Sample
Telephone and face-to-face interviews were conducted with representatives from AHLN member organizations as of February 2004. Organizations self-identified a representative who was best qualified to discuss that organization's AHLN-related activities. In general, respondents worked in the area of healthy living, specifically around healthy eating, active living, or tobacco reduction. Their organizational roles included executive directors, managers, and service providers. Different organizational roles can potentially provide different descriptions of an organization's relationships [10]. Three organizations chose to respond in a group format. In these instances, two to four respondents collaborated to provide a single response on behalf of their organization. These organizations were all traditional health organizations, specifically regional health authorities. While group respondents may have potentially identified more partnership ties, we allowed the potential variation to occur so as to maintain a participatory approach in our network mapping endeavor. When we excluded these three organizations in secondary analyses, we observed no significant change in our findings. The survey response rate was 100% (n = 54), although not all questionnaire items were answered.
Formal network analytic methods were used to identify and measure the character and intensity of ties among organizations. For the network modules, organizational representatives referred to the list of AHLN members as they answered questions regarding their relationships with those organizations.
Measures
Dependent Variable
To assess organizational perceptions of the funding environment, organizational representatives were asked to respond based on a four-point Likert scale from strongly disagree to strongly agree to the following statement: "Financial support for your organization's programs and activities in Healthy Eating, Active Living and Tobacco Reduction (HEALTR) is readily available." The question was designed to tap into an organization's perception of whether the funding environment was one in which they felt that they could readily obtain the resources necessary to carry out their programs and activities.
Organizational Attributes
To maintain stable statistical estimates with the small sample size, the number of organizational-attribute variables was kept to a minimum in this analysis. Two organizational characteristics were included in the analysis: i) organizational type and ii) size. Organizational size was based on the total number of employees hired by an organization in either a full- or part-time capacity.
For organizational type, each organization was classified as belonging to either the traditional or non-traditional health sector (Minke S.W. and Simpson T. AHLN Network Mapping: Report on Intersectoral Involvement in the Alberta Healthy Living Network. Unpublished Report, September 2004). The criteria used to classify organizations were established in consultation with key stakeholders in the AHLN, and based on rules around membership, mandates, and action strategies for AHLN organizations (Minke S.W. and Simpson T. AHLN Network Mapping: Report on Intersectoral Involvement in the Alberta Healthy Living Network. Unpublished Report, September 2004). For example, the primary mandate of traditional health-sector organizations was to improve health status. Provincial and federal government health departments, regional health authorities, chronic disease prevention charities, and health professional associations were classified as traditional members of the AHLN (n = 31). These organizations varied in their health promotion activities, with some focusing on primary prevention (e.g., school health coalition), others concentrating on secondary prevention (e.g., chronic disease non-profit), and a few targeting tertiary prevention (e.g., renal program). In contrast, the mandate of non-traditional health-sector organizations did not explicitly include improving health status, although the value of health activities may have been incorporated into their agendas. The organizations deemed to be non-traditional were active living organizations (e.g., fitness centres), education departments (e.g., university, provincial government), recreation and sport organizations, aboriginal organizations, and private businesses (n = 23).
Network Variables
The value of inter-organizational partnership ties within the AHLN network was ascertained by asking organizational representatives the following question for each of the other AHLN members: "Do you have a partnership arrangement with (name of other AHLN member), if so, on a scale of 1–5 where 5 is critically valuable and 1 is marginally valuable, how would you rate your partnership with (name of other AHLN member) to the success of your work in Healthy Eating, Active Living, and Tobacco Reduction (HEALTR)?" The importance of such partnership ties for the overall work of the AHLN was determined through consultation with the Partnership Development and Community Linkages Working Group (PDCLWG) of the AHLN Coordinating Committee.
To measure organizational centrality, we used a Freeman degree measure. In this case, Freeman degree measures the organizations with the most ties to other organizations in the network and has been used in previous studies as an "index of potential communication activity"[11]. In directed networks, we may distinguish between sending, i.e., the partnerships that organization x reports having with others, and receiving ties, i.e., the partnerships that other organizations report having with organization x. Out-degree scores capture an organization's influence; while in-degree scores measure an organizations'popularity and prestige in a network[12]. The distinction is important since a contrast can exist between how organization x values its relationships with other AHLN members and how other organizations value their relationships with organization x. In- and out-degree scores were calculated using UCINET [13].
In addition to measuring organizational positions in the network, we also examined the percentage of tie homophily among traditional and non-traditional organizations. Tie homophily refers in this instance to the idea that traditional organizations may maintain a higher percentage of their network ties with other traditional organizations, and non-traditional organizations may maintain a higher percentage with other non-traditional organizations. Higher percentage tie homophily thus indicates less cross-organizational type diversity in an organization's network ties.
Analysis
Using the statistical package SPSS, the analysis proceeded in three steps. First, we analyzed the distribution of organizational and inter-organizational variables for all AHLN members and then according to traditional or non-traditional organizational type. We examined if significant mean differences in our study variables existed between traditional and non-traditional organizations. Second, we used Pearson and Spearman-rho correlation analyses to examine significant associations among variables. Since there were no significant differences to report between Pearson and Spearman-rho correlation values, Pearson correlation analysis results are reported. Third, we constructed two ordinal logistic regression models. Model 1 regressed organizational-level perceptions of available financial support on two organizational characteristics. In model 2, we added our network measures of in-degree (prestige), out-degree(influence), and tie homophily to model 1. In secondary analyses, we also constructed a network effects model in which we included the term ρWY to adjust for autocorrelation between connected actors[14]. The weighting matrix W was based on i) organizations having equal levels of perceived support and ii) having direct reciprocal partnership ties. The basic premise is that the influence of other organizations' perceptions is strongest when those organizations perceive the same level of support and have reciprocal relationships. The network effects model allows adjustment for the influence of other organizations' perceptions on the ego organization's perception[15]. Secondary analyses found that the network effects term was not significant and did not alter the significance level of any variables in model 2. The regression procedure examines the odds of a variable predicting a higher organizational perception of support, i.e., agreeing or strongly agreeing with the statement that "financial support for your organization's programs and activities in Healthy Eating, Active Living and Tobacco Reduction (HEALTR) is readily available." Results are reported using maximum likelihood estimates.