This article describes an empirical research study that applies the spaced retrieval practice method that has been recommended by cognitive psychologist to improve retention. The article makes a significant contribution because it applies this practice within a college-level pre-calculus course and measures the impact of that practice on course level objectives and outcomes. By simply manipulating the frequency and timing of key (key objectives) practice sessions and quizzes (with hints) researchers were able to measure significant differences in performance that equated to half a letter grade between subjects.
Citation
Hopkins, R. F., Lyle, K. B., Hieb, J. L., & Ralston, P. A. (2015). Spaced Retrieval Practice Increases College Students’ Short-and Long-Term Retention of Mathematics Knowledge. Educational Psychology Review, 1-21.
Summary
Purpose of the research
To determine the utility of “spaced retrieval practice” at a real university setting, and determining retention of commonly encountered math content across courses and semesters (long-term retention).
Their goal was to “determine whether spaced retrieval practice can increase retention of a complex body of college-level mathematical knowledge in both short- and long-term -- i.e., within the pre-calculus course and into the calculus course respectively” (Hopkins et. al, 2015, p. 15).
Research questions
Although research questions were not explicitly stated, researchers stated that they intended to test the effect of spaced retrieval practice in multiple ways including
1) within-subjects by examining whether individuals in the experiment group retained spaced objectives better than massed ones.
2) between-subjects by testing across groups to see if spaced-objectives in the experimental group was greater than retention in the control group (massed study).
3) hybrid by testing retention of massed objectives in the experimental group to retention in the control group. The researchers were interested in seeing if using spaced retrieval practice on some objectives would benefit non-spaced objectives
Methods
Manipulated spaced retrieval practice versus massed practice within a pre-calculus for engineering course. Under the experimental condition, target (key) objectives (those considered to be extremely important) were given priority and treated with a spaced retrieval practice. Those not selected (non-targeted) were treated in the traditional massed practice process.
The Quiz Me tool in MyMathLab (Pearson) was used to control the timing of the presentation of quiz questions (practice quiz questions) for each group. Quiz questions that focused on the targeted objectives were presented three times over the course of the semester. The first practice quiz was presented at the time of acquisition, the second practice quiz was presented on the next quiz, the third practice quiz was presented on the fourth quiz (increasing the time interval by skipping a presentation in the third quiz). It is interesting to note that the researchers skipped the third quiz to make the retrieval more effortful (thinking about desirable difficulties).
Subjects were randomly assigned to either the control or spaced retrieval practice.
Spaced versus massed retrieval practice was manipulated in a hybrid between- and within-subjects design. Click here to view a nice graphical explanation of experimental designs in educational research (see slide #8 for between- and within-subject design).
A unique characteristic about he methods used in this study is that they tested retention over an extended time. Tests included retention of content from the beginning of the semester (2.5 weeks) to the final exam (3 months). Previous studies had tested shorter durations. In addition, researchers followed a subset of subjects into the subsequent Engineering Analysis 1 (EA1) course (this is the first calculus course for engineering students) and tested their retention of foundational calculus material. The examined three performance measures in the EA1 course that included 1) first exam, 2) third unit exam, and 3) cumulative final exam.
Subjects
All subjects were enrolled in Introductory Calculus for Engineering (ICE) and a subset was then enrolled in Engineering Analysis 1 (EA1).
Of those enrolled in ICE and that took all the ICE exams 46 experienced the control condition (traditional massing) and 40 the experimental condition (some objectives spaced, some massed).
Of those who continued into EA1 and took all the EA1 exams 29 had experienced the control condition (in ICE) and 25 had experienced the experimental condition (in ICE).
Results
Overall performance results were calculated based on performance on final exam questions that specifically covered target objectives (ignoring questions which covered non-target objectives). A proportion of questions answered correctly was calculated for each student.
Experimental group = 1) proportion of spaced objectives correct, and 2) proportion of massed objectives correct
Control group = 1) proportion of massed objectives correct
Within-Subjects Analysis
· Spaced to massed objectives (Control versus Experimental)
· Proportions correct submitted to repeated measures analysis of covariance (ANCOVA) – F(1,35)=7.96, p=.008, n2=.150
o Proportion correct on spaced objectives (M=.71) was higher than on massed objectives (M=.68) – See figure 1 (bar chart)
Between-Subjects Analysis
· Spaced objectives (in experiment group) to all-massed group
· Proportions correct submitted to repeated measure analysis of covariance (ANCOVA) – F(1, 80)=9.01, p= .004, n2=.009
o Proportion correct on spaced objectives (M=.71) was significantly higher than massed ones (M= .63)
· Also, calculated subjects’ proportion correct collapsed across spaced and mass objectives in the experiment group (M =.70) to target objectives in the control group (M= .62) – F(1, 80) = 6.58, p= .012, n2 = .074.
Engineering Analysis 1 (EA1)
· First Exam: Students in the experimental group (space retrieval practice) scored higher on the first unit exam (M= .62) than those in control (M= .56)
o Only approached significance F(1, 64)=3.21, p= .078, n2 = .043
· Third Exam: Groups almost identical on target objectives
o Experiment (M= .66) & Control (M = .67)
· Cumulative Final: Significant advantage for students who had been in experiment group (M= .53) and control group (M= .43) F(1, 48)=4.68, p= .035, n2 = .087
Other
o Satisfactory and unsatisfactory grades in EA1
o Also conducted satisfactory (C or better) versus unsatisfactory (C- or worse) overall grade analysis
o Control (48.3 % = satisfactory) & (51.7% = unsatisfactory)
o Experimental (68.0 % = satisfactory) & 32.0% = unsatisfactory)
o Those in experiment group were more than twice as likely as those in control group to earn a satisfactory grade
Discussion
Implications
This was a well-designed study that controlled for student differences through the randomized assignment of groups. In addition, groups were statistically indistinguishable on race, gender, high school GPA, and Math ACT. Furthermore, these variables were used as covariates in all analysis.
I agree with the authors that the between-group differences in both EA1 and ICE performance differences can be attributed with some confidence to the manipulation of retrieval practice (experimental condition) in the pre-calculus course.
Contribution and future research
Future research focus
Additional studies should attempt to replicate these findings in similar classes, including similar requirements of retention of materials over longer periods of time.
This study has made a significant contribution providing evidence that the spaced retrieval practice can make be applied in real college courses and that that impact can have a significant impact on student performance over time.
This research should be expanded to include other learning resources and methods beyond Pearson’s MyMathLab, to confirm findings.
Implications for the technology-enhanced learning (TEL) environment
The main manipulation in this research was accomplished through technology, mainly Pearssn's MyMathLab. This is extremely interesting because this type of automated manipulation can be scaled and, therefore, can have a larger impact if TEL's are designed or utilized with these principles at the forefront.
This research should be expanded to include other learning resources and methods beyond Pearson’s MyMathLab, to confirm findings.
This study has made a significant contribution providing evidence that the spaced retrieval practice can make be applied in real college courses and that that impact can have a significant impact on student performance over time.
This research should be expanded to include other learning resources and methods beyond Pearson’s MyMathLab, to confirm findings.
Implications for the technology-enhanced learning (TEL) environment
The main manipulation in this research was accomplished through technology, mainly Pearssn's MyMathLab. This is extremely interesting because this type of automated manipulation can be scaled and, therefore, can have a larger impact if TEL's are designed or utilized with these principles at the forefront.
This research should be expanded to include other learning resources and methods beyond Pearson’s MyMathLab, to confirm findings.
No comments:
Post a Comment