Abstract
Computer programming is inherently ill-suited to
traditional assessment with partial credit, as real-world
programming tasks have a binary evaluation scheme (works or
does not work) allowing multiple attempts. In this study, the
specifications grading paradigm is presented as an alternative
method of assessing computer programming in a more authentic
manner. The authors propose a specifications grading scheme for
an introductory programming course, and describe the
implementation of this scheme using a custom online learning
platform with automated grading capabilities. A set of criteria are
defined for assessing the quality of a grading system and these
criteria are used to discuss the merits of the proposed grading
system. The authors argue that the proposed system is superior to
the traditional assessment models in terms of fostering authentic
learning, providing more accurate and reliable assessment, and
saving the instructor time.