Hello,
I am conducting an expert study involving two distinct groups of psychology experts (study participants), each being specialized in a different concept. The first group is specialized in concept A, while the second is specialized in concept B.
Participants from each group are tasked with evaluating the 8 psychological resources in terms of their importance to the concept they are specialized in.
Using a scale ranging from 1 to 7, experts in the first group evaluate the importance of these resources to concept A, while experts in the second group evaluate the same resources in terms of their importance to concept B. The resulting data will be structured as illustrated in the attached images.
I intend to conduct two separate network analyses based on these evaluations:
- One network analysis using data from experts in group A.
- Another network analysis using data from experts in group B.
However, my objective is to identify the most central resources of a new concept A-B defined as the intersection of concepts A and B.
To achieve this, I would like to merge the two networks and identify the resources that are central within this combined network structure.
As I am relatively new to network analysis, I have a few questions:
- Is there a network analysis technique suitable for merging networks based on data collected from distinct participant groups (however, with parallel sets of nodes, i.e., same 8 resources in both networks)?
- Are there alternative network analysis approaches that could help achieve my study objectives, using the described methodology?
Thank you in advance for any insights you can share.
Best regards,
Dominik
There are multiple ways to think about your problem. One is to address it before or without a network projection/analysis step. Basic statistics could probably answer your questions.
Network of resources
You need to keep in mind that you are studying the importance of resources, not the participants themselves. This may sound strange, but one way is to transpose your current tables into incidence matrix tables representing two networks of resources (resources connected by participants; line = resource). With this data structure you could compute a centrality measure on the resources.
Network of participants
Another possible way to approach the problem from your current tables is to create a network of participants. I see three variables: participant ID, group affiliation, and an ordinal scale for the importance attributed to each resource.
If you want to use Gephi (a great software for network visualization and analysis), you need two tables: one attributes table for your participants, with two columns (Participant ID, Group affiliation), and a second that corresponds to an incidence matrix showing the importance of each resource by participant (similar to your table, but combined into one).
So, all participants must be included in the same table instead of one table by group. In Gephi, you can use a force-directed layout and set your vertex/node color according to group affiliation variable, and observe the results. You could probably use your ordinal scale values as an edge weight parameter. Perhaps you will see cluster of participants connected by few edges (resources).
With a network of participants (based on your current tables), if you want to calculate resource centrality, you need edge centrality measure. I am not sure if you can directly compute this type of measure in Gephi.
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude