Optimized scheduling - Revision history

Mbett at 15:43, 28 October 2009

2009-10-28T15:43:37Z

← Older revision		Revision as of 15:43, 28 October 2009
Line 14:		Line 14:
	Mathematical models may be used to produce optimized schedules by computing the [[knowledge component]] that will be most efficiently learned if practiced next.		Mathematical models may be used to produce optimized schedules by computing the [[knowledge component]] that will be most efficiently learned if practiced next.

	The spacing recommendation in the recent IES practice guide on [http://ies.ed.gov/ncee/wwc/practiceguides/ "Organizing Instruction and Study to Improve Student Learning"] describes a generalization of this principle, basically that spaced practice leads to better long-term retention than massed practice [this should probably be added as a principle in the hierarchy]. Recent work by Pavlik (2007) qualifies the conclusions because most (if not all) of the research referenced in the guide does not control for time on task.		The spacing recommendation in the recent IES practice guide on [http://ies.ed.gov/ncee/wwc/pdf/practiceguides/20072004.pdf "Organizing Instruction and Study to Improve Student Learning"] describes a generalization of this principle, basically that spaced practice leads to better long-term retention than massed practice [this should probably be added as a principle in the hierarchy]. Recent work by Pavlik (2007) qualifies the conclusions because most (if not all) of the research referenced in the guide does not control for time on task.

	===Operational definition===		===Operational definition===
Line 37:		Line 37:

	==Caveats, limitations, open issues, or dissenting views==		==Caveats, limitations, open issues, or dissenting views==
	A limiting condition of using the optimized scheduling of Pavlik is that it relies on a recency advantage for practice. Without this recency advantage, it is often true that maximal spacing is optimal as suggested in [http://ies.ed.gov/ncee/wwc/practiceguides/ "Organizing Instruction and Study to Improve Student Learning"] practice guide.		A limiting condition of using the optimized scheduling of Pavlik is that it relies on a recency advantage for practice. Without this recency advantage, it is often true that maximal spacing is optimal as suggested in [http://ies.ed.gov/ncee/wwc/pdf/practiceguides/20072004.pdf "Organizing Instruction and Study to Improve Student Learning"] practice guide.

	For example, if each practice trial has a fixed duration this will result in no recency advantage and maximal (or very wide) spacing will be optimal. However, many procedures use the test trials since the [[testing effect]] has shown that tests result in stronger learning than passive study. Tests often have a strong advantage when they occur with greater recency since this recency reduces the need for review in the case of failure.		For example, if each practice trial has a fixed duration this will result in no recency advantage and maximal (or very wide) spacing will be optimal. However, many procedures use the test trials since the [[testing effect]] has shown that tests result in stronger learning than passive study. Tests often have a strong advantage when they occur with greater recency since this recency reduces the need for review in the case of failure.

Bmclaren: /* Description of principle */

2008-03-19T19:25:35Z

Description of principle

← Older revision		Revision as of 19:25, 19 March 2008
Line 14:		Line 14:
	Mathematical models may be used to produce optimized schedules by computing the [[knowledge component]] that will be most efficiently learned if practiced next.		Mathematical models may be used to produce optimized schedules by computing the [[knowledge component]] that will be most efficiently learned if practiced next.

	The spacing recommendation in the recent IES practice guide on [http://ies.ed.gov/ncee/wwc/practiceguides/ "Organizing Instruction and Study to Improve Student Learning"] describes a generalization of this principle, basically that spaced practice leads to better long-term retention than massed practice [this should probably be added as a principle in the hierarchy]. Recent work by Pavlik (~~2207~~) qualifies the conclusions because most (if not all) of the research referenced in the guide does not control for time on task.		The spacing recommendation in the recent IES practice guide on [http://ies.ed.gov/ncee/wwc/practiceguides/ "Organizing Instruction and Study to Improve Student Learning"] describes a generalization of this principle, basically that spaced practice leads to better long-term retention than massed practice [this should probably be added as a principle in the hierarchy]. Recent work by Pavlik (2007) qualifies the conclusions because most (if not all) of the research referenced in the guide does not control for time on task.

