Aggregate Compilation Behavior: Findings and Implications from 27,698 Users

Abstract

The error quotient (EQ) was first reported in 2006 as a behavioral measure of novice programmers. The EQ scores how well students deal with correcting syntax errors (or not) in their programs. The original studies were carried out on data collected using BlueJ, a pedagogic Java programming environment; today, newly installed instances of BlueJ capture data similar to these early studies automatically, meaning data regarding nearly 2 million programmers is captured every year by the Blackbox project. In this paper, we apply Jadud's original error quotient algorithm to this new, massive data set, and discuss our results and analysis in light of related work. Further, we consider the implications of our findings for researchers and educators in applying the EQ to 27,698 users in 10 different countries during the fall term of 2013.