What the E-rater Grades
The E-rater is a “bot,” or a distant cousin of search engine spiders used to analyze and read web pages. The E-rater will read your essays and look for phrases that indicate competent reasoning.
The E-rater uses a stored battery of hundreds of graded essays for each of the 280 essay questions (this is partly why GMAT essays haven’t changed in years; to do so would require re-programming the E-rater). The E-rater has sample 1, 2, 3, 4, 5, and 6 score essays for each topic. The E-rater will evaluate your essay in terms of the stored essays in its database. If the essay you wrote resembles the stored “6” essays in the E-rater’s database, you will get that score. If your essay better resembles the “5s” in the database, you will get a “5” from the E-rater.
That is why it is so important to read the our 20 sample essays. You will see how well-written arguments are structured and you will learn the proper style necessary to impress both the E-rater and the human grader.
What the E-rater doesn’t grade
The E-rater cannot detect certain things, such as humor, spelling errors or grammar. It analyzes structure through the use of transitional phrases, paragraph changes, etc. It evaluates content by comparing your score to that of other students. If you have a brilliant argument that uses an unusual argument style, the E-rater will not detect this style.
The E-rater does, however, detect spelling and grammar indirectly. If your transition phrases and logical identifiers (e.g., “therefore”, “for example”) are not properly spelled, the E-rater will not detect them. Since the E-rater uses the presence of such transitional phrases as an indicator of effective writing, you are indirectly penalized if they are not spelled correctly.