Exploring the Multidimensionality of Trust in Participatory Health Partnerships - A Network Approach

Abstract

Introduction: Previous studies have identified "trust" as a key mechanism to achieve sustainable partnerships in participatory health research, which themselves can represent social networks. A recent review discussed the potential for social network analysis to investigate the development and maintenance of trust and its effects on partnership functioning in participatory health research partnerships. This review also recommended considering a comprehensive, nuanced and multidimensional approach to conceptualizing, operationalizing and measuring trust in research partnerships. Thus, this study aims to explore empirically the conceptualizing, operationalizing and measuring of trust in a multidimensional manner, approaching each trust dimension as an individual trust network, as well as combined as an overall trust network.

Methods: We sampled the whole network, recruiting from a newly established network of 57 individuals that must collaborate to achieve a common goal. These individuals represented academic, service and community organizations of an existing participatory partnership, the Public and Patient Involvement Ignite Network in Ireland. Of the 57 individuals invited to take part in the study, 75% (n = 43) individuals completed the network survey. A survey about trust was designed based on literature in the area and was administered via Qualtrics. The survey included eight network questions: one on collaboration, and seven on specific dimensions of trust. From this, we constructed a network for each trust dimension. We compared several core network measures of each to identify structural differences between the dimensions of trust. To statistically validate them, we compared them to a random and preferential null model.

Results: All the networks had a high reciprocity but were decentralized. Key differences were identified across trust dimensions, particularly in terms of integrity and shared values, visions and goals. None of the networks compared well to the null models indicating participants did not randomly or preferentially (based on how much trust they receive for a particular trust dimension) trust other partners.

Discussion/conclusion: This novel empirical social network analysis of trust in a real-world partnership elucidates the nuances and multidimensional nature of trust. This provides support for expanding this research direction to enhance understanding of and interventions for trust in participatory health research.

Keywords: community participation; community-based participatory research (CBPR); participatory health research; patient participation (patient engagement); public and patient involvement (PPI); social network analysis; social networking; trust.

Figure 1

Networks for trust dimensions 1 (vulnerability), 5 (shared values, visions and goals) and 6 (power sharing and co-ownership). (A) Network for trust dimension 1-vulnerability. This network was mapped by asking individuals to answer a question pertaining to vulnerability, specifically: “I would discuss with [name of network member X] how I honestly feel about my work, negative feelings and frustrations”. (B) Network for trust dimension 5-shared values, visions and goals. This network was mapped by asking individuals to answer a question pertaining to shared values, visions and goals, specifically: “I feel that [network member X] shares a vision with PPI ignite networks vision and goals?”. (C) Network for trust dimension 6-power sharing and co-ownership. This network was mapped by asking individuals to answer a question pertaining to power sharing and co-ownership, specifically: “I feel that [network member X] is open to discussion* about matters pertaining to the PPI ingnite network”.

Figure 2

The weighted in-degree for each trust question (i.e., dimension of trust), the combined trust network (yellow diamond) and the random (in red with the mean and 2.5 and 97.5 quantiles around it) and preferential (in blue with the mean and 2.5 and 97.5 quantiles around it) null models. Trust network 1 (vulnerability) and 2 (integrity) showed similar in-degree behavior to the preferential model, however the random model did not perform well as expected.

Figure 3

The weighted Freeman centralization about the in-degree for each trust question (i.e., dimension of trust), the combined trust network (yellow diamond) and the random (in red with the mean and 2.5 and 97.5 quantiles around it) and preferential (in blue with the mean and 2.5 and 97.5 quantiles around it) null models. Trust questions 5 and the combined network showed similar behavior to the preferential model, however, the random model yielded low values of centralization.

Figure 4

The reciprocity for each trust question (i.e., dimension of trust), the combined trust network (yellow diamond) and the random (in red with the mean and 2.5 and 97.5 quantiles around it) and preferential (in blue with the mean and 2.5 and 97.5 quantiles around it) null models. All the networks were beyond the 95% confidence interval for each null model.