Optimizing distributed practice

May combine two claims.

Including gaps of at least one day ensures the range includes at least one night of sleep, which may play a significant role in memory retention.

“Including gaps of at least one day insures that the range includes at least one night of sleep, which may play a significant role in memory retention”

• cites Peigneux, Laureys, Delbeuck, & Maquet, 2001 sourceCould not verify

  The indexed copy returns a repository listing and bibliography, but no passage from the paper’s body addressing whether a night of sleep aids memory retention. Searches for post-training sleep, overnight consolidation, and memory retention found no usable body-text passage; the repository abstract is not evidence under this rubric. (LLM, conf=0.70)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S1 (asserted), S3 (load-bearing).

Previous authors have suggested that test accuracy might follow an inverted U-function of gap.

“The possibility that test accuracy might follow an inverted U-function of gap has been suggested by previous authors”

• cites Balota, Duchek & Paullin, 1989Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Glenberg, 1976 sourceNot checked — no full text of the source
• cites Glenberg & Lehmann, 1980 sourcePartly supported

  The Abstract and General Discussion (pages unavailable in our copy) say that recall varies with the gap between repetitions. They describe a proportionality rule: recall increases with spacing when the retention interval is relatively long, but decreases with spacing when it is relatively short. They do not suggest an inverted-U curve with a peak at an intermediate gap. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Distributed practice research has had little effect on educational practice.

“area has had little effect on educational practice (Dempster, 1988(Dempster, , 1989;;Pashler, Rohrer, Cepeda, & Carpenter, 2007).”

• cites Dempster, 1988 sourceCould not verify

  The cited DOI identifies a horticulture paper, not an education paper, and it is not indexed in Semantic Scholar; its text and abstract were unavailable for verification. A snippet from “Beyond Time-on-Task: The Relationship Between Spaced Study and Certification in MOOCs” says the spacing effect was slow to enter educational practice and cites a Dempster (1988) work, suggesting the citing document may have linked to the wrong Dempster paper. That secondary snippet does not establish that the horticulture paper makes the claim, but the unavailable source text also prevents the definitive check required for a misattribution verdict. (LLM, conf=0.78)

  Experimental — accuracy not yet measured
• cites Dempster, 1989 sourceCould not verify

  The supplied DOI identifies Dempster’s *Theory and Applications of Belief Functions*, a different topic from distributed practice; two searches within that paper returned no snippets supporting the claim. An unscoped snippet in *Beyond Time-on-Task: The Relationship Between Spaced Study and Certification in MOOCs* says the spacing effect was slow to enter educational practice and attributes that observation to Dempster (1988), but its score is only 0.517. This suggests the supplied source is a bibliographic mismatch, but the available evidence does not establish the intended work strongly enough for a misattribution verdict under the specified threshold. (LLM, conf=0.72)

  Experimental — accuracy not yet measured
• cites Pashler, Rohrer, Cepeda, & Carpenter, 2007 sourceSupported

May combine two claims.

Welborn (1933) used a 28-day test delay but omitted feedback from Session 2, implying the second session probably provided no opportunity to learn items not learned in Session 1.

“Welborn (1933) used a 28-day test delay, but also omitted feedback from Session 2, implying that the second session probably provided no opportunity for learning those items not learned during Session 1”

• cites Welborn (1933)Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Pashler, Cepeda, Wixted, & Rohrer, 2005 sourceCould not verify

  The Materials and stimuli section says this paper’s own Session 2 occurred about one week later and gave no feedback. Searches for Welborn, 1933, and a 28-day delay found no passage describing Welborn’s study or supporting the inference about its second session. (LLM, conf=0.75)

  Experimental — accuracy not yet measured

May combine two claims.

The literature contains no studies supporting the idea that distributed practice aids concept learning, and recent work by the authors does not support it either.

“Additionally, Gagné et al. state that distributed practice improves concept learning but we have found no existing studies in the literature that support this claim, and our recent studies fail to support this claim (Pashler, Rohrer, Cepeda, & Carpenter, 2007).”

• cites Pashler, Rohrer, Cepeda, & Carpenter, 2007 sourcePartly supported

  The section “Spacing in perceptual categorization learning” reports no spacing benefit in the authors’ recent categorization studies, supporting that part of the claim. It says this task is “almost absent” from the spacing literature, not that no existing study supports distributed practice for concept learning. The paper also reports a spacing benefit for learning a mathematical principle, so its negative finding should not be generalized to all concept learning. (LLM, conf=0.85)

  Experimental — accuracy not yet measured

May combine two claims.

In two three-session laboratory studies, an optimal gap improved final recall by up to 150%.

“An optimal gap improved final recall by up to 150%.”

May combine two claims.

In Experiment 1, Tukey HSD tests showed the zero-day gap produced significantly worse recall than the 1-, 2-, 4-, and 7-day gaps, with no other pairwise comparisons significant.

