kirsten butcher

Download Kirsten Butcher

Post on 27-Jan-2016

21 views

Category:

Documents

0 download

Embed Size (px)

DESCRIPTION

Elaborated Explanations for Visual/Verbal Problem Solving:. Kirsten Butcher. Interactive Communication Cluster July 24, 2006. Visual & Verbal Information in Geometry. Geometry Cognitive Tutor: Angles and Circles Units. Research Goals. - PowerPoint PPT Presentation

TRANSCRIPT

  • Kirsten ButcherElaborated Explanations for Visual/Verbal Problem Solving:

    Interactive Communication ClusterJuly 24, 2006

  • Visual & Verbal Information in GeometryGeometry Cognitive Tutor: Angles and Circles Units.

  • Research GoalsTo understand how coordination between & integration of visual and verbal knowledge influences robust learning

    To explore the potential transfer of laboratory-identified multimedia principles to classroom context

    To inform the design of effective educational multimedia for classroom use

  • Relevant Learning ResearchLearning with MultimediaContiguity Effect (e.g., Mayer, 2001)Diagrams support inference-generation & integration of information (Butcher, 2006)

    Self-explanations & Cognitive TutorsSelf-explanations promote learning (e.g., Chi et al., 1994)Simple (menu-based) self-explanations support Geometry Learning (Aleven & Koedinger, 2002)

  • Hypotheses: Sense-making ScaffoldsContiguityWork & receive feedback in diagramIntegrated HintsApply verbal hints to visual problem situation (diagram)Elaborated ExplanationsVisual explanations to justify problem-solving

  • Hypotheses: Sense-making ScaffoldsContiguityWork & receive feedback in diagramIntegrated HintsApply verbal hints to visual problem situation (diagram)Elaborated ExplanationsVisual explanations to justify problem-solving

  • Connections to PSLC TheorySense-makingCoordinative Learning: Integrate results from multiple inputs & representations.Verbal informationVisual informationScaffolds change the format of the interface to promote coordinative learning.Contiguous representation: reduces mapping & supports inferences made directly from diagramIntegrated hints: reduce mapping & support recognition of critical visual elements

  • Hypotheses: Sense-making ScaffoldsContiguityWork & receive feedback in diagramIntegrated HintsApply verbal hints to visual problem situation (diagram)Elaborated ExplanationsVisual explanations to justify problem-solving

  • Connections to PSLC TheorySense-makingInteractive Communication: Tutor prompts explanationsStudents explain geometry principles that justify problem-solving stepsStudents receive feedback and hints on explanationsScaffold: Elaborated explanations require student to explain the application of geometry principlesRationale for explanations are visual in natureDiagram Condition: Visual format for explanationTable Condition: Verbal format for explanation

  • Existing Tutor: Explanations are verbal-only

  • Elaborated Explanations Tutor

  • Elaborated Explanations Tutor

  • Elaborated Explanations TutorDemo of the Geometry Cognitive Tutor with Elaborated ExplanationsNew & Improved! Now with more explanations!

  • Connections to PSLC TheoryWhat are the relevant knowledge components?(Verbal) Geometry principles.E.g., Inscribed Angle Theorem means that the measure of the angle is half the measure of the intercepted arc.(Visual) Geometry elements.E.g., Recognizing angles, arcs, and their relationships.(Integrated) Geometry inferencesE.g., Recognizing that an arc, which is associated with a known (or found) inscribed angle, can be found via the Inscribed Angle Theorem

  • Knowledge Components vs. Overall Visual Match

  • Knowledge Components vs. Overall Visual Match

  • Mapping Given Information to Elements

  • Integration of Principles and Elements

  • Superficial Strategies of Integration: Close = Connected

  • Robust Knowledge: Relationships connect Elements via Principles

  • Difficulty Factors Analysis (DFA): Problem Format & Explanation Type3 Problem FormatsDiagramQuadrantTable2 Explanation TypesSimple Explanations (Reasons Only)Elaborated Explanations (Reasons + Application)

  • DFA: Diagram Problem Format with Simple Explanations

  • DFA: Diagram Problem Format with Elaborated Explanations

  • DFA: Quadrant Problem Format with Elaborated Explanations

  • DFA: Table Problem Format with Elaborated Explanations

  • DFA Results: Given InformationLinear trend for Explanation Type, F (1, 88) = 3.8, p = .055

    Chart5

    86.590.41.51.51.71.7

    88.892.71.81.81.651.65

    88.289.31.61.61.651.65

    Simple Explanations

    Elaborated Explanations

    Problem Format

    Percent Correct

    Performance by Problem Format & Explanation Type

    DFA_Givens

    Means - GivensStdErrs_Givens

    Simple ExplanationsElaborated ExplanationsSimple ExplanationsElaborated ExplanationsDIV BY 2 for chart

    Diagram86.590.4Diagram3.431.71.5

    Quadrant88.892.7Quadrant3.33.61.651.8

    Table88.289.3Table3.33.21.651.6

    DFA_Givens

    001.51.51.71.7

    001.81.81.651.65

    001.61.61.651.65

    Simple Explanations

    Elaborated Explanations

    Problem Format

    Percent Correct

    Performance by Problem Format & Explanation Type

    DFA_NotGivens

    Means - GivensStdErrs_Givens

    Simple ExplanationsElaborated ExplanationsSimple ExplanationsElaborated ExplanationsDIV BY 2 for chart

