Policy and Complex Systems
Network analysis refers to this behavior as heterophily . Every agent in a network can be associated with a set of attributes that can be used to define homophily and heterophily . The choice to forge links in a network is labeled as either homophily or heterophily , depending on how an agent utilizes the relationship between their attribute values and those of the prospective partner ( Goodreau et al ., 2009 ). Simply put , homophily is when ‘ like attracts like ’ and heterophily is when ‘ different attracts different .’ When agents apply homophily , they will share a link with another agent who shares the same value on an attribute , such as when two non-profit organizations choose to interact . When agents apply heterophily , they will share a link with another agent who has a different attribute value , such as when a public program and a non-profit organization choose to interact .
An organization may choose homophilic pairings for one attribute , such as their jurisdiction , by preferring to work with other organizations in the same city or state . That same organization may simultaneously prefer a heterophilic pairing for another attribute , such as their sector . An example of this choice is when governmental programs partner with private business or non-profit organizations , due to the assets that private partners can bring to government , such as differing resource bases and constituencies ( Koliba , Meek , & Zia , 2010 ).
Network structure can also drive partner selection . Agents will sometimes find new partners when their current partners introduce them to potential new partners ( Wasserman & Faust , 1994 ). Network analysis refers to this behavior as transitivity . Transitivity occurs when agents select partners who share links to a common third agent , forming a complete triangle in the network ( Wasserman & Faust , 1994 ). For organizations , this could be the result of business referrals or public events that promote mixing , for example . Agents may also desire links to well-connected agents , finding value in the access to resources that a well-connected partner can provide , such as information or money , that the links direct , or in the influence that links provide , for example . When agents place intrinsic value in having links , they will seek to collaborate with other agents who already have a large number of links . This forms a pattern of preferential attachment , where a small number of agents are very highly connected while most of the agents have relatively few links . Networks that display preferential attachment are referred in network theory as “ scale free ” networks ( Albert & Barabasi , 2002 ).
Network Link Decay : Attributes and Bridges
Burt ( 2000 , 2002 ) identified that the probability that a link will cease to be present from one period to the next is related to the homophilic and heterophilic relationships between the nodes who share the link ( Burt , 2000 ). He has also determined that the probability is related to whether or not the link is a network bridge ( Burt , 2002 ) defined as a link that connects two otherwise unconnected components of a network ( Wasserman & Faust , 1994 ). In identifying bridges , researchers typically look for those components to contain many agents and links , though a bridge still exists if one or both of the agents would otherwise be an isolate node without the bridge . Burt ( 2002 ) operationalizes a bridge such that it is synonymous with Granovetter ’ s ( 1973 , 1983 ) definition of a weak network link , which is any link between two nodes in a network when those nodes do not link to any common third node . Burt ( 2000 , 2002 ) offers independent formulas for each decay process . As links age , they
6