“Tukey HSD tests show that the zero-day gap produced significantly worse recall than the 1, 2, 4, and 7-day gaps; no other pair-wise comparisons were significant.”

May combine two claims.

In Experiment 2, performance on the third and final test of Session 1 showed no main effect or interaction involving gap, although facts showed slightly greater learning than pictures (94 vs 90%), F(1,155)=10.3, p<.005.

“Likewise, performance on the third and final test of Session 1 showed no main effect or interaction involving gap, although facts showed slightly greater learning than pictures (94 vs. 90%), F(1,155) = 10.3, p < .005.”

May combine two claims.

The six-month gap results agree in kind and magnitude with the three-month gap results and with prior literature.

“The data from this group are qualitatively and quantitatively consistent with the three-month gap data and the literature more broadly.”

May combine two claims.

For delays of minutes, the best gap is nearly equal to the delay; for delays of many days, it is about one-tenth of the delay.

“At very short test delays, on the order of minutes, the ratio is close to 1.0; at multiday test delays, the ratio is closer to 0.1.”

May combine two claims.

In Experiment 2, the 28-day gap produced 151% greater retention than the 0-day gap, whereas the 1-day gap produced only an 18% improvement over the 0-day gap.

“the 28-day gap produced 151% greater retention than the 0-day gap, whereas the 1day gap produced only an 18% improvement over the 0-day gap.”

May combine two claims.

Gap had nonmonotonic effects on final recall even when test delays exceeded one week; accuracy rose and then fell as gap grew.

“Second, gap had nonmonotonic effects on final recall even with test delays longer than a week; accuracy first increased and then decreased as gap increased.”

• Also stated in: § INSERT FIGURE 5 ABOUT HERE · boosted

  “Studyphase retrieval theories predict -and our data show -an inverted-U shaped function of gap on performance following a test delay.”

May combine two claims.

In Experiment 2, first-test accuracy ranged from 75-82% by gap for facts and 45-64% by gap for objects.

“First-test accuracy ranged from 75 -82% by gap for facts and from 45 -64% by gap for objects.”

May combine two claims.

Fitting the quadratic function of log gap yielded best scores of 71%, 52%, and 21% and best gaps of 3.7, 25.6, and 37.1 days for Experiment 1, Experiment 2 facts, and Experiment 2 objects, in that order.

“Fits of this function to the data in Experiments 1, 2 (facts), and 2 (objects) produced optimal test scores of 71%, 52%, and 21%, respectively, and optimal gaps of 3.7, 25.6, and 37.1 days, respectively.”

• AssumptionRests on an assumption: The quadratic-in-log-gap function is an adequate description of the relationship between gap and final test score, despite not being theoretically motivated.

May combine two claims.

In Experiment 2, there were main effects of gap, F(5,155)=8.3, p<.001, and material, F(1,155)=502.2, p<.001, and an interaction between gap and material, F(5,155)=4.6, p<.005.

“There were main effects of gap, F(5,155) = 8.3, p < .001,and material, F(1,155) = 502.2,p < .001,and an interaction between gap and material, F(5,155) = 4.6, p < .005.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Those four studies suggest that a gap of roughly one day is optimal, but they do not demonstrate this with much certainty, especially given the restricted test delays.

“These studies suggest that a gap of roughly one day is optimal, but they hardly demonstrate this claim with any certainty, especially given the restricted set of test delays used.”

Classified as probably not a substantial claim. May combine two claims.

In Experiment 1, subjects were not taught to criterion in the second learning session because doing so would confound gap with number of trials.

“(Subjects were not taught to criterion in the second learning session, as they were in the first, because that would have confounded gap and the number of trials required during the second session, as explained in the Introduction.)”

May combine two claims.

Few data address retention overnight, much less over weeks or months.

“Few data speak to retention overnight, much less over weeks or months.”

Classified as probably not a substantial claim.

In Experiment 1, subjects were randomly assigned to one of six gap conditions.

“Subjects were randomly assigned to one of six conditions (0, 1, 2, 4, 7, or 14 day gap).”

Probably not a substantial claim: may not pass S4 (based).

No one has quantitatively characterized distributed practice functions over intervals much longer than a day.

“Because no one has quantitatively characterized the nature of distributed practice functions over time intervals much beyond a day”

Classified as probably not a substantial claim.

Little is known about how the distributed practice effect develops over long, nontrivial intervals.

“yet little is known about how this socalled distributed practice effect unfolds over nontrivial periods”

May combine two claims.

The current results and prior work support the possibility that the best gap grows with the test delay but becomes a smaller share of it.

“Instead, the present findings, in conjunction with the literature, are consistent with the possibility that the optimal gap increases with test delay, albeit as a declining proportion of test delay.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

In Experiment 2, items answered correctly during the pretest were excluded from analysis, leading to exclusion of less than 1% of items on average.

