1

Transitioning From Studying Examples to Solving Problems: Effects of Self-Explanation Prompts and Fading Worked-Out Steps

Author: Atkinson, K. Robert, Merrill, Margaret Mary & Renkl, Alexander. (2003).Journal: Journal of Educational Psychology, 95(4), 774-783.Speaker：陸虹妙Date ： 2005/5/5

2

前言 -- 研究動機與背景 &文獻探討

Worked-out examples Consist of a problem formulation, solution

steps, and the final answer itself.

Research findings Learning from worked-out examples can

be more effective than Learning by problem solving. (Sweller & Cooper, 1985)

Although have significant advantages, their use as a learning methodology does not guarantee effective learning.

3

前言 -- 研究動機與背景 &文獻探討Self-Explanation effect

Self-Explanation effect is actually a dual process: generate inference and repair the learner’s own mental model.(Chi,2000)

Four relatively distinct Self-Explanation styles (Renkl, 1997) Passive explainers Superficial explainers Anticipative reasoners

With relatively high level of prior knowledge Principle-based explainers

Tend to identify the essential meaning of a problem. Attempt to articulate its goal structure. Elaborate on the principle that the example was intended to

convey. With low prior knowledge

It is functional to Elicit Principle-based explanations to learners With low prior

knowledge & Encourage anticipative reasoning to learners with high prior

knowledge.

4

前言 -- 研究動機與背景 &文獻探討

Self-Explanation effect A short self-explanation training

procedure produced an increase in the frequency of self-explanation activities and enhanced both near-transfer and far-transfer. (Renkl, Stark, Gruber, & Mandl, 1998)

5

前言 -- 研究動機與背景 &文獻探討Computer-based learning environment Conati and VanLehn (1999,2000)

Support learning from worked-out examples by prompting self-explanation.

Fill in browser items (physics rules or subgoals in a solution plan) as building blocks of self-explanations.

Give hints when necessary for further self-explanations.

The environment did not foster learning gains. Hausmann and Chi (2002)

Typed self-explanations into computers. Did not find positive effects of a computer-based

facility. Aleven and Koedinger (2002)

Obtained positive results With prompting for during problem solving rather than during example study.

6

前言 -- 研究動機與背景 &文獻探討Method (Renkl et al.,2000)

First, a complete example was presented (model). Second, an example was given in which one single

solution step was omitted (scaffolded problem solving). Then, the number of blanks was increased step by step

until just the problem formulation was left. i.e. a problem to be solved.

An experiment for testing fading procedure (Renkl, Atkinson, Maier, & Staley, 2002)

EP (example-problem) pairs vs. BP (backward procedure) vs.FP (forward procedure)

The fading procedure produced reliable effects on near-transfer items but not on far-transfer items.

Backward approach was more advantageous than forward approach.

7

前言 -- 研究目的

Does the BF produce more favorable leaarning outcomes than EP pairs?Does the use of self-explanation prompts in comparison with the lack of such prompts lead to better learning outcomes?Is there an interaction between the use of fading(vs. EP pairs) and the use of self-explanation prompts (vs. no prompts)?

8

研究方法—研究設計

2x2 between-subjects factorial design

The first factor: instructional material (BF or EP pairs)

EP pairs(Example-Problem)

BF (backward fading) The second factor: the

presence or absence of self-explanation prompts

Domain probability calculation

No_prompts

prompts

BF 19 20

EP Pair

s19 20

9

研究方法 --研究樣本78 educational psychology students 27 freshmen, 27 sohpomores, 13 juniors,

and 11 seniors); 15 males, 63 females (mean GPA=3.07,

mean aCT score=21.99)

10

研究方法 --研究工具Pencil-paper materials

A demographic questionnaire Judge learner’s prior knowledge in statistics and math in

general. An overview of the fundamental principles of probability A nine-item pretest

Prior knowledge in probability calculation. Each item only required the application of one probability

principle. A 12-item posttest

6 near-transfer items & 6 far-transfer items Involved the coordinated application of several probability

principles.

Computer-based learning environment Director 6.0 Description of Instructional Material

11

研究方法 --研究工具

12

研究方法—實驗流程 & 資料分析

流程圖統計方法 ANCOVA Pretest as a covariate. (

) timenalinstructio:exception;05.

13

Fill out a demographic questionnaire.

Pretest

Reading instructional text on basic principles of probability calculation.

BF with prompting vs. no promptingStudy the worked-out examples & solve the practice problems in computer-based learning environment.

EP Pairs with prompting vs. no promptingStudy the worked-out examples & solve the practice problems in computer-based learning environment.

Posttest

開始Start

End

實驗實施流程

14

研究結果

Each measure was tested for homogeneity of regression, and the result were found to be nonsignificant – all Fs <1.

15

研究結果— Analysis of learning-process measures

There was no interaction between type of instruction and self-explanation prompting. (F(1,73)=1.18, P=.28)There was no significant main effect for prompting. (F(1,73)=0.66, P=.42)The participants in BF conditions outperformed their peers in the EP pairs conditions in terms of accuracy of anticipations. (F(1,73)=10.05, MSE=0.07, P<.05)

16

研究結果— Analysis of learning-outcome measures

There was no interaction between type of instruction material and self-explanation prompting. (F(1,73)=0.81, P=.37)There was a significant main effect for type of instruction material on near transfer. (F(1,73)=4.50, MSE=0.05, P<.05)

The participants in BF conditions significantly outperformed their peers in the EP pairs conditions.

There was a significant main effect for self-explanation prompting on near transfer. (F(1,73)=5.01, P<.05)

The participants with self-explanation prompting outperformed their peers without self-explanation prompting.

17

研究結果— Analysis of learning-outcome measures

There was no interaction between type of instruction material and self-explanation prompting. (F(1,73)=0.81, P=.37)There was a significant main effect for type of instruction material on far transfer. (F(1,73)=5.99, MSE=0.03, P<.05)

The participants in BF conditions solved significantly more far-transfer problems than their peers in the EP pairs conditions.

There was a significant main effect for self-explanation prompting on far transfer. (F(1,73)=4.50, P<.05)

The participants with self-explanation prompting solved significantly more far-transfer problems than their peers without self-explanation prompting.