This package supports a class social network exercise through a Qualtrics survey file and an online visualization dashboard. The dashboard also generates individualized network reports. The latest version of the dashboard is available at https://balachia.shinyapps.io/ob2-survey.
There are a few steps to get the network survey set up:
Before starting, download the network survey
files
containing various templates for setting up the survey. The survey
folder
contains templates, and the sample
folder contains sample results for the
Zachary Karate Club network.
First, you must create the email list/network roster for the survey. Open
survey/panel_template.xlsx
and fill it in with the class details:
FirstName
/LastName
: Self-explanatory, needed only to refer to the student.Email
: Critical, needed to distribute the survey and reportsFullName
: Critical, needed for students to identify each other in the rosterNodeID
: Critical, needed for the dashboard to link network respondents/respondees
Students can be listed in any order (alphabetical, by section, etc.). BUT, be sure that NodeID is sorted 1, 2, ..., else the dashboard will not correctly link students.
Save this panel as a .csv
file (e.g. section-aX-panel.csv
) for uploading to
Qualtrics.
In your Qualtrics account, create a new contact list (Contacts > Create Contact
List). Give it a meaningful name (e.g. "Panel A#"). Finally, upload the file
from the last step (section-aX-panel.csv
).
Verify that the preview contains the NodeID
, Email
, and FullName
fields.
In your Qualtrics account, use the survey template (survey/Network_Survey.qsf
)
to create a new survey (Projects > Create New Project > Survey > From File).
Rename the survey to something meaningful (e.g. "Network Survey - Section A#"), then upload the survey template file. Finally, click "Get Started" to create the survey.
In the survey, fill in the network roster. Return to your panel file
(section-aX-panel
) and COPY all names from the FullName
column. In the
survey, find the question called (network
), select the default text ("Click to
write Statement 1"), then PASTE the list into this field.
This will fill in the network roster. Click out of the text entry field, and make sure that the list of students looks correct. Finally, click "Publish" to finalize and save the survey.
Notes:
- Feel free to edit the text of the survey, BUT: make sure that the network
roster question is titled
network
and the anonymization consent question is titledconsent
. - If the
NodeID
column is sorted 1, 2, ... in the same order as theFullName
column pasted into the survey, the dashboard can link responders to respondees in the survey output.
When you're ready, distribute the survey over email.
From the survey page, create an email distribution (Distributions > Distribute Survey > Email > Compose Email).
Select the panel you've created (My Library > "Panel A#" > Select Entire Contact List). Set up the email and subject---check that it looks correct by sending yourself a preview email. Fill in the sending time, then click "Send" to send out the email.
You can also schedule reminder emails to those that have not completed the survey.
Once the survey is completed, download the data.
From the survey page, download the data as a CSV (Data & Analysis > Export & Import > Export Data > CSV). Check off "Download all fields" and "Use choice text", then click "Download" to download the data.
At this point, you should be ready to display the network data.
There are several ways to set up and use the network dashboard. The easiest is
to use the version available at https://balachia.shinyapps.io/ob2-survey. You
preview the dashboard by using sample results mimicking the Zachary Karate Club
Network (sample/ZKC Qualtrics.zip
).
Upload the survey results file (Upload Qualtrics file). The network will display as soon as it finished loading. If there is an error or the screen turns gray, please let me know, as there is probably some kind of bug.
Select whether the dashboard should display the "Advice" or the "Support" network (Select network). Then select which tie direction to display (Select network direction): directed ties, mutual ties (both respondents selected each other), or any ties (either responded selected the other). In the directed network, mutual ties are thick, while one-directional ties are thin.
You may use this to select the network that looks cleanest and least dense.
Highlight node centrality on various metrics (Color by). In each case, darker nodes have a higher score (i.e. the darkest nodes have the highest eigenvector score, betweenness score, etc.).
- Out-degree: the number of people that ego selected. "Self-perceived popularity"
- In-degree: the number of people that selected the ego. "True popularity"
- Closeness: the average distance of ego to all others. "Personal Kevin Bacon score"
- Betweenness: how often paths between individuals must pass through ego. "Network control ability"
- "Eigenvector": how connected ego is to other well-connected people. "Influence/status"
- "Cluster": what fraction of ego's ties are mutually connected. "Local community"
You can also highlight the names of individual respondents that have consented to have their names revealed.
First, select "Highlight consenters" to highlight the nodes whose names are available: consenters are large, while non-consenters become small. Then, select "Reveal names" to be able to reveal individual names. When you click on a node, its name will appear below it. Anonymous respondents will display "???".
If you would like to present the networks offline, you can export any view of the network by clicking "Export as png" in the bottom right of the map.
You can also download the network data to analyze offline in other softare (Download network data > Download).
You might start by briefly discussing the features of the network, such as any apparent clusters or bridges. Then you step through various measures of centrality, and contrast e.g. out-degree with in-degree popularity, or show how popular respondents may not be the most central on closeness or betweenness metrics. Finally, you might move to highlight individual respondents, and ask the class to guess their identity or discuss whether their position matches prior guesses.
The dashboard can also generate individualized reports for every node in the
network (Generate reports > Download). Wait for this process to complete; once
it is done, you can download all the reports as a zip file. See sample/ZKC Reports.zip
for a sample reflecting the Zachary Karate Club network.
The reports are anonymized and show the student's position in the advice and support networks, as well as their percentile scores on various centrality measures.
Reports are named by their NodeID. It should be possible to distribute these reports through some kind of mail merge. I have not yet set up a pipeline for this.