Posted July 8, 2019 鈥� Far too many students end up in high school with poor writing skills, setting them up to struggle in college or careers.
Milena Keller-Margulis, an associate professor at the 色花堂 College of Education, is working to change that. She has received a $1.39 million grant from the federal Institute of Education Sciences to develop a quicker and more reliable way to assess younger students鈥� writing abilities. College of Education Professor Jorge Gonzalez is the co-investigator on the grant, and Sterett Mercer, an associate professor at the University of British Columbia 鈥� Vancouver, is a co-principal investigator.
Their project involves studying whether computer-based scoring of short writing samples is effective 鈥� which would allow teachers to identify more easily which elementary school students are struggling with basic writing skills and need extra help.
鈥淲e are going to make some big steps toward being able to screen students in the area of writing, which we cannot do well right now,鈥� said Keller-Margulis, a professor of school psychology. 鈥淭his will allow us to ultimately change the trajectory of students who are struggling in writing. The earlier and more accurately we can identify them, the more time we have to change their performance.鈥�
The research comes amid ongoing criticism of lengthy test preparation happening in some schools across the country, with students spending hours on practice exams.
Instead, Keller-Margulis focuses on educators using a short but effective approach to check students鈥� basic academic skills and then track their progress. The method is called curriculum-based measurement, or CBM. While the approach does not yield a comprehensive analysis of a student鈥檚 skills, it鈥檚 comparable in value to taking a child鈥檚 temperature to find out if he or she is sick.
鈥淲e don鈥檛 need to do a shut-down day and give kids the whole state test,鈥� Keller-Margulis said.
Research suggests, for example, that the curriculum-based measurement for reading 鈥� counting the number of words a student reads correctly in a grade-level appropriate passage in one minute 鈥� can predict whether the student will pass the state test with decent accuracy.
鈥淚t鈥檚 a screener. It鈥檚 not going to be 100% perfect,鈥� Keller-Margulis said. 鈥淏ut the idea is that you鈥檙e not wasting a lot of instructional time with testing.鈥�
For the grant-funded project, Keller-Margulis has teamed with Gonzalez and Mercer to develop an improved version of the standard curriculum-based measurement approach for writing for elementary school students.
The typical method involves having students respond to a prompt, thinking for one minute and writing for three minutes. However, Keller-Margulis has found that writing for longer 鈥� seven minutes or perhaps 15 鈥� may be more effective.
In part, the research team plans to study the effectiveness of using automated scoring of students鈥� writing samples through a free or open-source program or a commercial program versus hand scoring. They also will look at whether the genre of the writing 鈥� narrative, informational or persuasive 鈥� has an impact.
The researchers are not suggesting that automated scoring is better than human scoring. Instead, they will evaluate whether an automated process gives teachers enough information to accurately screen students鈥� skills, thus prompting extra help from the teacher.
鈥淲hen you do curriculum-based measurement, it鈥檚 like taking someone鈥檚 pulse, or tracking where is a child in this particular domain such as writing?鈥� said Gonzalez, a professor of school psychology whose research focuses on early language and literacy. 鈥淎nd by knowing where a child is, it allows teachers to differentiate instruction when necessary for children who may need more supports. To do it well and timely is critical, especially for young children."
鈥揃y Ericka Mellon