Auto TA: Contextual Coding Assistance for Novice Data Scientists

My role: UX researcher, learning experience designer, and author (and implemented some features!)

AutoTA Demo Video

Abstract

Novice R users without programming experience often struggle with debugging due to the lack of systematic knowledge and timely help. Existing built-in debugging tools in R are insufficient to help novice users solve problems. Despite the abundance of external debugging tools in other programming languages for different user groups, similar tools for novice R users are still scarce if any. AutoTA is created to teach debugging strategies within authentic debugging tasks. AutoTA’s activity-centered problem-shooting flow and timely feedback provide learners contextual debugging assistance immediately when they get into trouble. Scaffolding features allow AutoTA to cater to individual variability. Approachable user interface relieves psychological and emotional stress for novice debuggers.

Design Overview

AutoTA embeds debugging pedagogy into the real debugging situations that emerge during the coding process. Rather than creating additional tasks for users to complete, AutoTA seeks to walk users through the bugs they encounter and provide contextual support, and cater to individual variability through scaffolding. The friendly interface and conversational wording choices also aim at relieving the psychological stress users may experience while debugging.

Approachable User Interface

AutoTA adopts an approachable interface to relieve the emotional stress novice users usually experience while debugging according to our user interviews. The personification of the tool is embodied through a smiling robot asking “What went wrong?” so that users can feel emotionally connected and encouraged. Stylistically, messages in AutoTA are written in a conversational, concise, and informative tone to afford a quick and easy understanding of error messages.

The sections are expandable to avoid information overload for users (see Figure 2 and Figure 3). The default AutoTA interface collapses the detailed information so that the users will not feel overwhelmed by the additional reading task in the debugging process. The signals on each main block afford clicking, and they can expand the specific block to get further information. Once they finish reading the information, they can scroll down or collapse the block again.

The colorful interface also differentiates the tool from the RStudio environment. When AutoTA pops up, there will be 3 pink blinks to attract users' attention. Pink is derived from the red color that has the connotation of errors, while it is milder visually to avoid nervousness from users. The green color for each block signifies the solution as this color has both calming and energetic attributes, and is also the color for “exit” signs. In addition, users can customize the color settings by clicking the palette symbol on the upper right corner.

Final project for CS402 Beyond Bits and Atoms: Design Technological Tools, Stanford University In collaboration with Will Crichton and Hung Nguyen at Stanford CS Department