	===Operational definition===		===Operational definition===

PhilPavlik: /* In vivo experiment support */

2008-01-08T18:07:59Z

In vivo experiment support

← Older revision		Revision as of 18:07, 8 January 2008
Line 28:		Line 28:

	===In vivo experiment support===		===In vivo experiment support===
			Fall 2007 Chinese I results produced by Pavlik show increased time on task (motivational effect) for optimized vocabulary practice. In another experiment in Fall 2007, students using an optimized Chinese radical trainer experienced a gain in their future learning of Hanzi characters.

	==Theoretical rationale==		==Theoretical rationale==

PhilPavlik: /* Experimental support */

2008-01-08T18:03:25Z

Experimental support

← Older revision		Revision as of 18:03, 8 January 2008
Line 23:		Line 23:

	==Experimental support==		==Experimental support==
	~~See references below. [These should be summarized here.]~~
	===Laboratory experiment support===		===Laboratory experiment support===
	Optimal scheduling (an expanding schedule computed by the ACT-R based algorithm) results in easier (and faster) practice and better one week retention in the lab (Pavlik, accepted).		Optimal scheduling (an expanding schedule computed by the ACT-R based algorithm) results in easier (and faster) practice and better one week retention in the lab (Pavlik, accepted).

PhilPavlik at 18:00, 8 January 2008

2008-01-08T18:00:31Z

← Older revision		Revision as of 18:00, 8 January 2008
Line 48:		Line 48:

	==References==		==References==
			* Pavlik Jr., P. I., & Anderson, J. R. (accepted). Using a model to compute the optimal schedule of practice. Journal of Experimental Psychology: Applied.
	* Pavlik Jr., P. I. (2005). The microeconomics of learning: Optimizing paired-associate memory. Dissertation Abstracts International: Section B: The Sciences and Engineering, 66(10-B), 5704.		* Pavlik Jr., P. I. (2005). The microeconomics of learning: Optimizing paired-associate memory. Dissertation Abstracts International: Section B: The Sciences and Engineering, 66(10-B), 5704.
	* Pavlik Jr., P. I. (2007). Timing is an order: Modeling order effects in the learning of information. In F. E., Ritter, J. Nerb, E. Lehtinen & T. O'Shea (Eds.), In order to learn: How order effects in machine learning illuminate human learning (pp. 137-150). New York: Oxford University Press.		* Pavlik Jr., P. I. (2007). Timing is an order: Modeling order effects in the learning of information. In F. E., Ritter, J. Nerb, E. Lehtinen & T. O'Shea (Eds.), In order to learn: How order effects in machine learning illuminate human learning (pp. 137-150). New York: Oxford University Press.

PhilPavlik: /* Laboratory experiment support */

2008-01-08T18:00:06Z

Laboratory experiment support

← Older revision		Revision as of 18:00, 8 January 2008
Line 25:		Line 25:
	See references below. [These should be summarized here.]		See references below. [These should be summarized here.]
	===Laboratory experiment support===		===Laboratory experiment support===
			Optimal scheduling (an expanding schedule computed by the ACT-R based algorithm) results in easier (and faster) practice and better one week retention in the lab (Pavlik, accepted).

	===In vivo experiment support===		===In vivo experiment support===

PhilPavlik at 17:57, 8 January 2008

2008-01-08T17:57:16Z

← Older revision		Revision as of 17:57, 8 January 2008
Line 31:		Line 31:

	==Conditions of application==		==Conditions of application==
			One main condition of application is that the task requires some form of repetition of related [[knowledge components]]. In cases with single trials of unrelated items the schedule is trivial and cannot be optimized.

	==Caveats, limitations, open issues, or dissenting views==		==Caveats, limitations, open issues, or dissenting views==
Line 39:		Line 39:

	==Variations (descendants)==		==Variations (descendants)==
	While Pavlik (2005, 2007) has pioneered optimization of spacing for independent items, schedules ~~might~~ also be ~~optimize~~ by controlling practice quantity, or ~~optimized~~ by controlling the order (and or spacing) of scaffolding exercises for dependent items.		Examples of optimized scheduling include [[learning event scheduling]] (see [[Optimizing the practice schedule\|Pavlik's research program]]), the [[knowledge tracing]] algorithm used in [[Cognitive Tutors]] (see [[Using learning curves to optimize problem assignment\|Cen's study]]), and adaptive [[fading]] of [[scaffolding]] or [[assistance]] (see [[Does learning from worked-out examples improve tutored problem solving? \|Renkl's study]]).
			While Pavlik (2005, 2007) has pioneered optimization of spacing for independent items, schedules can also be optimized by controlling practice quantity (Cen) or by controlling the order (and or spacing) of scaffolding exercises for dependent items (Renkl).

