A Computer Learning Environment for Novice Java Programmers That Supports Cognitive Load Reducing Adaptations and Dynamic Visualizations of Computer Memory

Abstract

Learning to program a computer is difficult for many. The Learning Edge Momentum hypothesis suggests that the difficulty may be due to the tightly integrated nature of programming concepts and adapting the way curriculum is offered may have a significant influence on the outcomes. We investigate applying cognitive load reducing methods to instruction of the introductory programming concepts of declaration, assignment and sequence, using a new learning environment that an instructor can adapt for a specific example or that a student can personalize for amount and modality of content provided. Our study has three learning surveys. Each learning survey has short instructional videos designed using cognitive load reducing methods and then asks participants to solve novel problems using the presented materials. Our first learning survey was completed by 123 participants recruited on Amazon's Mechanical Turk (AMT). We found that 23% that watched the instructional video without computer memory (n=61) answered the three code tracing questions correctly. Our second learning survey included instructional videos prepared after analyzing the results of the previous survey and emphasized cognitive load reducing methods in preparing the new instruction. This second survey was completed by 220 participants also recruited via AMT. We found that 57% of the participants that watched the instructional video without computer memory (n=72) answered the three tracing questions correctly. Our third learning survey with 322 participants recruited via AMT confirmed that the difference between the two videos was statistically significant with medium effect size. In the third survey, 29% of the participants watching the first survey instructional video without computer memory and 45% of those that watched the second survey instructional video without computer memory answered all three tracing questions correctly. In the third learning survey, the gain from 29% from our first short video that we thought was a reasonable presentation to 45% in the second short video seems to lend strong support to the hypothesis that our typical methods of instruction for introductory programming simply overwhelm the cognitive capabilities of many of the students. Our results suggest that cognitive load reducing methods may be very helpful for teaching introductory programming concepts.