Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
344241 | Assessing Writing | 2014 | 9 Pages |
Abstract
This is a response to Dr. Les Perelman's critique of Phase I of the Hewlett Trials. His argument is that the construct validity of the study was undermined because there was a high correlation between word count and vendor predicted scores. The response addresses the argument by showing that correlations do not mean causation. Further the reply illustrates how predications are actually formulated in automated essay scoring. The response concludes with an appeal for more research on the underlying constructs associated with writing.
Related Topics
Social Sciences and Humanities
Arts and Humanities
Language and Linguistics
Authors
Mark D. Shermis,