ASSESS ING PARTNERSHIP DEVELOPMENT UNDER THE MICHIGAN D EPARTMENT OF N ATURAL R S HABITAT GRANTS PROGRAMS USING SOCIAL NETWORK ANALYSIS By Sarah Burton A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Fisheries and Wildlife Master of Science 2019 ABSTRACT ASSESSING PARTNERSHIP DEVELOPMENT UNDER THE MICHIGAN DEPARTMENT MS USING SOCIAL NETWORK ANALYSIS By Sarah Burton In recent years s tate wildlife agencies have begun to realize the need to build collaborations and partnerships among their constituen t s in order to further their conservation goals. This has been done through a variety of avenues including grant programs such as t he Michigan Department of MI DNR) wildlife habitat grant program s , which continue to provide fund s to gover nment, profit and non - profit organizations to develop land for wildlife habitat. Regarding grant programs facilitating p artnerships, there has been no systematic assessment of whether this is a successful method to do so . The goal of this research was to e valuate the effectiveness of the MI DNR grant programs in building relationships. In this work, a s ocial n etwork a nalysi s was conducted to assess the nature of partnerships among gran t receiving and non - grant receiving conservation organizations. A selecti on model approach was used to determine what characteristics were driving the partnerships of this network . The outcome variable being modeled was support received from a partner. The results deliver ed visualization s of the network and insight into why these organizations wer e selecting one another as partners. Major driving forces in partnership selection were found to be grants, the scale of management, having received prior support and distance between organization s . This valuable information will serve as a platform to better understand the networks surrounding wildlife conservation and allow the MI DNR to address any sh ortcomings and gaps within the partnership network. iii ACKNOWLEDGEMENTS I w ish to thank the conservation organizations who participated in th is research . I thank the Michigan Department of Natural Resources, Wildlife Division for providing funding ( award #WLD1804 ) and support throughout the research , especia lly recognizing Clay Buchannan . I would like to thank my committee members, Dan Kramer, Ken Frank and Shawn Riley for their valuable feedback and support to improve the quality of this work. iv TABLE O F CONTENTS LIST OF TABLES ................................ ................................ ................................ ........................ v LIST OF FIGURES ................................ ................................ ................................ ..................... vi INTRODUCTION ................................ ................................ ................................ ......................... 1 Background Information ................................ ................................ ................................ ............. 1 Habitat Grant Programs ................................ ................................ ................................ ............. 4 Significance of the Research ................................ ................................ ................................ ....... 6 METHODS ................................ ................................ ................................ ................................ .... 8 Semi - Structured Interviews ................................ ................................ ................................ ......... 8 Survey Methods ................................ ................................ ................................ ........................... 8 Structural Analysis ................................ ................................ ................................ .................... 10 Selection M odel for S upport ................................ ................................ ................................ ...... 11 RESULTS ................................ ................................ ................................ ................................ .... 17 S emi - Structured Interviews ................................ ................................ ................................ ....... 17 Structural A nalysis ................................ ................................ ................................ .................... 1 8 Selection M ................................ ................................ ................................ ...................... 23 DISCUSSION ................................ ................................ ................................ .............................. 26 CONCLUSIONS ................................ ................................ ................................ ......................... 29 APPENDICES ................................ ................................ ................................ ............................. 31 A PPENDIX A: Interview Protocol ................................ ................................ ............................ 32 A PPENDIX B: Survey Protocol ................................ ................................ ................................ 33 REFERENCES ................................ ................................ ................................ ............................ 41 v LIST O F TABLES Table 1. Descriptive s tatistics of variables included analysis . . . 1 6 Table 2 . Organization types mentioned in survey .. 1 7 Table 3. Model 1 results of selection model by support type (N=1,764 pairs) 3 T able 4. Model 2 results of selection model by support type (N=8,464 pairs) . ................ 5 vi LIST OF F IGURES Figure 1. Geographical extent of WHGP, DHIPI, and Deer PLAN grant recipients across Michigan . Figure 2 . Geographical d istribution of the network 1 9 Figure 3 . Indegree centrality 20 Figure 4 . Betweenness centrality 2 1 Figure 5 . Closeness centrality 2 2 1 INTRODUCTION Background Information With an ever - growing interest in conservation issues and reduced federal and state budgets for natural resource management, agency managers believe partnerships are a n increasin gly necessary t ool to achieve their goals (Rocha & Jacob son, 1998; Bender, 2004) . The emergence of untraditional alliances between public agencies and private organizations in recent decades has allowed for more efficient and effective strategies to addres s conservation concerns ( Robertshaw et al., 1993 ). Not only can p artnerships have a positive impact on the resources being managed, they are also considered an advanced form of public participation thus increasing citizen engagement in the participatory p rocess of management (Arnstein, 2011) . As outlined i to have some level of input into the management process in order to ensure the success and longevity of the relati onships . These partner ship s must also exhibit more d irect involvement and commitment from partners to on - the - groundwork rather than merely being a cooperat ive or contractual agreement (Trauger et al. , 1995). The rise of partnerships can also be a result f rom an increasing recognition of a nee d to address human dimension s of fish and wildlife management. H uman dimension s are described as the process es - making process (Decker & En ck, 199 6 ; Decker et al., 2012 ) . P artnerships with stakeholders enables them to have this direct involvement. F or agencies to make good use of partnerships, agency personnel must first realize the network of organizations that they are working with . Underst anding these networks reveals insight into how these systems operate and where the resources are flowing (Provan et al. , 2005). 2 Social networks can be viewed as a way to develop social capital. In social capital theory, networks are key to the su ccess of an actor in obtaining resources (Burt, 1992; Lin, 2002; Wellman & Frank, 2001) . This theory can be helpful in understanding some of the drivers of partnership development within a network. Constructing and analyzing the network of a community can show not only which actors interact with one another , but also the strength of those interactions , driving forces behind them and th e flow of information between stakeholders . Social network analysis (SNA) is used to understand these networks and can allow manager s to facilitate meaningful partnerships among stakeholders based on resource needs within the community . For wildlife agenc ies, gaining insight into the network of their stakeholder organizations that are partnering with one another is beneficial to unders tanding if the current network structure is even conducive to participation and how best to involve them in the participatory process (Holman, 2008). I be involve d in participatory management can also be identified with SNA a nd can help bridge the gaps between divided segments of the network (Reed et al., 2009). On the side of conserva tion organizations, t his valuable knowledge can provide them access to external resources by becoming more involved with other network members t hus leading to an increase of their social capital and in turn their ability to manage resources effectively (Ba rnes et al. , 2013). S ocial Network Analysis was first conceptualized in the 1930s using sociometry, a technique