	==Generalizations (ascendants)==		==Generalizations (ascendants)==

PhilPavlik: /* Variations (descendants) */

2008-01-08T17:53:06Z

Variations (descendants)

← Older revision		Revision as of 17:53, 8 January 2008
Line 39:		Line 39:

	==Variations (descendants)==		==Variations (descendants)==
	While Pavlik has pioneered optimization of spacing for independent items, schedules might also be optimize by controlling practice quantity, or optimized by controlling the order (and or spacing) of scaffolding exercises for dependent items.		While Pavlik (2005, 2007) has pioneered optimization of spacing for independent items, schedules might also be optimize by controlling practice quantity, or optimized by controlling the order (and or spacing) of scaffolding exercises for dependent items.

	==Generalizations (ascendants)==		==Generalizations (ascendants)==

PhilPavlik at 17:52, 8 January 2008

2008-01-08T17:52:39Z

← Older revision		Revision as of 17:52, 8 January 2008
Line 31:		Line 31:

	==Conditions of application==		==Conditions of application==


			==Caveats, limitations, open issues, or dissenting views==
	A limiting condition of using the optimized scheduling of Pavlik is that it relies on a recency advantage for practice. Without this recency advantage, it is often true that maximal spacing is optimal as suggested in [http://ies.ed.gov/ncee/wwc/practiceguides/ "Organizing Instruction and Study to Improve Student Learning"] practice guide.		A limiting condition of using the optimized scheduling of Pavlik is that it relies on a recency advantage for practice. Without this recency advantage, it is often true that maximal spacing is optimal as suggested in [http://ies.ed.gov/ncee/wwc/practiceguides/ "Organizing Instruction and Study to Improve Student Learning"] practice guide.

	For example, if each practice trial has a fixed duration this will result in no recency advantage and maximal (or very wide) spacing will be optimal. However, many procedures use the test trials since the [[testing effect]] has shown that tests result in stronger learning than passive study. Tests often have a strong advantage when they occur with greater recency since this recency reduces the need for review in the case of failure.		For example, if each practice trial has a fixed duration this will result in no recency advantage and maximal (or very wide) spacing will be optimal. However, many procedures use the test trials since the [[testing effect]] has shown that tests result in stronger learning than passive study. Tests often have a strong advantage when they occur with greater recency since this recency reduces the need for review in the case of failure.

	~~==Caveats, limitations, open issues, or dissenting views==~~

	==Variations (descendants)==		==Variations (descendants)==
			While Pavlik has pioneered optimization of spacing for independent items, schedules might also be optimize by controlling practice quantity, or optimized by controlling the order (and or spacing) of scaffolding exercises for dependent items.

	==Generalizations (ascendants)==		==Generalizations (ascendants)==
	[[~~Spaced practice~~]]		[[Instructional schedule]].

	==References==		==References==

Koedinger: The prior description did not specific a contrast between the suggested approach and some alternative .

2008-01-04T02:23:11Z

The prior description did not specific a contrast between the suggested approach and some alternative .

← Older revision		Revision as of 02:23, 4 January 2008
Line 6:		Line 6:

	==Brief statement of principle==		==Brief statement of principle==
	~~Applying an [[instructional~~ schedule~~]] that has been ordered to maximize [[robust learning]]~~. ~~Mathematical models may often be used to produce optimized schedules by computing the [[knowledge component]] that will be most efficiently learned if practiced next~~.		Optimized scheduling yields better long-term retention than a practice schedule based on fixed intervals (whether massed or spaced) or intervals self-determined by students (e.g., in flash card use).

