Data Collection and Management

Survey Design

• Network Generator – question or prompt which generates a list of alters related to a specific relationship or connection
  • Connect, interact, communicate, influence
• Name interpreters – questions designed to collect information regarding the alters listed above
  • Gender, age, frequency of contact, perception of activity/support
• Alter interrelater – questions designed to determine connections between alters
  • Does Tom know Bob?
  • Details structural holes

Egocentric

• Can use all three – Generator, Interpreter, Interrelater
• Names are not important in this setting
• Collect information on alters from the ego perception
• Alter Limits – Some surveys have limited the number of alters an ego can nominate

Egocentric Example

Whole Network

• May only use name generator as all other elements are reported by the others in the network
• Roster based - supplies a roster of names from the bounded network
  • Can be helpful to match names
  • May be difficult with large networks or not possible if you do not have all of the names
• Free recall - the ego supplies names from memory
  • Larger networks or networks in which you do not know all members
  • May be difficult to match names (Bob/Robert)
• Both come with a level of bias – roster may lead to over reporting, free recall may lead to under reporting

Whole Network Example

Survey Administration

• Researcher administered – industry standard
• Online surveys – can be difficult based on software
• Nomination limit concerns
• Roster / recall concerns

Software Available (collection)

• EgoWeb2.0
• PARTNER
• Network Canvas
• Enso (formerly OpenEddi)
• Qualtrics

Other types of collection

• Observational networks
  • Observational techniques similar to SOPARC, SOPLAY, SOFIT can be modified to track target individuals and the interactions between them and others
  • Multiple timepoints are needed to detail multiple connections
• Natural Networks
  • Networks already in place in which you would like to determine impact of their structure
  • Example: Parks or crossings connected by trails / paths
    • Nodes – parks and attributes of the park
    • Ties / Edges – trails which connect them
    • May determine important trails to maintain or parks which are important midpoints in trail networks
• Two-Mode Networks
  • Nodes are not connected to each other but are connected through a second type of node (mode)
  • Example: people using a park – person -> park -> person
    • This can then tell the parks of PA resources people may have “shared”
• Cognitive Mapping
  • Type of whole network interviewing based on perception
  • Every member asked to map all connections between every person in the network
  • Perceptions of all individuals are overlapped and condensed to develop a final network
  • Example: adolescents connections at after-school program
• Public Records
  • Social media accounts – scraping data from accounts
  • Email records
  • Public announcements
  • Organization member lists

Data Files

• Relational data: connects one node to another
  • Edgelist: easiest form – A-B, B-C, A-D
  • Matrix: all members are listed on X and Y axis, 1 is placed in each cell which a connection is present a 0 is placed if there is no connection
• Attribute table – file containing all ego information
  • Demographics, outcome variables, etc.

Slides