Modern Learning Theories and Mathematics Education - Robert Siegler

Modern Learning Theories and Mathematics

Education

The research reported here was supported by the Institute of Education Sciences, U.S. Department of Education, through Grant R305H050035 to Carnegie Mellon University. The opinions expressed are those of the author and do not represent views of the Institute or the U.S. Department of Education.

Like many developmental psychologists, most of my early research was theoretical (Definition: “Without any likely application”)

Over time, my kids (and granting agencies) motivated me to think harder about ways in which the research could be applied to important educational problems without sacrificing rigor

One outcome has been my current research applying theories of numerical cognition to improving low-income preschoolers’ mathematical understanding

A Little Personal Background

Another outcome has been to increase my interest in broader issues of application, i.e., educational policy issues

This growing interest in applications led me to abandon my traditional “just say no” policy regarding commissions and panels and accept appointment to the National Mathematics Advisory Panel (NMAP). Main role was in learning processes group

The present talk combines perspectives gained from doing the applied research and from participating in the learning processes group of NMAP

“9. Encouraging results have been obtained for a variety of instructional programs developed to improve the mathematical knowledge of preschoolers and kindergartners, especially those from low-income backgrounds. There are effective techniques – derived from scientific research on learning – that could be put to work in the classroom today to improve children’s mathematical knowledge.”

“14. Children’s goals and beliefs about learning are related to their mathematics performance. . . When children believe that their efforts to learn make them ‘smarter,’ they show greater persistence in mathematics learning.”

Conclusions of NMAP:

A basic issue in many modern learning theories involves how knowledge is represented

In mathematical cognition, this issue involves the underlying representation of numerical magnitudes (Dehaene, 1997; Gelman & Gallistel, 2001; Case & Okamoto, 1996)

Empirical research indicates that linear representations linking number symbols with their magnitudes are crucial for a variety of important mathematics learning outcomes

Theoretical Background: The Centrality of Numerical Magnitude Representations

The Number Line Task

0 100

71

NumberPresented

NumberPresented

NumberPresented

Progression from Log to Linear Representation — 0-100 Range

(Siegler & Booth, 2004)

Progression from Log to Linear Representation — 0-1,000 Range

(Siegler & Opfer, 2003)

Sixth Graders

Number Presented

R2lin = .97

Number Presented

Med

ian

Est

imat

e

Second Graders

R2log = .95

Linearity of magnitude representations correlates positively and quite strongly across varied estimation tasks, numerical magnitude comparison, arithmetic, and math achievement tests (Booth & Siegler, 2006; 2008; Geary, et al., 2007; Ramani & Siegler, 2008; Whyte & Bull, 2008).

The Centrality of Numerical Magnitude Representations

Correlations Among Linearity of Magnitude Representations on Three Estimation Tasks

(Booth & Siegler, 2006)

Grade Task Measurement Numerosity

2nd Number line .65** .55*

Measurement .54**

4th Number line .84** .70**

Measurement .60**

**p < .01; *p < .05

Correlations Between Linearity of Estimation and Math Achievement

(Booth & Siegler, 2006)

Number Line Measurement Numerosity

2nd .53** .62** .48**

4th .47* .54** .35

Estimation TaskGrade

**p < .01; *p < .05

Causal Evidence: External Magnitude Representations and Arithmetic Learning

(Booth & Siegler, 2008)

0

25

50

Pretest Posttest% C

orre

ct A

rith

met

ic A

nsw

ers

Feedback + Magnitude RepresentationFeedback **p < .01

**

% C

orre

ct S

ums

Issue in Mathematics Education: Low-Income Children Lag Behind in

Mathematical Proficiency Even Before They Enter School

1. Children vary greatly in mathematical knowledge when they enter school

2. Numerical knowledge of kindergartners from low-income families trails far behind that of peers from higher-income families (ECLS, 2001)

3. Kindergartners’ numerical knowledge strongly predicts later mathematical achievement — through elementary, middle, and high school (Duncan, et al., 2007; Jordan et al., 2009; Stevenson & Newman, 1986)

4. Large, early, SES related differences become even more pronounced as children progress through school

Applying Theory to Educational Problem

Might inadequate representations of numerical magnitudes underlie low-income children’s poor numerical

performance?

