PEG operates on the principle that "good writing can be predicted" by analyzing specific linguistic features. The system uses a two-stage process:
: The algorithm is fed a sample of essays (typically 100 to 400) already graded by expert humans. It calculates "beta weights" to determine which text features correlate most strongly with high scores. PEG operates on the principle that "good writing
The system distinguishes between "trins" (intrinsic quality like diction and grammar) and "proxies" (measurable correlations like average word length). It typically evaluates: Page, a former high school English teacher, conceived
: New essays are analyzed for these same features—known as "proxies"—and a score is calculated using the established statistical model. Key Analytical Features a former high school English teacher
: Dr. Page, a former high school English teacher, conceived PEG to help educators manage the overwhelming workload of grading. The first version was born at the University of Connecticut in 1964.
Project Essay Grade (PEG) is a historic milestone in the field of educational technology, representing the first major attempt to automate the evaluation of student writing. Developed in the 1960s by Dr. Ellis Batten Page, PEG laid the groundwork for modern automated essay scoring (AES) by demonstrating that computers could analyze prose with a level of reliability comparable to human graders. The Evolution of PEG