to represent individuals and th e relationships shared between them (Moreno, 1934). In recent decades, there has been a burst of interest in using SNA in a variety of social science contexts including natural resource management (Crona & Bodin, 2006; Borgatti et al., 2009; Turner et al., 2014). In SNA, actors and their ties with one another are mapped i nto a network using nodes and edges. The nodes, or actors, represent the individuals or organizations involved 3 in the community. These interactions of nodes in a network are represented by the ties, or edges, which can represent relationships or communication between actors (Borgatti & Foster , 2003). Relationships are as equally important as the actors themselves since relationship s define them within the network (Hanneman & Ri ddle , 2005). Several metrics describing the structure of social networks are often ana lyzed . One of the most measured aspects of a network is centrality , which identifies the most influential members of the network. Types of centrality include : degree cent rality , the number of tie s to each node; betweenness centrality : the number of times a node acts as a bridge between two other nodes; and closeness centrality , the average distance to all other nodes . Two fundamental models are used in SNA to investigate t he network behind the structural aspects: the influence model and the selection model (Frank, 2011) . The influence model expresses how the beliefs or actions of an actor can be affected by those with whom they interact. Selection models, on the other hand, aim at understanding how actors choose with whom to interact. For example, Frank et a l. ( 2011 ) describes how people provide help to those with whom they have developed an emotional attachment or those of a certain status with regards to natural resource u se. There are different network mechanisms in selection models that can drive connecti ons such as transactional costs , aspects that may inhibit an interaction (proximity); information seeking, where the goal is to gain new knowledge; or homophily , where ac . Applications of SNA in the natural resource disciplines have increased steadily in recent years (Bodin et al. , 2006; McAllister et al. , 2008; Moore et al. , 2015). SNA can be a useful tool in natural resource management to identify stakeholders, address con flicts, and ensure a diverse representation (Prell et al. , 2009). Moreover, SNA can be especially important when seeking to 4 understand the behaviors of stakeholders. While there have been several studies on influence in so cial networks for natural resource management (Stevens et al. , 2015; Barnes et al. , 2016; Kramer et al. , 2016), fewer have focused on what drives selection. With public participation becoming increasingly embedded in natural resource management, the need to unde rstand the stakeholders inv olved also increases. Over the years there have been many efforts across conservation agencies to encourage partnerships through incentive s or grant programs (e.g. National Fish Habitat Conservation Through Partnerships Ac t, The Regional Conservation Partnership Program, and many state/federal wildlife action plans) but these have not always been targeted towards specific types of partnership s nor have they been sufficiently evaluated to determine effectiveness in partnersh ip building (Bidwel l & Ryan, 2006). Thus far, there has been research linking state wildlife agencies (SWA) to their stakeholders in order to increase public participation ( Chase et al., 2004; Johnson et al., 1993; Lord, 2006) and research on partnerships among agency stakeholders ( Bidwell, 2006 ) but little that has taken both into consideration. In this research I examined the influence that SWA have on conservation organiza tions through means of grant funding and how the knowledge produced from this resea rch could be used to facilitate better participation and increase resource flows. Habitat Grant Programs The MI DNR awards various grants for purpose s of improving and increasing the quantity and quality of wildlife habitat in Michigan and foster lasting p artnerships with conservation organizations (Michigan Department of Natural Resources, Wildlife Division , 2017b). This study focuses on three main wil dlife grant programs which consist of government, profit and non - profit organizations, and individuals . Th ose are: the Wildlife Habitat Grant Program 5 (WHGP), the Deer Habitat Improvement Partnership Initiative (DHIPI), and the Northern Lower Peninsula Deer Habitat Improvement Grant Program (Deer PLAN) (Fig ure 1). Figure 1. Geographical extent of WHGP, DHIPI, and Deer PLAN grant recipients across Michigan The statewide WHGP , established in 2013 , is the largest of the three grant programs with the most funds allocated each year and is funded in part by hunting and fishing licenses (Michigan Department of Natur al Resources, Wildlife Division, 2017b) . T he p riorities developed The points represent an approximate location of the grantees as the centroids of zip code areas matching the database from the US Census (https://www.census.gov/geo/maps - data/data/cbf/cbf_zcta. html). This 6 for the WHGP include increasing opportunities for recreational wildlife use and m aintaining and increasing habitat for waterfowl, upland bird species, small game, and big game. The other two grant programs have similar intentions but differ in the region of the state in which they are administered. The DHIPI is the granting program for the upper peninsula of Michigan, and the Deer PLAN is the granting program in the Northern portion of the lo wer peninsula of Michigan (Michigan Department of Nat ural Resources, Wildlife Division , 2017c and Michigan Department of Natural Resources, Wildlife Division , 2017a). The goal of both plans is to improve and increase habitat specifically for deer on privat e ly owned lands. Public Act 106 of 1971 dedicated $1. 50 of every hunting and fishing license sold to deer habitat improvement and created the Deer Range Improvement Program fund (DRIP). The act has provided the funds for these programs since they began in 2009. The parallels between these three ls in habitat development for game species and their funding schemes are what warranted their inclusion in this study over other MI DNR grants. Significance of the Research The WHGP, DHIPI and Deer PLAN programs aim to assist the MI DNR in 1) developing and improving habitat for wildlife, and 2) partnership development . The first objective targets Goal 2 (manage habitat for sustainable wildlife populations and wildlife - based recreation) of the MI DNR str a tegic plan and is already within the evaluative scope of the various grant administrators (Michigan Department of Natural Resources, Wildlife Division , 2016). However, there has been no systematic assessment of the equally important second objective. The g oal of this research was to evaluate the effectiveness of the MI DNR grant programs in building relationships . Th is research addresse d Goal 5 in the MI to improve and maintain public communication, strong relationsh i ps, and partnerships but more 7 specifically to measure and evaluate the success of existing partnerships (Michigan Department of Natural Resources, Wildlife Division , 2016) . S ocial N etwork A nalyses , among grantee organization s and among all conservation org anizations in Michigan , determine d whether and how the grant programs were strengthening partnerships. I hypothesize d that partnership building within this network of conservation organizations was explained by characteristic s of the organizations involved and was facilitated by the grant programs . Th erefore , the results of this project should enable the MI DNR to improve the administrative and strategic efficacy of these and potential future grant programs to create partnerships through eligibility require ments and application criteria, clarification and prioritization of program goals, and the identification of grant characteristics associated with partnerships. 