Description:

Co-Principal Investigator: McDaniel, Mark A.

Purpose: Students report that re-reading and highlighting text are their main study strategies, but such strategies are not as effective as ones that involve retrieval practice. Even though essays maximize retrieval practice, minimize exposure to errors, and encourage organization and elaboration of to-be-learned material, research on retrieval practice has not focused on the potential benefits of essay testing. On the other hand, the writing-to-learn literature has emphasized how writing can be an effective learning strategy. This project will bring together cognitive and educational psychology research approaches with a programmatic set of experiments aimed at understanding the malleable factors that determine when essay testing promotes learning and retention of course content. The goal is to understand which version of essay writing yields the most benefits for a range of learning outcomes, and to identify potential implications for education practice.

Project Activities: This project will tie together the laboratory-based research on retrieval practice with research on essay writing in education to explore the malleable factors that determine when essay testing will provide significant learning benefits. First, in a laboratory setting, the research team will compare the benefits of essay writing to those observed following standard practice, including both standard study activities (e.g., note-taking) and standard tests (e.g., short answer). Next, in a laboratory setting, the research team will identify the key parameters that will optimize the benefits of essay testing for memory and learning. Finally, the research team will conduct a classroom study, using the practices that promoted the most learning in the laboratory studies, to provide important data about feasibility that will be useful for the future development of an intervention.

Products: The products of this project will be preliminary evidence of potentially promising essay-writing and essay-testing strategies and peer reviewed publications.

Structured Abstract

Setting: Research will be conducted in laboratories at Duke University as well as in classrooms in urban and suburban schools in North Carolina.

Sample: Approximately 240 Duke University undergraduates (aged 18-24 years), 410 Washington University in St. Louis undergraduates (aged 18-24 years), and 300 high school juniors and seniors will participate in the laboratory studies. Approximately 150 high school juniors and seniors will participate in the classroom study. All high school students will be from urban and suburban counties in North Carolina.

Intervention: The goal of this grant is to understand which aspects of essay writing instructional tasks yield the most benefits for learning outcomes. The components that will be addressed are whether essay testing should be open or closed book, the contribution of essay-writing training, the effect of different essay-writing instructions, and the effect of different types of feedback. The findings from this grant can be used to inform the development of an essay-writing intervention.

Research Design and Methods: The researchers' plan includes a total of six laboratory studies and one classroom study, organized into three steps. For all experiments, students will study textbook excerpts in different ways depending on the experiment and condition and then will be tested on their memory for the text later. In the first step of the research plan, the research team will use laboratory experiments to explore the claim that essays promote long-term retention and transfer relative to typical study activities and test-for-learning formats. In the second step, the team will use laboratory studies to identify the particular malleable factors that yield the most benefits for essay writing by manipulating particular components of essay writing. In the third step, the team will conduct the classroom study, where the most powerful essay activity (as identified in the laboratory studies) will be compared to standard practice in a high school classroom setting. Of the experiments proposed, some will use a between-subjects design, where participants are randomly assigned to only one condition, and others will use a within-subjects design, where each participant experiences each condition in the experiment.

Control Condition: For the laboratory studies, the control or comparison condition being used will vary depending on what is being tested. In some experiments, the control conditions are typical study habits (e.g. note-taking) or other testing techniques (e.g. short answer questions). In other experiments, there is no standard control condition, but rather a comparison condition such as a different variation of essay writing (e.g. open-book versus closed-book). For the classroom study, the control condition will be business as usual.

Key Measures: The key measures that will be used to assess performance are memory tests designed by the researchers. The tests will be free recall, multiple choice, and problem solving and will be designed to assess students' memory for the texts that they had studied earlier. In addition to these measures, the research team plans to collect individual differences data such as performance on the multi-media comprehension battery (MCCB), to assess individual differences in structure building, as well as (when available) American College Testing (ACT) and Scholastic Aptitude Test (SAT) Writing Section scores, to assess individual differences in writing ability.

Data Analytic Strategy: The researchers plan to use a General Linear Model (GLM) approach, including ANOVA and regression models. The research team plans to include contrasts for learning condition, school level, individual difference variables, and interactions between these variables.

Products and Publications

Journal article, monograph, or newsletter

Arnold, K.M., Daniel, D.B., Jensen, J., McDaniel, M., and Marsh, E.J. (2016). Structure Building Predicts Grades in College Psychology and Biology. Applied Cognitive Psychology, 30(3): 454–459.