skip to main content


Title: Matrix multiplication on batches of small matrices in half and half-complex precisions
Award ID(s):
1740250
NSF-PAR ID:
10190584
Author(s) / Creator(s):
; ;
Date Published:
Journal Name:
Journal of Parallel and Distributed Computing
Volume:
145
Issue:
C
ISSN:
0743-7315
Page Range / eLocation ID:
188 to 201
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Abstract Background

    Undergraduate STEM instructors want to help students learn and retain knowledge for their future courses and careers. One promising evidence-based technique that is thought to increase long-term memory is spaced retrieval practice, or repeated testing over time. The beneficial effect of spacing has repeatedly been demonstrated in the laboratory as well as in undergraduate mathematics courses, but its generalizability across diverse STEM courses is unknown. We investigated the effect of spaced retrieval practice in nine introductory STEM courses. Retrieval practice opportunities were embedded in bi-weekly quizzes, either massed on a single quiz or spaced over multiple quizzes. Student performance on practice opportunities and a criterial test at the end of each course were examined as a function of massed or spaced practice. We also conducted a single-paper meta-analysis on criterial test scores to assess the generalizability of the effectiveness of spaced retrieval practice across introductory STEM courses.

    Results

    Significant positive effects of spacing on the criterial test were found in only two courses (Calculus I for Engineers and Chemistry for Health Professionals), although small positive effect sizes were observed in two other courses (General Chemistry and Diversity of Life). Meta-analyses revealed a significant spacing effect when all courses were included, but not when calculus was excluded. The generalizability of the spacing effect across STEM courses therefore remains unclear.

    Conclusions

    Although we could not clearly determine the generalizability of the benefits of spacing in STEM courses, our findings indicate that spaced retrieval practice could be a low-cost method of improving student performance in at least some STEM courses. More work is needed to determine when, how, and for whom spaced retrieval practice is most beneficial. The effect of spacing in classroom settings may depend on some design features such as the nature of retrieval practice activities (multiple-choice versus short answer) and/or feedback settings, as well as student actions (e.g., whether they look at feedback or study outside of practice opportunities). The evidence is promising, and further pragmatic research is encouraged.

     
    more » « less