8 METHODS Semi - Structured Interviews Phone i nterviews were conducted f rom December 2017 t o February 2018 to gain qualitative information on the partnerships formed between grant - receiving organizations. All grant recipients over the last four years were identified by the MI DNR grant administrators . Although there were a few individuals who ap plied for the WHGP and the Deer PLAN, they were the minority of applicants and an even smaller number of them received a grant. In this study, I focused on surveying only the organizations involved in the network. Once the relevant grantees were identified , a group of 20 organizations who received a grant were randomly selected . Of the 20 selected, 16 phone interviews were successfully conducted. The remaining fo u r grantees did not respond to the interview request. The grantees were asked questions aimed at understanding their experience with the grant and their partnerships for habitat management (Appendix A) . T his included asking about how and why they applied, the nature of their relationships with partner ing organizations, and if/ how the grant had affect ed t heir o rganization. Survey M ethods In addition to surveying grant recipients , other Michigan - based conservation organizations that ha d not received or applied to the grant programs , but had potential to do so, were also identified. In order to determin e the boundary of inclusion for organizations as a potential grantee, I used a definitional focus ed search for organizations, which implement ed some restrictions based on specific characteristic s ( Laumann et al., 1989 ; Prell, 2012 ). I sought to find organiz ations that were similar to the organizations that received the grants in order to identify matches a - I 9 collected these additional organization names included through the semi - structured i nterview process , sear ching grantee webpages and discussing potential applicants with MI DNR st aff (Frank, personal communication, Nov. 22, 2017) . These additional organizations allowed me to compare and identify differences with the grant - receiving organi zations. The survey was distributed on June 25 th , 2018 to 2 06 conservation organizations via a weblink sent to the organization with the MI DNR grant coordinators and was available to complete until October 16 th 2018 . Because there were th ree types of respondents: successful applicants, unsuccessful applicants, and non - applicants, each type was sent a different survey version to allow for unique data needs and requests . Each of the three survey s consisted of roughly 3 0 questions regarding o f t 1) attributes and demographics, 2) grant status, 3) relationships with partner ing organizations , 4) perceptions of the MI DNR and 4) leveraged resources as a result of the grant (Appendix B) . All organizations who were sent the survey were put into a roster from which respondents could choose their partners for habitat projects (Butts, 2008) . To understand how these organizations were partnering for habitat management and based on the results of the phone interv iews , I asked respondents to identify the type of support they receive d from each partner. Data are more reliable when survey respondents are asked who they feel supports them rather than asking the organizations who they provide support to (Frank et al. , 2004). There were f our ty pes of support included shar ed equipment, funding, volunteer , and knowledge . The three different survey versions were largely the same but one major distinction was that information was collected on partnership support received pr ior to receiving the gran t in the successful applicant version, but not in the unsuccessful applicant and non - applicant versions. The survey was developed according to the Tailored Design Method to 10 help identify and reduce the four main sources of survey error : coverage, sampling, nonresponse and measurement (Dillman et al. , 2009). Of the 2 06 surveys distributed, I received 113 responses for an overall response rate of 55 % , although 21 of the surveys we re removed due to missing data resulting in a sample size of 9 2 . By respondent type , 4 2 of 50 successful applicants responded ( 8 4 % ); 1 0 of 25 unsuccessful applicants responded ( 4 0 % ) ; and 4 0 of 131 non - applicants responded ( 3 1 % ) . Structural Analysis I analyzed the network of organizations from a structural perspective as well . This analysis allowed me to identify th e organizations that play an important role in resource flow and who might act as brokers : those who can help bridge gaps and bond connections within this network (Bodin et al., 2006) . One of the most measured aspects of a network is centrality , which identifies the most influential members of the network. In a highly centralized network, there are fewer actors that have more con nections than other actors (Bodin et al. , 2006). There are a three main types of centrality : deg ree centrality , betweenness centrality and closeness centrality . Degree centrality is indicated by the number of connections to each node. High degree centralit y scores imply that those actors have increased access to the amount of information and resource s available in the network (Hanneman & Riddle , 2005). Along with calculated overall degree centrality, it can be calculated as in - degree (number of links going to the node) or out - degree (number of links going from the node). In this case, I focus ed on in - degree centrality since I was interested in the selection process. Betweenness centrality i s the number of times an actor l ies along the shortest path between two pairs of actors , normalized by the highest betweenness possible (Borgatti et al., 2009). Betw eenness indicates the potential power in the position of that actor to disrupt, facilitate or distort information flow. Closeness centrality is the average path 11 length between one actor and all other actors in the network. Shorter path lengths allow for ea sier transfer of information between actors (Borgatti, 2005) . Another metric that can be m easured is network density. Density is described as the proportion of potential connections to actual connections in the network . A density score of one indicates th at all possible connections occurred and a score of 0 indicates that the actors in the network are completely disconnected from one another ( Wasserman & Faust , 1994). The three centrality measures and density each bring relevant , unique information that i s important for understanding the stakeholder dynamics involved in this network. All four metrics were calculated for each of the organizations that participated in the survey. Visualizations and calculations of these metrics wer e done using the igraph pac kage in the R software ( R Core Team , 2018 ; Csardi & Nepusz, 2006) . Selection M odel for S upport Using the data collected on organization partnerships, I modeled the effects of various covariates on the likelihood of receiving support. I used each of the fo ur different support types identified : shared equipment, funding, volunteers, and knowledge as dependent variable s ( Support : 0=no, 1=yes) , as well as overall support , indicating whether any support was received, reg ardless of the types. Given that I collec ted different information from each of the three respondent types, two separate selection models were developed . The first pertain ed only to successful applicant s from whom information was collected on partnership support received prior to receiving the gr ant. The framework for m odel 1 is: 12 where represent ed support at time two (after receiving the MI DNR grant) an d can be interpreted as the probability that organization i i.e. the receiver) will provide resources to organization i ( i.e. the sender) (Frank, 2011). Model 1 covariates include d distance of the organizations to one another, support received prior to receiving the MI DNR grant , organization type , the scale of management, and the amount of funds received . Distance was measured in meters as the exact straight - line distance between organization s . Research suggests organizations working in the same region will be more likely to become partners. S horter geographical distance between organizations allows for more in person interactions which facilitates richer exchanges and stronger rel ationship (Torre & Gilly , 2000) . Prior support received indicate d if suppo rt was received prior to receiving the grant, such that prior support equal ed one when received and zero when not received. Investigating prior support helps to determine the nature of the relationship before a certain event (Frank & Fahrbach, 1999) . Diffe rent o rganization type equal ed one when paired organizations wer e of the different type and zero when they we re same . Various types of organizations (non - profit, government, local community - based, etc.) have their own competitive advantages . F or example, w here governments can provide more financial resources, nonprofits tend to work closer with the local communities and thus can provide more social supports (Brinke rhoff , 2002). It may be more beneficial for an organization to partner with those offering 13 dif ferent advantages, which typically would come from a different organizational typ e (Brown & Korten , 1991). The first three covariates : distance , prior support , and different organization type were all pairwise , meaning the values were based on individual pairs of organizations. The last two covariates were modeled as receiver - level effects, meaning that they were considered for individual organizations rather than dependent on a pair. Scale refer red to the geographical scale at which these organizations ma nage land at. It was collected as a categorical vari able with local, regional, statewide, and multi - state being the options. Although little research exists for this type of variable, it was included in the analyses as a way to help account for the effects of high er status levels vs local status levels ( Ber kes, 2002 ). The final variable for this model was fund amount . This variable was measured using the exact dollar amount each organization received divided by $1000. With the interest in evaluating the gra nt programs, I thought adding a variable to this fir st model that account ed for the amount of funds would allow for better understanding of that effects . The second model includ ed all three respondent types , which excludes the prior support variable s ince I only ha d one time point for the unsuccessful applicants and non - applicants. The framework for m odel 2 is: where represent ed supp ort at tim e one and can be interpreted as the probability that the organization i will provide resources to the organization i (Frank, 2011) . 