“For each subject, items that were answered correctly during the pretest were excluded from analysis of their data, leading to the exclusion of less than one percent of items, on average.”

May combine two claims.

The results demonstrate that a one-day gap is not always optimal, since zero- and one-day gaps were not significantly different, and recall was significantly worse for one-day versus 28-day gaps.

“Quite dramatically, this demonstrates that a one-day gap is not always optimal, since zero-and one-day gaps were not significantly different, and recall was significantly worse for one-day versus 28-day gaps.”

Probably not a substantial claim: may not pass S1 (asserted).

Because the four prior studies and Experiment 1 used approximately equal test delays, a much longer test delay might yield an optimal gap other than one day.

“However, because these four studies and Experiment 1 used approximately equal test delays, the possibility remains that a much longer test delay would yield an optimal gap other than one day.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

In Experiment 2, dropout rates did not vary across conditions except that the six-month gap condition had fewer subjects.

“Table 2 shows fewer subjects in the six-month gap condition, partly due to the increased difficulty maintaining contact with these subjects; otherwise, dropout rates did not vary across conditions.”

Probably not a substantial claim: may not pass S3 (load-bearing).

On many demographic measures, the six-month gap participants matched the other participants.

“Subjects in the six-month gap condition were equivalent to other subjects on a wide range of demographic measures.”

May combine two claims.

Using massed practice and overlearning wastes time.

“Morrison et al. (2001) suggest writing facts over and over to learn them; this prescription for massed practice and overlearning is a highly inefficient use of time (Rohrer, Taylor, Pashler, Wixted, & Cepeda, 2005).”

• cites Rohrer, Taylor, Pashler, Wixted, & Cepeda, 2005Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim. May combine two claims.

When textbooks do mention distributed practice, they usually give it a paragraph or less and make inconsistent, frequently wrong recommendations about spacing.

“Those texts that mention the distributed practice effect often devote a paragraph or less to the topic”

• Also stated in: § Practical Implications · plain

  “and offer widely divergent suggestionsmany incorrect -about how long the lag between study sessions ought to be”

• cites Glaser, 2000Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Jensen, 1998Not checked — no full text of the source
• cites Ormrod, 1998Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

The results suggest that the 28-day gap was optimal and support a claim that final-test recall gradually declines with too-long gaps.

“This suggests that the 28day gap was optimal and supports a claim that final-test recall gradually declines with too-long gaps.”

The quote only partly states this claim.

In Experiment 2, pilot testing confirmed that the multiple-choice options were about equally likely to be chosen by subjects with no prior knowledge of the fact or object.

“Pilot testing confirmed that the options in the multiple-choice test were about equally likely to be chosen by subjects with no previous knowledge of the fact or object.”

Classified as probably not a substantial claim.

In Experiment 2, subjects were randomly assigned to one of six gap conditions.

“Subjects were randomly assigned to one of six conditions (0, 1, 7, 28, 84, or 168 day gap).”

Probably not a substantial claim: may not pass S3 (load-bearing). The quote only partly states this claim.

In Experiment 2, all coding was done blind to experimental condition.

“All coding was done blind to experimental condition.”

May combine two claims.

The interaction between gap and material likely reflects different degrees of improvement relative to baseline for fact versus visual object materials, with no obvious qualitative differences.

“The interaction between gap and material likely reflects the different degrees of improvement, relative to baseline, for fact versus visual object materials; there are no obvious qualitative differences in the results.”

May combine two claims.

In Experiment 2, pilot data suggested Swahili-English word pairs would produce floor effects after a 6-month delay, so the authors chose material shown to produce lesser forgetting.

“Because pilot data suggested that Swahili-English word pairs (which were used in Experiment 1) would produce floor effects after a 6-month test delay, we chose material that was shown to produce lesser rates of forgetting.”

• AssumptionRests on an assumption: The material chosen for Experiment 2 produces lesser rates of forgetting than Swahili-English word pairs.

May combine two claims.

A published encoding variability model and the authors' preliminary modeling both support encoding variability theory.

“Both a published encoding variability model (Raaijmakers, 2003) and our own preliminary modeling efforts (Mozer, Cepeda, Pashler, Wixted, & Cepeda, 2008) lend support to this theory.”

• cites Raaijmakers, 2003Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Mozer, Cepeda, Pashler, Wixted, & Cepeda, 2008Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

With only a limited set of gaps tested, the actual best gap could differ slightly from the observed one, leaving the data unable to confirm or deny a further decline in the ratio for multi-day delays.

“Because we only used a limited range of gaps in the present studies, and the true optimal gap in each of our studies might be slightly shorter or longer than the observed optimal gap, our current data neither support nor refute the existence of a further decreasing ratio, within the multi-day test delay period.”