1. How does EvoGrader work?

EvoGrader uses sophisticated machine-learning methods to analyze student responses. A corpus of more than 10,000 human-graded written responses to many different types of questions was used to build the models that score student responses.

2. How do I use EvoGrader?

It's easy! All you need to do is upload an Excel (csv format) file with students' written responses, and EvoGrader analyzes the responses for a variety of concepts--both scientific and naïve--and provides data about how the students are performing. The videos show you how to set up your Excel file for analysis. If you need a data file to practice EvoGrader, please let us know. We will send you the file.

3. What information does EvoGrader provide about student responses?

Currently, EvoGrader can reliably detect up to nine standards-based concepts in student responses. It also provides information about general student reasoning models--pure scientific, "mixed" (scientific and naïve), and pure naïve. The videos and publications page will tell you more about these concepts. We are currently building additional models, so more concept detection tools are forthcoming!

4. How accurate is EvoGrader?

Published research has shown that EvoGrader is as accurate as trained human raters. Keep in mind that humans are not perfect raters! For this reason, we recommend that users study overall class-- or sample-level responses rather than individual student responses. EvoGrader is constantly being improved, and keeps getting better! You can learn more specifics about scoring quality in the publications page.

5. What types of questions can EvoGrader score?

EvoGrader was designed to score short answer written evolutionary explanations, like those produced in response to the ACORNS instrument. We have a database of more than 80 questions that you can use.