14 Similar to m odel 1, m odel 2 also include d the covariates distance , different organization type , scale, and fund amount but additionally include d the age of the organization . The organization age variable was measured using the exact year th e organization became officially established. It was included more as a control variable to account for variations in the longevity of the organizations. Like scale and fund amount , age was also analyzed as a re ceiver - level effect in model 2. Fund amount was similar to that described above for model 2, however it also included zeros for those organizations that did not receive a grant at all. In this way, the variable represented whether receiving a grant and the amount of the grant influenced being selected as a partner. There are several frameworks of selection models that can be used for analyzing social networks (Carrington et al., 2005). One of the more widely us ed models is the exponential random graph mode l (ERGM) (Lusher et al., 2013). This method utilizes the presence (or absences) of relationships as a means to explain new, emerging relationships. ERGMs emphasize network structures (e.g. reciprocated ties and relationship triangles) while also accounting for individual attributes (Robins et al., 2007). Another suite of network models that are commonly used are latent space models (LSMs). Differently from ERGMs, emerging relationships between actors in LSMs are dependent on the space between their social p ositions in the network (i.e. their latent space), which is measured in terms of Euclidean space (Hoff et al., 2002). The additive and multiplicative effects model (AME) is a more recently developed framework f or social network analysis. These models also use latent space to account for emerging relationships but different from LSMs & ERGMs , AMEs allow one to account for pair - wise interactions as well as sender - specific and receiver - specific factors (Minhas et a l., 2019 ; Snijders , 2011 ). AMEs are built on a generalized linear modeling framework and accounts for several network 15 dependencies including heterogeneity, reciprocity, transitivity and stochastic equivalence (Hoff, 2015). Accounting for such dependencies is limited in ERGMs and LSMs. With its ease to implement, straightforward interpretation of results, and flexibility o f use, the AME model was the most appropriate choice for my analysis. These are logit models where the coefficients are presented as log odds ratios and can be converted to odds ratios for interpretation . I used the R software , version 3.6.0, for statistic al analysis of this data (R Core Team, 201 8 ) and the R package AMEN to analyze the AME selection models (Hoff, 2015). 16 Table 1 . Descriptive s tatistics of variables include d analysis Mean Std. Dev. Min Max Model 1 (N=1,764 pairs ) Outcome Variables Overall Support 0.06 - 0 1 Shared Equipment Support 0.02 - 0 1 Funding Support 0.02 - 0 1 Knowledge Support 0.05 - 0 1 Volunteer Support 0.03 - 0 1 Covariates Prior Support - Overall 0.06 - 0 1 - Shared Equipment 0.01 - 0 1 - Funding 0.02 - 0 1 - Knowledge 0.04 - 0 1 - Volunteer 0.03 - 0 1 Distance (meters) 314.78 192.30 1.38 1031.22 Different Organization Type 0.33 - 0 1 Scale 1.81 - 1 4 Fund Amoun t (in $1,000 increments) 40.77 45.84 2.48 174.44 Model 2 (N= 8 , 464 pairs ) Outcome Variables Overall Support 0.04 - 0 1 Shared Equipment Support 0.01 - 0 1 Funding Support 0.02 - 0 1 Knowledge Support 0.03 - 0 1 Volunteer Support 0.01 - 0 1 Covariates Distance (meters) 302.27 195.08 0.05 1431.40 Different Organization Type 0.30 - 0 1 Scale 2.03 - 1 4 Fund Amount (in $1,000 increments) 18.61 36.72 0 174.44 Age 59.45 27.34 2 115 17 R ESULTS S emi - Structured Interview s The main de cision to apply was needs based since many of these organizations are grant funded and have no steady revenue. For several of the grantees, habitat development is part of their mission, making this grant even more appropriate. When asked how receiving the grant affected their organ ization, most interviewe es responded that the grant allowed them to develop long - term habitat in areas of need. A lso mentioned was that the grants helped to foster partnerships/relationships and encourage volunteers to get involve d. A few of the grantees i ndicated that relationship building with the MI DNR was one of the most successful outcomes of their projects . Table 2 is a list of the types of organizations mentioned as being top partners and the number of interviewees that men tioned each type . The row in red indicate s th e number of organizations that are recipient s of a MI DNR habitat grant , for each type . Aside from gaining information on who the grantees are partnering with, I also learned how they partne r with these organiza tions. As mentioned ab ove, r eceiving some form of support (i.e. shared equipment, funding, volunteers, or knowledge) came across in most of the interviews as why the grantees formed partnerships. Table 2. Organization types mentioned in survey Government Non - profit Sportsmen's Club Conservation Districts Other 6 15 1 1 3 0 6 1 1 0 *The numbers shaded in red = grantees. 18 Structural A nalysis Figure 2 provides a geographical distribution of the network , showing the relative location of each organizati on and the general structure. The density of the network wa s 0.046. This measure, being on a scale of zero to one , wa s very low indicating that the network is disconnected . Although there we re several organizations involved in t he network (9 2 ) many we re pe ripheral and only ha d one or two connections. The values for the mean connections in the different support types (range: 0.01 - 0.06) in table 1 above provide d further evidence for the low density of the network. The average in - degree centrality score was fo ur ties with a range of 0 - 43 (Figure 3) . Organization 67 , a governmental organization, ha d the most ties in the network. The organization with the second highest in - degree central ity score (org. 42) wa also governmental , followed 3 rd and 4 th by non - profit organizations (org. 86 and org. 28) . The average betweenness centrality score was 0.016 (Figure 4), indicating that this network ha d few organizations that provide a bridge among different sections of the network. The organizations with the three highest b etweenness scores we re all non - profit organizations (org. 74, org. 69, and org. 62). The average closeness centrality score was 0.04 9 (Figure 5), indicating that this was a disconnected network. With a range of 0.01 - 0.06, most organizations share d a very s imilar score and no actors st oo d out from the rest as high scorers . 19 Figure 2. Geographical d istribution of the network 20 Figure 3 . Indegree centrality Figure 3 depicts the results of in - degree centrality of the nodes of the network . The larger the size of the node, the higher the in - degree centrality score that organization received. The color s correspond to the various organization types indicated in the legend above. 67 42 86 28 21 Figure 4. Betweenness centrality Figure 4 il lustrates the results of betweenness centrality of the nodes in the network. The larger the size of the node, the higher the betweenness centrality score that organization received. The color s correspond to the various organization types indicat ed in the l egend above. 74 6 9 6 2 22 Figure 5. Closeness centrality Figure 5 illustrates the results of closeness centrality of the nodes in the network. The larger the size of the node, the higher the closeness centrality score that organization received. The col or s correspond to the various organization types indicated in the legend above. 23 Select ion M Results for m odel 1 and model 2 we re reported in unstandardized coefficients (log odds ratios) and standardized coefficients ( standard deviation s) . By looking at the standardized coefficients, comparisons can be made among the covariates since they have been normalized ( Menard, 2011 ). In model 1, the prior support a grantee organization received was statistically significant (p < 0.001) and positive for all support types as well as overall support (Table 3 ) . This prior support was correlated with their current partnerships. Among the different support types, the effect of prior support on the likelihood of selection varies . For example, with overall support, the odds of receivin g support we re 98 . 3 (e 4. 