• String
  • Posts
  • Sociograms: mapping social relationships in the classroom

Sociograms: mapping social relationships in the classroom

Planning group work, class camps and early identification of risk

AuthorAuthorAuthor

String Team, Kahhow Lee & Rimjhim Relan
February 26, 2023

A number of resources introduced in this post (namely All Ears → LeaderBoardX) were created by Mr Poon Zi Li, Head of Department of Dept. of Physical Education Bowen Secondary School; Kahhow (author) adapted and introduced flows using FormSG, ChatGPT/GPT-3 and wrote up this piece after interviewing other teachers:

Application in the Singapore Education Context

  • Sociograms were used in Bowen Secondary to help determine class camp allocations.

  • For ACSI, independently administered applications of sociograms helped to determine the assignment of class committee roles. Questions used are similar to what will be shown: who hangs out with who; who do you hang out with most

  • Sociograms were also helpful in a particular school to identify “circles of vulnerability”

  • Deciding seating preferences while causing the least upset especially for younger students

  • … among other schools that we have yet to interview! (:

  • Note from the teachers who administered sociograms: we cannot administer them too early since that wouldn’t yield meaningful results (students need time to mix)

There are a few ways explored in this piece:

  1. LeaderboardX with FormSG and ChatGPT*

  2. LeaderboardX with All Ears and Excel*

  3. Hummi - more complex but feature rich

*This interim is needed in order to prepare data in a way that can be processed

What is LeaderboardX? Built by Timothy Pratley who also later developed Hummi, it is a free tool that allows one to draw interactive sociograms. Because of its simplicity and ease of use, this article will focus on that while also introducing Hummi.

1 Using LeaderboardX with FormSG and ChatGPT (for data cleaning)

This was demo-ed (albeit rather haphazardly) live during Tech Talk for Teachers on 20 Feb 2023.

Data Pipeline:

FormSG → data clean (using ChatGPT) → feed into Leaderboard X

If you go to LeaderBoardX and try to reverse engineer how the data is structured, you will arrive at something like this: 

"Person","Endorses"
"Angeline","likes", "Kahhow", "Angelia"
"Hwee Hwee","likes","Adalyn","Angelia", "Huihui"
"Grace","likes", "Angeline", "Huihui", "Hwee Hwee"
"Rich","likes", "Kahhow", "Huihui", "Rimjhiim"
"Hwee Hwee","likes", "Kahhow", "Rich", "Rimjhiim"
"Rimjhiim","likes","Kahhow", "Adalyn", "Hwee Hwee"
"Rich", "likes","Kahhow", "Adalyn", "Hwee Hwee"

So we notice, 2 columns: person, endorses

And the structure is that of: [person] [likes] [up to 3 names]

  • Beyond being comma separated values (csv), every word is also wrapped with “”

  • This is the data prep that we need to factor

Your choice of Form builder doesn’t really matter but will just use FormSG out of familiarity (all ears, google form, typeform, any form works just recommending FormSG and All Ears in a Singapore-MOE setting).

You can see the same form introduced in the session here just to get a sense of what question types are best. You can also see the video below for a live demo:

💡 Kudos to Hwee Hwee from ACSI for refining prompt engineering (screenshot below) to arrive at a cleaner iteration of how to use ChatGPT for data cleaning.

  • While it is a bit if ‘hard-coding’, this gives you a sense of how you can refine the data. With a bit of reflection, a prompt like the following should work: 

Add "likes" after the first word. Wrap each word with "" and comma separate the values. Remove , before each line break

Recording of the session held on 19 Feb can be found below:

3 Using LeaderboardX with All Ears + excel template

💡 Download link for the guided walkthrough using All Ears (MOE Form Builder) + excel template

All credits go to Mr Zi Li Poon from Bowen Secondary for generously sharing his excel template

  1. Slides for more info on what to look out for if you want to create an all ears form for a level to collect the responses (in the download link)

    • How to use the “display condition” in all ears to direct students to questions that will contain dropdown list of their classmates (need to use dropdown list so that the names selected are consistent)

    • How to use ranking qn in all ears to prevent them from selecting the same friend 3 times.

    • But cannot prevent them from selecting their own name.

  2. Text generator file:

    • Basically just some formulas to string the input into the format required by leaderboardx.

    • Copy and paste the date into the yellow cells.

    • The name in the sociogram will follow the name in the data by default.. but to make the sociogram less cluttered, it will be good to shorten their names.. Example: Christopher Andrews to Chris instead.

So you can choose to manually shorten the name in the column “Shortname” then all the vlookup formula will settle it in the text file.

  • Just copy the columns as indicated in the sheet and paste it to a new txt file.

  • Load from leaderboards

3 Hummi

Again not a sponsored post but the developer of LeaderBoardX, Timothy made this tool after LeaderboardX with advanced features that might be worth exploring, including deep configuration down to the DB schema:

Interactive Factors Framework was something new to me when I prepared for the session and might be worth exploring as a possible way to learn about complex students.

We are used to qualitatively processing and thinking about our students but having visual tools may then assist and enhance our work as educators.

4 Suggested (academic) readings

5 Follow-ups

  • Keep a lookout for the survey by ESTL (Experimental Science and Tech Lab) and ETD (EduTech Division)! Both nested under Ministry of Education HQ, they are co-developing a sociogram/ network analysis tool that is easy to use and compliant with data security standards

  • Join our discord to follow-up on such conversations and share with us your implementation of sociograms(: