“Great works are performed, not by strength, but by perseverance.”

A famous quote by Samuel Johnson explains how perseverance is a better indicator of success than any other strength. A team of students in SCET’s Data-X course developed a tool to quantify and help build the grit levels in an individual. 

Inspired by Angela Duckworth’s Research highlighting key indicators of success, a big picture of this project was initially proposed by Elias Castro Hernandez (Technical Program Manager AI/ML at SCET) during one of the course lectures. This immediately drew the attention of five CS undergrads along with one statistics major from UC Berkeley’s College of Engineering, who was keen to work on areas focusing on Human Intent Prediction. On Extensive Research about the topic, the team identified that success of Grit lies not only within the traits which Duckworth points out but in other characteristics as well and existing tools hugely lacked proper statistical analysis as they were solely dependent upon simple uni-dimensional scoring of responses which mostly resulted in inaccurate results. 

Over the due course of fewer than four months, through the constant support of mentors from SCET, Team BIT has managed to build a research-backed state of the art tool which surveys users with select questions and estimates their GRIT level. This involves quantifying a user’s ability to strenuously work towards challenges, maintaining effort and interest over a period of time; despite failure, adversity, and plateaus in progress.

Cemented upon statistical validations, BIT (Berkeley Index for Tenacity) provides users with a visual analysis of their primary component GRIT and fundamental sub-components: Passion, Perseverance, Courage, Resilience, and Conscientiousness. Apart from these visualizations, the tool also gives a comprehensive analysis of scores along with specific recommendations in order to keep the users engaged and gradually build their grit levels. 

The team has worked courageously to develop a sophisticated tool in direct partnership with Cognitive Scientists, I/O Psychologists, Psychometricians, Designers, and Engineers. In this project, UC Berkeley’s very own Bear Assessment System has been implemented which combines Behavioral analytics, Machine learning, and development Engineering in order to assess individuals and help them to be better versions of themselves. It also involved the incorporation of Item Response Theory with a partial credit model for statistical analysis and multi-dimensional scoring of survey responses based upon a complex node-based Construct Map.

Framework Behind Scoring

Project Framework Behind Scoring

Since every individual in the team was new to the realm of Psychometrics, in order to build an end-to-end system, the team encountered several problems lasting from understanding and validation of extensive literature for survey design, this was overcome by reconfiguring the time allocation for the initial phase in Agile sprint to around 3 months; as most of the team came from strong CS backgrounds, they trusted themselves on implementing the platform in shorter duration. The initial phase was efficiently utilized for reiterating through constant consultation with field experts like Ikhlaq Sidhu, Paul Li, and Shruthi Bathia which helped the team in achieving a high-quality questionnaire. This schedule change in turn affected the team’s data collection process owing to which sufficient data was not generated through tool marketing strategies in a short span of 1 week. Yet, team BIT tactically overcomes this issue through Synthetic Data Generation of the missing rows.

The current pandemic situation helped the entire world realize the Grittiest ones would be the most benefited. Let it be adapting to classes, battling the ailment or even continually taking a shot at individual objectives. Hence, why don’t you consider testing out this project yourselves and let the team know how well their system quantified your personality traits!

Project by: Jahangir Abbas Mohammed, Jaideep Cherukuri, Sajan Kumar, Juliann Nguyen, Sagun Suryavanshi, Hoi Wei Thong