Data collection Instruments for Egocentric Networks
EPH 647: Community Based Participatory Research and Social Network Analysis
2024-03-27
Ego-centric Data Types
- Ego-centric (also called personal network) data, typically has three main data types:
Data about the respondent (called ego). which consists of each row as the participant/ego response, and each column represents a variable or characteristic of the ego (e.g., egos age, BMI, smoking status, etc.)
Which consist of the egos nominated alters (e.g, peers, friends, etc.) characteristics (e.g., alter age, ethnicity, etc.) and ego-alter dynamics (e.g., closeness to alter, frequency of meeting, shared a meal together etc.)
Data that includes the alter-alter ties as reported by ego. This is also called an “edge list” or can be described as an adjacency matrix. For example:
alter 1 and alter 2 know each other.
alter 1 and alter 2 meet when ego is not there.
alter 1 and alter 2 have shared a meal together.
In ego-centric (also called personal networks) data, alters can only belong to one ego, and alter-alter ties only exist between alters that were nominated by the same ego. In other words, in this type of analysis, there is no overlap between the different networks of different egos.
Data Collection Software
Social network analysis (SNA) is a growing field and there are many different data collection software, which you can check out a list of different data collect software here.
However, for today we will focus on Network Canvas.
Major Components of Network Canvas
Network Canvas has three main programs:
which is the software that allows primary investigators and researchers the ability to build their Social Network protocol. (e.g., this is where the researcher can define the different data types and the different questions)
which is the software that helps deploy and run the network canvas interview. (e.g., This is where the research assistant and the participant complete the questionnaires)
The central hub software to keep multiple network canvas protocols updated if you are running a major, multi-site study.
The Network Canvas (NC) team is expecting to launch a “studio” version, to be able to support for longitudinal project designs in about 2-3 years, so this software workflow may change.
For this lesson, we will not be talking about server today, as the typical way for social network researchers to use Network Canvas is through the architect and interviewer software.
Components of Personal Network Research
When collecting personal network data, there is a few fundamental principles that researchers are interested in collecting when running SN projects:
The way that network canvas is structured, it tried to capture social network data in its most fundamental way, by capturing nodes and edge attributes. To collect this type of data, there are generally 4 main components needed:
A survey that can collect ego characteristics, such as, age, BMI, smoking status, etc.)
This type of data can be collected via NC, but it can also be collected through other survey software. My favorite, in my opinion the golden standard, is [REDCap.](https://www.project-redcap.org/)
Keep in mind that REDCap does not really support network data like NC does, but it is great when you’re interested in collecting ego data.
- for example: mental health scales like MACVS, PTSD scales, demographic information.
also called Name Generator:
NC ability to produce alters, that the ego can nominate in their personal network.
This allows for building out the personal networks. There are different name generator form types (depending on your data needs), such as quick add per network, quick add per alter, using roasters for recall etc.)
- For example: “Think about the people you know: Who are the people in which you discuss personal matters with? List up to 13 individuals.”
Also called Alter Ego Attributes:
NC ability to capture alter-ego dynamics. (e.g., closeness, frequency of contact, etc.)
This is also where the researcher has the ability to capture ordinal categorical type of data. (e.g., how close is the ego to the alter? Very close, somewhat close, not close at all)
This allows for the participant and researcher, to describe and build the layout of their personal networks.
NC allows for different types alter-alter tie connections.
For example, the tie between two alters represents that the alters “know” each other, or “hang out” together when the ego is not there, or “have fought” with each other.
A sociogram is one type of method to building personal networks, but there are other types that are supported by network canvas, such as dyad-census (which goes through every possible dyad connection to build the ties)
![]()
Input control
Input control is the ability to modify the way in which the data is presented and how it will collected.
- For example, you can use a calendar type input control for age, or a text field for name.
Edge Types
In personal network research, edges describe the relationship between alters.
Network Canvas supports different edge types, which can be used to capture different types of relationships between alters.
Burden Reduction
In all human subject research (HSR), especially in epidemiological designs with complicated protocols, there’s always a give-and-take regarding burden level.
This is something to always keep in mind when conducting HSR, and therefore important to implement tips in any protocol workflow.
Tactics to reduce protocol burden includes skip logic:
General Protocol Process Workflow 🔄
Generally, when running the protocol with a potential participant, the process would look something like this:
