I am curious about the relationship between Bayesian Knowledge Tracing and the Teacher Dashboard. Can you say more about what the teacher sees or might see on the Dashboard?

(I had seen BKT in a different context altogether of an online physics game that examined student growth, then generated specific feedback to that student based his/her patterns of play.) 

Hi Sherry!

The dashboard provides two views for the teacher. The first view shows student progress and performance (P&P) and the second shows conceptual understanding (CI).

The P&P view lets teachers know where each student is in the game, how many challenges they have completed, and their performance on each challenge, which is indicated by the color crystal earned. Teachers also see an indicator of how many times each student attempted each challenge. Teachers have found this view to be a game-changer. (No pun intended!)

The CI view displays all the concepts tracked by the ITS and provides teachers with a window into student understanding. Teachers quickly see the specific concepts that students are struggling with (red flag) or have already mastered (green check). This information is provided for each individual student as well as in summary form for the entire class. Teachers can dive deeper into the report to learn more about each concept and make connections to related traits and specific game levels and challenges. For example, teachers may see that a majority of their students are struggling with incomplete dominance, which is addressed in Mission 2 with the Armor trait.

Teachers have found the P&P and CO views helpful both in and outside of class time. Teachers use the dashboard to track students during the class period and often go back to the dashboard later, as they prepare for the next day's lesson. 

Our BKT-infused ITS was built using past play data and is now being used to track student understanding of the concepts in real-time, delivering hints and remedial challenges as appropriate. Hints, which are text-based with visual highlights, are provided to the student after they meet a certain threshold of struggle. It's a fine balance--we let them have a little room to play but we don't want them to suffer too much! Hints are contextualized to each challenge type and address concepts and known misunderstandings with specific traits and patterns of inheritance. BKT is also used to determine if a student needs more than just a hint. If necessary, a simplified, remedial challenge is offered. Remedial challenges are highly scaffolded and guide students as they work through challenges that address one specific conceptual misunderstanding at a time. The dashboard alerts teachers when remediation is used. 

Thanks!
Trudi ~:)

Thanks Scot! 

Currently, Geniventure is being tested in two levels of biology classrooms: middle school and high school. We've found that high school students can typically get through all of Geniventure in approximately 5five 45-minute class periods. However, that is roughly 45-minutes of strict gameplay. We hope to design teacher materials that would scaffold effective discussion points and opportunities for additional exploration outside of Geniventure into the classroom agenda.

Through the GeniConnect project, Geniventure is also being used in more informal settings such as afterschool programs and summer camps. We are coupling the online game with hands-on activities that will not only foster middle school students' understanding of genetics concepts but also an awareness of STEM careers.

Thanks Rachel! We are still researching students' learning gains using Geniventure. The concepts that we've identified and are tracking were carefully chosen using Evidence Centered Design. However, we have not yet done a comparison against some other modes of teaching genetics. I'd be happy to share when we do!

Hi Jessica - thank you and great questions! We do hope to design teacher materials as a part of our effort in developing a cohesive piece of technology for use in the classroom. The handouts you see in the video were developed by our teachers who are piloting Geniventure for us. We hope to work closely with these teachers as well as other experienced biology teachers to incorporate handouts and other guides for a productive classroom experience.

Students playing in pairs is actually not part of our study design but turns out that it naturally happens in the classroom and we definitely encourage it when we set students up to play the game in class. We've observed that when students are given the opportunity to engage with each other during gameplay, they often learn from each other. Students are able to explain to each other, in their own words, how the mechanisms of genetics work, which hopefully will deepen their understanding of the concepts.

Hi Robert,

We do have some preliminary data on student interest in science and STEM careers after our summer pilot of the GeniConnect program last year in 2017. We are collecting data using a retrospective survey designed by The PEAR Institute, specifically identifying interest in genetics (more broadly) and STEM careers. Our hope is that students will notice the characters in the game and their role in the “underground lab” in Geniventure. However, the connection to STEM careers in the GeniConnect program cannot solely be attributed to the characters in the game alone. The GeniConnect project couples Geniventure gameplay with engaging activities and experiences such as field trips to biotech labs, visits from volunteer scientists, and hands-on laboratory experiments.

We have another implementation of the GeniConnect program scheduled to take place this summer with a local group of middle school students. We are eager to find out if our suite of activities yields the type of interest in STEM careers that we are anticipating!

Hi Eli,

I'm a computer scientist on Geniventure's Intelligent Tutoring System (ITS) project. We are using a human-authored rule system to evaluate the student's move/answer in each step of the problem. The binary result (correct/incorrect) is then evaluated using a BKT-based tutor to calculate the probability that the student knows the skill/concept. The ITS then uses this probability (compared against a threshold) to decide whether to present a hint or remediate for the skill associated with the step.

We don't specifically attempt to deduce student confusions. We simply evaluate whether the student's answer was correct or incorrect and use that to update the probability that the student knows the skill. The BKT algorithm has parameters to account for 'slips' (making a mistake even though they understand the concept) or 'guesses' (the student got it right even though they don't understand the concept).

Regards,

-Robert