588 ) times hi gher when having received prior support vs not . Compared that to funding support where the odds of receiving support were 13.7 (e 2.618 ) times higher. Table 3. Model 1 results of s election m odel by support type (N=1,764 pairs ) Shared Equipment S upport Funding Support Knowledge Support Volunteer Support Overall Support Unstd coeffs Std. coeffs Unstd coeffs Std. coeffs Unstd coeffs Std. coeffs Unstd coeffs Std. coeffs Unstd coeffs Std. coeffs (SE) (SE) (SE) (SE) (SE) Scale 0.376 * (0.166) 2.272 0.222 (0.167) 1.326 0.296 (0.183) 1.614 0.42 6* (0.190) 2.241 0.129 (0.233) 0.551 Fund Amount 0.006* (0.003) 2.15 0.006* (0.003) 2.311 0.008** (0.003) 2.563 0.004 (0.003) 1.513 0.008* (0.004) 2.379 Prior Support 3.169*** (0.611) 5.186 2.618*** (0.334) 7.839 4.848*** (0.57 8) 8.387 3.110*** (0383) 8.125 4.588*** (0.417) 11.007 Distance - 0.002*** (0.001) - 4.963 - 0.002*** (0.001) - 4.169 - 0.002** (0.001) - 2.52 - 0.002*** (0.001) - 4.023 - 0.001 (0.001) - 1.575 Different Org Type - 0.142 (0.163) - 0.869 - 0.178 (0.1 64) - 1.083 - 0.429* (0.212) - 2.024 - 0.020 (0.176) - 0.112 - 0.380 (0.241) - 1.578 Intercept - 2.333 (0.365) - - 2.182 (0.359) - - 2.946 (0.435) - - 2.594 (0.406) - - 3.092 (0.518) - *p < 0.05; **p < 0.01; ***p < 0.001 24 The other covariates were statistically si gnificant for some of the support types but not all. Distance wa s found to be statistically significant in all support types except for overall support and the effect was negative which indicates that larger geographic distance between organizations decreased the likelihood of partnerships. Different organization type was only found to be statistically significant for knowledge support and on average had a much smaller effect on partnership selection than the other covariates . Looking at volunteer support, the effect of d ifferent organization type ( - 0.112 ) wa s more than 1 3 times smaller than the next most effective covariate, fund amount ( 1.513 ). Scale was found to be statistically significant and positive for shared equipment support and volunteer support. Analyzing the variable further for volunteer support, it showed that an organization is 1.5 times more likely to be selected as a partner with every increase in scale level. The last variable in Model 1 to review wa s fund amount . This variable was statistically significant and positive for all support types accept volu nteer support. When significant, the effect of the variable on the outcome was very similar. In model 2, my main variable of interest , the fund amount receiv ed , was statistically significant (p < 0.05 or p < 0.01) in all models except for shared equipment support . This finding not only suggest s tha t organizations that receive d a grant ha d higher probability of being selected but that the more money those organizations received also increased their probability of being selected as a partner . Along wit h the e ffect of the grants , distance wa s also statistically significant with a negative effect in every support type (p < 0.001). This indicate d that the further away organizations we re from one another, the less likely they we re to partner. The effect size was a bout the same for all support types, for every meter increase in distance , the odds of receiving support decrease d by one (e - 0.002 ). Although this seems small, I was working with a small unit (meters) relative to the overall distances. Another important co variate to note in these 25 results was scale . This covariate was statistically significant and positive for all support types but shared equipment support . This covariate also had the second highest standard coefficient meaning it was the second most influ en tial covariate on the outcome. D ifferent organization type was also included in this model and was only found to be statistically significant (p < 0.05) for overall support. The final covariate included was the organization age . This variable had a positiv e effect and was statistically significant for funding support and knowledge support. The effect size was relatively the same for the different support types . With every one year increase in age, the odds of being selected as a partner was one times higher . Table 4. Model 2 results of selection model by support type (N=8,464 pairs) Shared Equipment Support Funding Support Knowledge Support Volunteer Support Overall Support Unstd coeffs Std. coeffs Unstd coeffs Std. coeffs Unstd coeffs Std . coeffs Unstd coeffs Std. coeffs Unstd coeffs Std. coeffs (SE) (SE) (SE) (SE) (SE) Organization Age 0.005 (0.004) 1.41 0.008* (0.003) 2.385 0.006* (0.003) 2.186 0.002 (0.003) 0.614 0.005 (0.003) 1.803 Scale 0.131 (0.080) 1.643 0.339*** (0.073) 4.662 0.225*** (0.063) 3.568 0.165** (0.061) 2.693 0.248*** (0.059) 4.199 Fund Amount 0.003 (0.002) 1.389 0.005* (0.002) 2.331 0.005** (0.002) 3.039 0.005** (0.002) 3.057 0.005** (0.002) 3.164 Differen t Org Type 0.110 (0.142) 0.773 - 0.085 (0.106) - 0.797 - 0.159 (0.084) - 1.904 - 0.068 (0.115) - 0.593 - 0.192* (0.078) - 2.48 Distance - 0.002*** (0.000) - 4.313 - 0.002*** (0.000) - 6.523 - 0.002*** (0.000) - 9.412 - 0.002*** (0.000) - 6.187 - 0. 002*** (0.000) - 9.778 Intercept - 3.398 (0.341) - - 3.516 (0.327) - 2.640 (0.261) - 2.789 (0.262) - 2.457 (0.245) - *p < 0.05; **p < 0.01; ***p < 0.001 26 D ISCUSSION Our findings from the structural analysis indicated a discon nected network of conservation organizations, given a low - density score, few actors with high in - degree centrality and low closeness centrality scores. However, information gained from this study can be of value to agency managers. Identifying the organiza tions that were selected the most (high in - degree centrality) as important partners show ed who might be the most influential in the network moving forward while highlighting the actors that may need more assistance with making connections. Organizations wi th hig h betweenness centrality we re important because of their ability to act as brokers along main communications paths in the network, allowing them to facilitate the flow of information (Tang et al., 2010). With this knowledge, i nterventions might be im plemen ted to better allocate resources to organizations that are less connected to the network, thus increasing their social capital and facilitating cooperation among organizations (Pretty, 2003). One interesting covariate from the models that helps expl ain th e flow of resources the structure of the network is depicting is scale . In model 1, t his covariate was only statistically significant for shared equipment and volunteer support, both of which are physical resources, and indicated that the larger the scale of an organization the more likely they were to be selected as a partner. Sc ale was meant to distinguish between institutional status levels of the organizations, so this finding could be interpreted that the flow of physical resources through the gr ant receiving network was driven by the high - level organizations whereas nonphysical resources were less tied to the status of the organization. Since these organizations received funds from the MI DNR, it could be that they are less likely to see k additio nal high - level organizations for that resource. Differently than model 1 , model 2 sho ws scale being statistically 27 significant for all support types b ut shared equipment. Since model 2 shifted to include non - grant receiving organizations, the wider signific ance of scale could indicate that without MI DNR funds organizations seek more high - l evel organizations for necessary physical and nonphysical resources ( Edwards , 2001 ) . This study investigate the effects grant s had on partnership selec tion among survey respondents for all support types . The AME selection model indicated that a relationship exist ed between the MI DNR grants and being selected as a partner . Although little research exists that directly links an network with receiv ing grant funding, social capital theory can be used to explain this relation ship. In some interpretations of social capital theory, networks are key to the success of an actor in obtaining necessary resources (Burt, 1992; Lin, 2002; Wellman & Frank, 2001) . The stronger the social network of an organization the more likely they are to succeed in procurement of resources (in this case grants) that otherwise would be unavailable to the m ( Bourdieu & Richardson, 1986 ; Burt, 1992; Lin, 2002). This theory has bee n previously used to explain factors that contribute to making a nonprofit organization successful (Hager et al., 2004) . The authors argue d that successful organizations tend ed to create a more developed network and tie themselves to centers of power which led to such organizations being viewed as legitimate and increases their access to resources. In this study, those organization that received a grant had a stronger network and the more grant funds they receive d increased their network as well. The dista nce covariate is statistically highly significant with a negative effect in both model 1 and model 2 for almost all support types. This finding indicates that the further organizations are to one another the les s likely they are to partner. This effect is in line with much of the literature on geographic distance 28 Where human resources are involved (i.e. volunteer and shared equipment support), being geographically clo se to one another has been found to facilitat e those types resource flows (Vedovello, 1997). These findings suggest that these conservation organizations tend to work locally or regionally and partner with one another in the same way. Although these findings for the prior support, scale, grant s and d istance covariates provide d a strong case for what drives partnership selection in this network, this study show ed some limitations that would need to be addressed i n future research on this topic. The data collected in this study was not truly longitudina l, meaning network actors were not surveyed at two or more different time points to determine if some event (e.g. receiving a grant) had a true l., 2005). Although analyzing longitudinal data has its ow method used in this study. An additional covariate that could have been collected to determine the strength of these relationships was reciprocity, i.e. a mutual exchan ge support (Wasserman & Faust, 1994) . Given time constraints and a lack of knowledge on the full extent of the network, it was difficult to take reciprocity into consideration. 29 C ONCLUSIONS As state wildlife managers look to new ways to achieve their cons ervation goals, partnerships with private organizations such as cooperatives and non - profit organizations have become vital to the success of such agencies (Trauger et al., 1995; Wigley & Sweeney, 1993). Partnerships can result in an increase in public par ticipation from stakeholders which has been shown to be beneficial to wildlife agencies to ensure their management techniques are relevant and build trust with citi zens (Decker et al., 2015). T o facilitate these relationships among conservation organizatio ns, the MI DNR has implemented partnership development as a goal of their habitat grant programs. However, this has not been a requirement for obtaining a grant. In this study, I present ed evidence that grants we re a driving force in partnership selection among the surveyed conservation organizations, along with distance to one anothe r , having received prior support, and the scale at which the organization manage s resources . These findings we re in line with my hypothesis and hold significance in shaping th e future of these grant programs and conservation policy. By understanding the social networks of their stakeholders and partners, state wild life managers and policy makers can be in a better position to engage citizens for decision making for conservation (Prell et al., 2009). Understanding what drives partnerships selection among conservation organization s provides a unique opportunity for m anagers to intervene and incentivize networks that will maximize the potential for success in resource management a nd conservation . High - level organizations were sought out for partnerships within this network and are somewhat common among the grant receiving organization pool , so a strategic on the part of the MI DNR to help facilitate these connections could be to im plement certain requir ements for partnering between different scaled organization s or organization types . This would ensure that the funds th e MI DNR award go to 30 those high - l evel organizations that typically have a reputation for good management practices while still including and assisting the smaller scale organizations in the management process . Future research could capture the changes in the network using our preliminary work and reevaluate the drivers for selection should a similar intervention be imp lemented. 31 A PPENDICES 32 A PPENDIX A : Interview Protocol Interview Protocol Form Project: Evaluating the Habitat Grant Program Date ___________________________ Time ___________________________ Grantee being interviewed ______________________ Release of information? ______________ Notes to interviewee: Thank you for your participation. I believe your input will be valuable to this research and in helping grow all of our professional practice. Confidentiality of res ponses is guaranteed Approximate length of interview: 10 minutes Purpose of research: This research will focus on evaluating the effectiveness of the DNR grant programs and aims to improve the process to allow the program to run successfully. The foll owing questions were formulated to gain useful information that can be used to help guide the development future surveys for this project. 1. How did you come to learn about the Habitat Grant Program? What made you decide to apply? 2. Who are the organizat ions and/or individuals with whom you regularly work? What kind of organizations are they and what kind of work do they do? 3. Do you work with other organizations or individuals when carrying out the tasks to complete your project? How do you work with the m? What are their names? 4. Since receiving the grant, have you soug ht out additional resources to help carry out your habitat project? What resources? (Financial, Volunteer, Shared Equipment, etc.) 5. Does any information on the work you do for your projec t get publicized? If so, through what outlets? (Newsletters, social media, website posts, etc.) 6. What effects, if any, has the grant had on your organization; operationally, values, etc? 7. What do yo u feel was the most successful outcome of the grant pro cess or work? 33 A PPENDIX B: Survey Protocol DNR Habitat Grant Programs Survey Dear Participant , you are invited to contribute to a research project being conducted for a Master's thesis project of a MS U student in the Department of Fisheries and Wildlife. The results of the study will be used by the Michigan Department of Natural Resources to inform the management of their habitat grants. You should feel free to ask the researchers any questions you may have. Study Title : Using Social Network Analysis to Me asure and Assess Relationship Building and Partnerships under the Michigan DNR's Habitat Grants Programs Researcher and Title : Sarah Burton (Graduate Student) & Daniel Kramer (Professor) Department and Institution : Department of Fisheries and Wildlife, Mic higan State University Address and Contact Information : Rm 13 Natural Resource Building, East Lansing, MI 48824; burtons7@msu.edu or dbk@msu.edu; 517 - 614 - 6965 This stud y aims to understand how ms may foster successful partnerships with and perceptions of the DNR. We are asking your organization and others to answer some questions about the type and extent of their interactions with other organizations for habitat management projects. From this s tudy, we hope to understand the success of these grant programs in building successful direct and indirect relationships with the MDNR. Participating in this survey will take about 20 minutes of your time, depending on your answers. You must be at least 18 years old to participate in this research. This survey is completely voluntary and you may choose to end the survey at any point. By continuing with the surv ey, you indicate your voluntary agreement. Confidentiality will be kept and information obtained i n this survey will only be shared with the research team. You are being sent this survey because your organization has received one or more of the fol lowing Michigan Department of Natural Resources grants: Wildlife Habitat Grant Program (WHGP) De er Habitat Improvement Partnership Initiative (DHIPI) Northern Lower Peninsula Deer Private Land Assistance Network (NLP Deer PLAN) 34 Q1 What organization are you representing? ________________________________________________________________ Q2 What yea r(s) did your organization receive the grant(s)? Please separate years by commas. ________________________________________________________________ The following series of questions will provide the necessary information to evaluate the partnerships dev eloped between organizations. You will be asked to select up to 10 of your top par tners that you regularly work with on habitat management projects. Here, a partner is defined as an organization that you receive support from in order to further your organi zation's habitat work. This support can be in the form of volunteers, funding, sha red equipment or knowledge/expertise. The list of organizations have been broken down into 6 different types. A brief description/example of the types are below: Conservatio n Districts : A local government entity that carries out natural resource managemen t programs. In Michigan they are developed at the county levels. Sportsmen's Clubs/Co - Ops : These are local organizations or collectives of individuals who have a passion for hunting, fishing and/or wildlife conservation. Example - Hiawatha Sportsman's Club , Tower Road Area QDM Cooperative, Canada Creek Ranch. Non - Profit Organizations : An organization dedicated to conserving natural resources and/or wildlife habitat; most havi ng official non - profit status. Example: Ducks Unlimited, Great Lakes Commission, C hippewa Nature Center. Federal/State Agencies : A recognized government entity that operates at a federal or state level. For - Profit Organizations : A company that conducts c onservation activities for a profit, sometimes in a consulting role. Example: Gree n Timber Consulting, Upper Michigan Land Management and Wildlife Services, Boyne Outfitters. Miscellaneous : This grouping includes tribes, universities, and allows for entry of other organizations not mentioned in the above categories, such as local municipalities. Q3Below is the list of organization types, as described above. Please indicate the types of organizations your organization regularly partners with. Check all th at apply. ˛| Conservation Districts (1) ˛| Sportsmen's Clubs/Co - Ops (2) ˛| Non - Profit Organizations (3) ˛| Federal/State Agencies (4) ˛| For - Profit Organizations (5) ˛| Miscellaneous (6) ˛| My organization doesn't have any partners. (7) 35 Q4 Please indicate the Conservation Districts you regularly work with from the list below. Sel ect all that apply. To select multiple, hold CTRL