A system and method for automated assessment of constrained constructed responses provides for automatic, e.g., programmatic, assessing of a test subject response to a constrained constructed response item or question. In one embodiment, paper test materials with test subject markings thereon are converted to electronic form. Test subject markings are further isolated, for example, by registration, sizing and removal of markings other than a graphic response. The subject markings are further compared with a baseline response, correctness deviation criteria and incorrectness deviation criteria (e.g., correctness, incorrectness, substantial correctness, substantial incorrectness, and so on) from which one or more of a score, further learning or other assessment results may be determined. Cluster analysis or other post-evaluation or post-assessment processing may also be conducted, or re-evaluation or re-assessment may also be conducted.