the items may strengthen the scale. More research is needed to confirm the validity of the questionnaire and to examine how nurses’ workplace social capital is related to other concepts (i.e. its nomological network), and to examine its application to other work settings.