while clicking the organizations. Omitted List of Conservation Districts Q5 Please fill out the corresponding columns for each org anization indicated, or write in additional organizations. These questi ons are asking about your organization's relationships in terms of the most recent grant cycle your organization was involved in. The first 3 questions are asking to identify support r eceived from organizations since your organization's most recent grant. The last 2 questions are asking to identify support received from organizations prior /before your organization's most recent grant. Q6. Please indicate the sportsmen's clubs/co - op s you regularly work with from the list below. If not included in the list, you will have the opportunity to add organizations in the following section. Select all that apply. To select multiple, hold CTRL while clicking the organizations. Omitted List of - ops Q7 Please fill out the corresponding columns for each organization indicated, or write in additional organizations. These questions are asking about your organization's relationships in terms of the most recent grant cycle you r organization was involved in. The first 3 questions are asking to identify support received from organizations since your organization's most recent grant. The last 2 questions are asking to identify support received from organizations prior/before your organization's most recent grant. 36 Q8 Please indicate the Non - Profit Organizations you regularly work with from the list below. If not included in the list, you will have the opportunity to add organizations in the following section. Select all that ap ply. To select multiple, hold CTRL while clicking the organizations. Omitted List of Non - Profit Organizations Q9 Please fill out the corresponding columns for each organization indicated, or write in additional organizations. These questions are asking about your organization's relationships in terms of t he most recent grant cycle your organization was involved in. The first 3 questions are asking to identify support received from organizations since your organization's most recent grant. The last 2 que stions are asking to identify support received from or ganizations prior/before your organization's most recent grant. Q10 Please indicate the Federal/State Agencies you regularly work with from the list below. If not included in the list, you will have the opportunity to add organizations in the following section. Select all that apply. To select multiple, hold CTRL while clicking the organizations. Omitted List of Federal/State Agencies Q11 Please fill out the corresponding columns for each organiz ation indicated, or write in additional organizations. These questions are asking about your organization's relationships in terms of the most recent grant cycle your organization was involved in. The first 3 questions are asking to identify support recei ved from organizations since your organi zation's most recent grant. The last 2 questions are asking to identify support received from organizations prior/before your organization's most recent grant. 37 Q12 Please indicate the for - profit organizations you regularly work with from the list below. If not included in the list, you will have the opportunity to add organizations in the following section. Select all that apply. To select multiple, hold CTRL while clicking the organizations. Omitted List of For - Profit Organizations Q13 Please fill out the corresponding columns for each organization indicated, or write in additional organizations. These questions are asking about your organization's relationships in terms of the most recent grant cycle your or ganization was involved in. The fir st 3 questions are asking to identify support received from organizations since your organization's most recent grant. The last 2 questions are asking to identify support received from organizations prior/before your orga nization's most recent grant. Q14 Please indicate the organizations you regularly work with from the list below. If not included in the list, you will have the opportunity to add organizations in the following section. Select all that apply. To select multiple, hold CTRL while clickin g the organizations. Omitted List of Miscellaneous Organizations Q15 Please fill out the corresponding columns for each organization indicated, or write in additional organizations. These questions are asking about you r organization's relationships in terms of the most recent grant cycle your organization was involved in. The first 3 questions are asking to identify support received from organizations since your organization's most recent grant. The last 2 questions are asking to identify support receiv ed from organizations prior/before your organization's most recent grant. 38 Q16 Why did your organization apply for the MI DNR grant? ________________________________________________________________ Q17 Does your orga nization partner with the MI DNR f or other projects, unrelated to the habitat grants? o Yes, please explain what projects. (1) ________________________________________________ o No (2) Q18 How dependent on grants is your organization for habitat develop ment? Click on the circle and drag to the desired interval. Q19 Please indicate on the scale below what percent of your organization's funding comes from the Michigan Department of Natural Resources - Wildlife Division. Q20 Please indicate the agree ment your organization shares with the following statements. 39 Q21 Please indicate your agreement with the statements below in regard to whether the Michigan DNR has had an effect on the following aspects of your organization: Q22 On a scale of 1 - 10, how likely would you be to recommend the DNR - Wildlife Division as a partner? Q23 What year did your organization establish? ________________________________________________________________ Q24 How many employees does your organization have? o 0 (5 ) o 1 - 10 (1) o 11 - 30 (2) o 31 - 50 (3) o 50+ (4) Q25 How many members does your organization have? 40 o 0 (6) o 1 - 100 (1) o 100 - 1,000 (2) o 1,000 - 5,000 (3) o 5,000 - 10,000 (4) o 10,000+ (5) Q26 At what scale does your organization work at? o Local (1) o Regional (2) o Statewide (3) o Multi - state (4) Q27 After receiving the MI DNR grant(s), did your organization seek out additional resources to help complete the work of the grant? o Yes (1) o No (2) Q28 Of the list below, what type of resources di d your organizatio n leverage after receiving the DNR habitat grant and in what amounts? ________________________________________________________________ __________________ ______________________________________________ Q30 Please provide any additional comments regarding the Michigan Department of Natural Resources - Wildlife Division's habitat grant programs. ____________________________________________ __________________ __ ________________________________________________________________ 41 REFERENCES 42 REFERENCES Barnes, M. L., Lynham, J., Kalberg, K., & Leung, P. (2016). Social networks and environmental outcomes. Proceedings of the Na tional Academy of Sciences, 113(23), 6466 - 6471. Barnes - Mauthe, M., Arita, S., Allen, S., Gray, S., & Leung, P. (2013). The influence of ethnic diversity on social network structure in a common - pool resource system: implications for collaborative managemen t. Ecol ogy and Society, 18(1). Bender, K. A. (2004). Evaluating Agency Perceptions of Partnerships: Lessons Learned from the US Fish and Wildlife Service (Doctoral dissertation, University of Florida). Berkes, F. (2002). Cross - scale institutional linkage s: pers pectives from the bottom up. The drama of the commons, 293 - 321. Bidwell, R. D., &; Ryan, C. M. (2006). Collaborative partnership design: the implications of organizational affiliation for watershed partnerships. Society and natural resources, 19(9) , 827 - 843. Bodin, Ö., Crona, B., & Ernstson, H. (2006). Social networks in natural resource management: what is there to learn from a structural perspective?. Ecology and Society, 11(2). Borgatti, S. P., & Foster, P. C. (2003). The network paradigm in o rg anizational research: A review and typology. Journal of management, 29(6), 991 - 1013. Borgatti, S. P. (2005). Centrality and network flow. Social networks, 27(1), 55 - 71. Borgatti, S. P., Mehra, A., Brass, D. J., & Labianca, G. (2009). Network analysis i n the social sciences. science, 323(5916), 892 - 895. partnerships: A study on French data. Papers in Regional Science, 91(2), 355 - 375. Bourdieu, P., & Richardson , J. G. (1986). Handbook of Theory and Research for the Sociology of Education. The forms of capital, 241 - 258. Brinkerhoff, J. M. (2002). Government nonprofit partnership: a defining framework. Public administration and development, 22(1), 19 - 30 . Brown, L. D., & Korten, D. (1991). Working more effectively with NGOs. In IDR Reports (Vol. 8, No. 1). Institute for Development Research. Burt, R. S. (2009). Structural holes: The social structure of competition. Harvard university press. 