A computer-assisted grading tool has been developed and tested at Missouri S&T. This software automatically grades multiple choice assignments and exams using image processing technology. The tool speeds grading time, increases grading accuracy and enables instructors to focus on student learning instead of time for grading massive assignments and exams.
As large section classes and massive open online courses continue to generate interest, there is more demand for implementing automated grading. A computer-assisted grading tool that has been developed and tested at Missouri University of Science & Technology will be presented and discussed in this presentation. The tool was created by Dr. Nicolas Libre, Assistant Teaching Professor of Civil Engineering at Missouri S&T. It is currently under beta test and will be available for all users in 2018. The software is compatible with all devices with a Windows or Mac operating systems.
This software is ideal for large-enrollment courses requiring proctored, paper assessments. Because of limited capacity in computer labs and scheduling difficulties, paper assessments are often selected over computer-based assessments in large classes. This choice eliminates scheduling difficulties and reduces the potential for networking glitches during high-stakes assessments. This method is also less costly than using proctoring services for online assessments and eliminates the various software and hardware difficulties which may arise when students are completing work from personal computers in off-campus locations.
The software enables instructors/graders to automatically grade multiple choice, exams, assignments and surveys using image processing technology. Additionally, the grading is completed without the need for costly special hardware or paper. Specifically, the software scans three items on an assessment cover sheet. First, the software scans a numerical code assigned to each student. This code is the identifier used for the student for all assessments. As a safeguard, students are also required to write their names on the cover sheet. Second, the software scans a barcode which indicates the version of the assessments. The software has been tested and proven accurate while using four versions of the same assessments scanned simultaneously. Third, the software scans the answers selected by the student and uses this information to calculate a grade.
During the grading process, the software generates images of the graded files including correct and incorrect answer indicators and an overall assessment score. The tool also has built-in warnings and flags so that any errors can be addressed and corrected as the software is grading. The grader can also choose to ignore flags and address all issues after all scanned files have been graded. Finally, the software generates analytics in both graphical and spreadsheet form for analysis of class, question, and specific student learning objective statistics.
The software was successfully used by the developer for grading 4 midterms as well as a final assessment including more than 1500 exam papers. In April of 2017, the tool was also successfully used by another instructor at Missouri S&T. In this second trial use of the software, the instructor was able to run the software on a virtual Windows computer via a Macintosh computer. This environment slowed the progress of grading, but the system was still much faster than grading by hand and the analytics were also automatically generated. Since that time, this instructor has adopted the technology for regular use and two additional instructors plan to pilot the software in their classes during the summer of 2017 with anticipated adoption in their large-enrollment courses in the fall of 2017.
Such an automated tool can speed up the grading time, increase grading accuracy and enable instructors to focus on student learning instead of spending time for grading massive assignments and exams. The pros and cons of the developed tool will be presented and potential applications in other courses will be discussed.