	==Description of principle==		==Description of principle==

			This principle involves applying an [[instructional schedule]] that has been ordered to maximize [[robust learning]]. Optimized scheduling involves maximizing instructional efficiency (i.e., robust learning gains per instructional time spent) by mathematically deriving when a student should repeat practice of a [[knowledge component]]. The time interval between practice is optimal (neither too short or too long) when it best balances the benefit of enhanced memory strength due to retrieval at a long interval ([[spaced practice]]) and the cost of time to retrain due to retrieval failure at a long interval.

			Mathematical models may be used to produce optimized schedules by computing the [[knowledge component]] that will be most efficiently learned if practiced next.

			The spacing recommendation in the recent IES practice guide on [http://ies.ed.gov/ncee/wwc/practiceguides/ "Organizing Instruction and Study to Improve Student Learning"] describes a generalization of this principle, basically that spaced practice leads to better long-term retention than massed practice [this should probably be added as a principle in the hierarchy]. Recent work by Pavlik (2207) qualifies the conclusions because most (if not all) of the research referenced in the guide does not control for time on task.

	===Operational definition===		===Operational definition===
	Scheduling practice to maximize some future measure of learning given a fixed time of current practice.		Scheduling practice to maximize some future measure of learning given a fixed time of current practice.
Line 16:		Line 23:

	==Experimental support==		==Experimental support==
	See references.		See references below. [These should be summarized here.]
	===Laboratory experiment support===		===Laboratory experiment support===
	===In vivo experiment support===		===In vivo experiment support===

	==Theoretical rationale==		==Theoretical rationale==
	Optimized scheduling is often a method for providing optimal [[repetition]]. The optimized scheduling of Pavlik (2005; 2007) balances the speed (reduced time cost of practice) advantage of [[recency]] with the long-term learning advantage of [[spaced practice]]. This speed advantage typically occurs for [[drill]] practice because more recent drill practice has fewer failures and therefore less need for costly review practice.		Optimized scheduling is often a method for providing optimal [[repetition]]. The optimized scheduling of Pavlik (2005; 2007) balances the speed (reduced time cost of practice) advantage of [[recency]] with the long-term learning advantage of [[spaced practice]]. This speed advantage typically occurs for [[drill]] practice because more recent drill practice has fewer failures and therefore less need for costly review practice.

	==Conditions of application==		==Conditions of application==
	A limiting condition of using the optimized scheduling of Pavlik is that it relies on a recency advantage for practice. Without this recency advantage, it is often true that maximal spacing is optimal as suggested in ~~the practice guidelines.~~ [http://ies.ed.gov/ncee/wwc/practiceguides/ "Organizing Instruction and Study to Improve Student Learning"]		A limiting condition of using the optimized scheduling of Pavlik is that it relies on a recency advantage for practice. Without this recency advantage, it is often true that maximal spacing is optimal as suggested in [http://ies.ed.gov/ncee/wwc/practiceguides/ "Organizing Instruction and Study to Improve Student Learning"] practice guide.

	For example, if each practice trial has a fixed duration this will result in no recency ~~advanatge~~ and maximal (or very wide) spacing will be optimal. However, many procedures ~~uses~~ the test trials since the [[testing effect]] has shown that tests result in stronger learning ~~tha~~ passive study. Tests often have a strong advantage when they occur with greater recency since this recency reduces the need for review in the case of failure.		For example, if each practice trial has a fixed duration this will result in no recency advantage and maximal (or very wide) spacing will be optimal. However, many procedures use the test trials since the [[testing effect]] has shown that tests result in stronger learning than passive study. Tests often have a strong advantage when they occur with greater recency since this recency reduces the need for review in the case of failure.

	==Caveats, limitations, open issues, or dissenting views==		==Caveats, limitations, open issues, or dissenting views==

	==Variations (descendants)==		==Variations (descendants)==

	==Generalizations (ascendants)==		==Generalizations (ascendants)==
			[[Spaced practice]]

	==References==		==References==
	* Pavlik Jr., P. I. (2005). The microeconomics of learning: Optimizing paired-associate memory. Dissertation Abstracts International: Section B: The Sciences and Engineering, 66(10-B), 5704.		* Pavlik Jr., P. I. (2005). The microeconomics of learning: Optimizing paired-associate memory. Dissertation Abstracts International: Section B: The Sciences and Engineering, 66(10-B), 5704.