    Diagram30.536.3Diagram4.34.42.152.2

    Quadrant31.831.5Quadrant4.54.32.252.15

    Table36.547.2Table4.752.352.5

    DFA_NotGivens

    002.152.152.22.2

    002.252.252.152.15

    002.352.352.52.5

    Simple Explanations

    Elaborated Explanations

    Problem Format

    Percent Correct

    Performance by Problem Format & Explanation Type

    Sheet3

  • DFA Results: Problem SolvingLinear trend for Explanation Type, F (1, 88) = 2.9, p = .09Quadratic effect for Problem Format, F (1, 88) = 3.8, p = .053Trend for interaction, F (1, 88) = 3.0, p =.088

    Chart6

    30.536.32.152.152.22.2

    31.831.52.252.252.152.15

    36.547.22.352.352.52.5

    Simple Explanations

    Elaborated Explanations

    Problem Format

    Percent Correct

    Performance by Problem Format & Explanation Type

    DFA_Givens

    Means - GivensStdErrs_Givens

    Simple ExplanationsElaborated ExplanationsSimple ExplanationsElaborated ExplanationsDIV BY 2 for chart

    Diagram86.590.4Diagram3.431.71.5

    Quadrant88.892.7Quadrant3.33.61.651.8

    Table88.289.3Table3.33.21.651.6

    DFA_Givens

    001.51.51.71.7

    001.81.81.651.65

    001.61.61.651.65

    Simple Explanations

    Elaborated Explanations

    Problem Format

    Percent Correct

    Performance by Problem Format & Explanation Type

    DFA_NotGivens

    Means - GivensStdErrs_Givens

    Simple ExplanationsElaborated ExplanationsSimple ExplanationsElaborated ExplanationsDIV BY 2 for chart

    Diagram30.536.3Diagram4.34.42.152.2

    Quadrant31.831.5Quadrant4.54.32.252.15

    Table36.547.2Table4.752.352.5

    DFA_NotGivens

    002.152.152.22.2

    002.252.252.152.15

    002.352.352.52.5

    Simple Explanations

    Elaborated Explanations

    Problem Format

    Percent Correct

    Performance by Problem Format & Explanation Type

    Sheet3

  • Preliminary Results: ProcessObservational pilot dataLonger latency of responses in table condition BEFORE entering quantitiesLonger latencies AFTER quantities entered when elaborated explanations are required Classroom FeedbackTeachers report student preference for diagram tutorStudents report no perceived differences in the amount of work for the elaborated explanationsStudents adapt quickly to the elaborated explanations, but performance far from ceiling even after successful completion of tutor with simple explanations.

  • Next StepsLog files??????!!!!Think-aloud protocols with elaborated explanationsSummer 2006Lab testing of elaborated explanations Summer 2006In-vivo testing with the elaborated explanations & contiguous interface (2 X 2)Late Fall 2006

  • Research TeamVincent Aleven: Research Scientist, CMU HCIIKirsten Butcher: Research Postdoc, Pitt LRDCShelley Evenson: Assoc Prof, CMU School of DesignOctav Popescu: Research Programmer, CMU HCIIAndy Tzou: Masters Student: CMU HCII Honors ProgramCarl Angiolillo: Masters Student: CMU HCII Honors ProgramGrace Leonard: Research Associate, CMU HCIIThomas Bolster: Research Associate, CMU HCII

  • Questions?

  • Extra Slides

  • Existing Tutor: Multiple Verbal Inputs

  • Existing Tutor: Multiple Visual Inputs

  • Table Condition = Noncontiguous

  • Diagram Condition = Contiguous

  • Methods: Contiguity (Study 1)Geometry Cognitive Tutor: 2 conditionsTable (noncontiguous)Diagram (contiguous)

    ProcedurePretest (in class)Training (classroom use of tutor, grade-matched pairs randomly assigned to conditions within classes)Posttest (in class)

  • Assessment: 3 types of itemsAnswers

  • Assessment: 3 types of items

  • Assessment: 3 types of itemsTransfer

  • Preliminary Results: AnswersMain effect of test time: F (1, 38) = 29.5, p < .01

    Chart7

    23281028

    50563548

    Table Low

    Table High

    Diagram Low

    Diagram High

    Test Time

    % Correct

    Higher and Lower Ability Students' Performance on Answers (Solvable)

    Sheet1

    ANSWERSPretestPosttesta-pre-stdva-pst-stdv

    TableLow23502332

    High28562232

    DiagramLow10351728

    High28482523

    REASONSPretestPosttestr-pre-stdvr-pst-stdv

    TableLow0.130.320.150.26

    High0.130.570.160.28

    DiagramLow0.080.370.140.24

    High0.190.480.230.25

    Transfer-PRETransfer-POSTt-pre-stdvt-pst-stdv

    TableLow0.420.110.350.25

    High0.10.350.170.36

    DiagramLow0.060.230.160.25

    High0.230.270.340.39

    Sheet1

    23281028

    50563548

    Table Low

    Table High