43 Butts, C. T . (2008). Social network analysis: A methodological introduction. Asian Journal of Social Psychology, 11(1), 13 - 41. Carrington, P. J., Scott, J., & Wasserman, S. (Eds.). (2005). Models and methods in social network analysis (Vol. 28). Cambridge university press. C hase, L. C., Decker, D. J., & Lauber, T. B. (2004). Public participation in wildlife management: what do stakeholders want?. Society and Natural resources, 17(7), 629 - 639. Crona, B., & Bodin, Ö. (2006). What you know is who you know? Communicat ion patter ns among resource users as a prerequisite for co - management. Ecology and society, 11(2). Csardi, G., & Nepusz, T. (2006). The igraph software package for complex network research. InterJournal, Complex Systems, 1695(5), 1 - 9. Decker, D. J., & E nck, J. W. (1996). Human dimensions of wildlife management: knowledge for agency survival in the 21st century. Human Dimensions of Wildlife, 1(2), 60 - 71. Decker, D. J., Riley, S. J., & Siemer , W. F. (Eds.). (2012). Human dimensions of wildlife management. JHU Press . Decker, D. J., Forstchen, A. B., Pomeranz, E. F., Smith, C. A., Riley, S. J., Jacobson, C. A., Organ , J. F., & Batcheller, G. R. (2015). Stakeholder engagement in wildlife manage ment: Does the public trust doctrine imply limits?. The Journal of Wildlif e Management, 79(2), 174 - 179. Dillman, D. A., Smyth, J. D., & Christian, L. M. Internet, mail, and mixed - mode surveys: the tailored design method. 2009, Hoboken, N.J: Wiley & Sons. Edwards, B., Goodwin, M., Pemberton, S., & Woods, M. (2001). Partnership s, power, and scale in rural governance. Environment and Planning C: Government and Policy, 19(2), 289 - 310. Frank, K. A., & Fahrbach, K. (1999). Organization culture as a complex system: Balance and information in models of influence and selection. Organi zation Science, 10(3), 253 - 277. Frank, K. A., Zhao, Y., & Borman, K. (2004). Social capital and the diffusion of innovations within organizations: The case of computer tec hnology in schools. Sociology of Education, 77(2), 148 - 171. Frank, K. A. (2011). So cial network models for natural resource use and extraction. Social networks and natural resource management: Uncovering the social fabric of environmental governance, 180 - 205. 44 Frank, K. A., Maroulis, S. P. I. R. O., Belman, D. A. L. E., & Kaplowitz, M. D . (2011). The social embeddedness of natural resource extraction and use in small fishing communities. Sustainable fisheries: Multi - level approaches to a global problem, 30 9 - 331. Graddy, E. A., & Chen, B. (2009). Partner selection and the effectiveness of interorganizational collaborations. The collaborative public manager: New ideas for the twenty - first century, 53 - 70. Hager, M. A., Galaskiewicz, J., & Larson, J. A. (2004 ). Structural embeddedness and the liability of newness among nonprofit organization s. Public Management Review, 6(2), 159 - 188. Hanneman, R. A., & Riddle, M. (2005). Introduction to social network methods. Hoff, P. D., Raftery, A. E., & Handcock, M. S. ( 2002). Latent space approaches to social network analysis. Journal of the american S tatistical association, 97(460), 1090 - 1098. Hoff, P. D. (2015). Dyadic data analysis with amen. arXiv preprint arXiv:1506.08237 Holman, N. (2008). Community participation: using social network analysis to improve developmental benefits. Environment and P lanning C: Government and Policy, 26(3), 525 - 543. Johnson, K. N., Johnson, R. L., Edwards, D. K., & Wheaton, C. A. (1993). Public participation in wildlife management: Opinions from public meetings and random surveys. Wildlife Society Bulletin (1973 - 2006) , 21(3), 218 - 225. Kramer, D. B., Mitterling, A., & Frank, K. A. (2016). Understanding peer influence in hunter harvest decisions using social network theory and analysis. Human Dimensions of Wildlife, 21(5), 414 - 426. Laumann, E. O., Marsden, P. V., & Pr ensky, D. (1989). The boundary specification problem in network analysis. Research methods in social network analysis, 61, 87. Lin, N. (2002). Social capital: A theory of social structure and action (Vol. 19). Cambridge university press. Lord, J. K., & Cheng, A. S. (2006). Public involvement in state fish and wildlife agencies in the US: a thumbnail sketch of techniques and barri ers. Human Dimensions of Wildlife, 11(1), 55 - 69. Lusher, D., Koskinen, J., & Robins, G. (Eds.). (2013). Exponential random gr aph models for social networks: Theory, methods, and applications. Cambridge University Press. McAllister, R. R., Cheers, B., Da rbas, T., Davies, J., Richards, C., Robinson, C. J., ... & Maru, Y. T. (2008). Social networks in arid Australia: a review of c oncepts and evidence. The Rangeland Journal, 30(1), 167 - 176. 45 Menard, S. (2011). Standards for standardized logistic regression c oefficients. Social Forces, 89(4), 1409 - 1428. Michigan Department of Natural Resources, Wildlife Division. (2016). The GPS: Gu iding Principles and Strategies - Wildlife Division Strategic Plan 2016 - 2020. Retrieved from Wildlife Division Publications: http://www.michigan.gov/dnr/0,4570,7 - 153 - 10370 - 232 589 -- ,00.html Michigan Department of Natural Resources, Wildlife Division. (2017a). 2018 Northern Lower Peninsula Deer Private Land Assistance Network (NLP Deer PLAN) Handbookhttp://www.michigan.gov/documents/dnr/NLP_Deer_PLAN_Program_Guide_ 450993_7.pdf Michigan Department of Natural Resources, Wildlife Division. (2017b). Wildlife Habitat Grants Handbook. http://www.michig an.gov/documents/dnr/WHGP_2017_ Handbook_572241_7.pdf Michigan Department of Natural Resources, Wildlife Division. (2017c). 2017 Uppe r Peninsula Deer Habitat Improvement Partnership Initiative Executive Summary Handout. http://www.michigan.gov/documents/dnr/DHIPI_Exec_Summary_Handout_548246_7.p d f Minhas, S., Hoff, P. D., & Ward, M. D. (2019). Inferential Approaches for Network Analysis: AMEN for Latent Factor Mo dels. Political Analysis, 27(2), 208 - 222. Moore, J. A., Xu, R., Frank, K., Draheim, H., & Scribner, K. T. (2015). Social network analy sis of mating patterns in American black bears (Ursus americanus). Molecular e cology , 24 (15), 4010 - 4022. Moreno, J.L. ( 1934) Who Shall Survive? Nervous and Mental Disease Publishing Company, Washington, DC. Provan, K. G., Veazie, M. A., Staten, L. K., & analysis to strengthen community partnerships. Public Administration Revi ew, 65(5), 603 - 613. Prell, C., Hubacek, K., & Reed, M. (2009). Stakeholder analysis and social network analysis in natural resource ma nagement. Society and Natural Resources, 22(6), 501 - 518. Prell , C. (2012). Social network analysis: History, theory and methodology. Sage. Pretty, J. N. (2003). Social capital and connectedness: Issues and implications for agriculture, rural development and natural resource management in ACP countries: a review pap er for CTA. R Core Team (2018). R: A language and environment for statistical computing. R Foundation fo r Statistical Computing, Vienna, Austria. URL https://www.R - project.org/ . 46 Reed, M. S., Graves, A., Dandy, N ., Posthumus, H., Hubacek, K., Morris, J., ... & Stringer, L. C. (2009). Who's in and why? A typology of stakeholder analysis methods for natural resource management. Journal of environmental m anagement, 90(5), 1933 - 1949. Robertshaw, F., Harvey, M. M., & Molinaro, L. (1993). Conservation partnerships: A field guide to public - private partnering for natural resource conservation. National Fish and Wildlife Foundation. Robins, G., Snijders, T., Wang, P., Handcock, M., & Pattison, P. (2007). Recent developme nts in exponential random graph (p*) models for social networks. Social networks, 29(2), 192 - 215. Rocha, L. M., & Jacobson, S. K. (1998). Partnerships for conservation: protected areas and nong overnmental organizations in Brazil. Wildlife Society Bulletin, 26, 937 - 946. Snijders, T. A. (2011). Statistical models for social networks. Annual review of sociology, 37, 131 - 153. Stevens, K., Frank, K. A., & Kramer, D. B. (2015). Do social networks influence small - scale Plo S one, 10(3), e0121431. Tang, J., Musolesi, M., Mascolo, C., Latora, V., & Nicosia, V. (2010, April). Analysing informat ion flows and key mediators through temporal centrality metrics. In Proceedings of the 3rd Workshop on Social Network Systems (p. 3). A CM. Torre, A. and Gilly, J.P. (2000). On the analytical dimension of proximity dynamics. Regional Studies, 34(2), 169 1 80 Trauger , D., Tilt, W., & Hatcher, C. (1995). Partnerships: Innovative Strategies for Wildlife Conservation. Wildlife Society Bulletin (1973 - 2006), 23(1), 114 - 119. Retrieved from http://www.jstor.org/st able/37832 08 links between information flow and fishing success in the Northumberland lobster fishery. Ecology and Society, 19(2). Vedovello, C. (1997). Sc ience parks and university - industry interaction: geographical proximity between the agents as a driving force. Technovation, 17(9), 491 - 531. Wasserman, S., & Faust, K. (1994). Social network analysis: Methods and applications (Vol. 8). Cambridge universit y press. Wellman, B., & Frank, K. (2001). Network capital in a multilevel world: Getting support from personal communities. Social capital: Theory and research, 233 - 273. 47 Wigley, T. B., & Sweeney, J. M. (1993). Cooperative partnerships and the role of pr iv ate landowners. In: Finch, Deborah M.; Stangel, Peter W.(eds.). Status and management of neotropical migratory birds: September 21 - 25, 1992, Estes Park, Colorado. Gen. Tech. Rep. RM - 229. Fort Collins, Colo.: Rocky Mountain Forest and Range Experiment Sta ti on, US Dept. of Agriculture, Forest Service: 39 - 44, 229.