• Counting experience is likely helpful and necessary, but insufficient

• Children can count in a numerical range more than a year before they can generate a linear representation of numerical magnitudes in that range (Condry & Spelke, 2008; LeCorre & Carey, 2007; Schaffer et al., 1974)

Applied Goal Raised New Theoretical Question: What Leads Anyone to Form

Initial Linear Representation?

• Board games might play a crucial role in forming linear representations of numerical magnitudes

• Designed to promote interactions between parents and peers

• Also provides rich experiences with numbers

Playing Board Games

Chutes and Ladders

The greater the number a token reaches, the greater the• Distance that the child has moved the token • Number of discrete hand movements the child

has made• Number of number names the child has spoken• Time spent by the child playing the game

Thus, playing number board games provides visuo-spatial, kinesthetic, auditory, and temporal cues to links between number symbols and their magnitudes

Key Properties of Board Games Like Chutes and Ladders

Number Board Game

Color Board Game

Goal was to investigate whether playing the number board game:

• Improves a broad range of numerical skills and concepts

• Produces gains that remain stable over time

Effects of Playing the Number Board Game

(Ramani & Siegler, 2008)

Participants: 129 4- and 5-year-olds from Head Start classrooms (mean age = 4.8), 52% African-American

Experimental Conditions:• Number Board Game (N = 69)• Color Board Game (N = 60)

Design: Pretest, 4 training sessions, posttest, 9 week follow-up.

Methods

Training Procedure:• Children play a total of 20 games over 4 sessions

in a 2 week period, 15-20 minutes/session• Child spins spinner, gets 1 or 2, says while

moving token (e.g.) “5, 6” or “blue, red”• Feedback and help if needed

Measures:

• 0-10 Number Line Estimation

• 1-9 Numerical Magnitude Comparison

• 1-10 Counting

• 1-10 Numeral Identification

50

60

70

80

90

100

Pretest Posttest Follow-up

Number Board Game Color Board Game

Numerical Magnitude ComparisonM

% C

orr e

ct **

*p < .001

6

7

8

9

10

Pretest Posttest Follow-up

Number Board Game Color Board Game

CountingM

Mea

n C

ount

s W

itho

ut E

rror **

*p < .001

0

10

20

30

40

50

Pretest Posttest Follow-up

Number Board Game Color Board Game

Number Line Estimation:Linearity of Individual Children’s

EstimatesM

Mea

n R

2 lin *

*

*p < .001

50

60

70

80

90

100

Pretest Posttest Follow-up

Number Board Game Color Board Game

Numeral IdentificationM

**

% C

orr e

ct

*p < .001

0

10

20

30

40

50

Pretest PosttestLinear Board GameCircular Board GameControl Numerical Activities

Percent Correct Addition Answers(Siegler & Ramani, 2009)M

% C

orre

ct

*p < .05

*

Experiment 2: Game Playing Outside the Lab

Participants: 145 4- and 5-year-olds• 115 children from prior Head Start sample• 30 children recruited from middle- to upper-

middle income familiesProcedure: Informal activities questionnaire

• Board games, card games, & video games played outside of school

• Name the different games

10

30

50

70

90

Board Games Card Games Video Games

Head Start Children Middle-Income Children

Percent of Children Who Had Played Each Type of Game

M

*

*%

of

Chi

ldr e

n*

*p < .01

Correlations Between Game Playing Experience and Numerical Knowledge Among Head Start Children

M

Number Line Linearity

Magnitude Comparison

Counting

Numeral Identification

No. of Board Games .38** .26** .20* .25**

No. of Card Games .18 .28** .11 .13

No. of Video Games .21* .02 .00 .07

**p < .01; *p < .05

Correlations Between Playing Chutes and Ladders and Numerical Knowledge

M

Number Line Linearity

Magnitude Comparison

Counting

Numeral Identification

Played Chutes & Ladders .20* .18 .19* .24*

*p < .05

Young students in East Asia and some European countries spend more time on math, encounter more challenging and conceptually richer curricula, and learn more. No reason why we can’t do the same. Belief that young children aren’t ready to learn relatively advanced concepts contradicts both national and international data.

Conclusion 15 from NMAP:“Teachers and developers of instructional materials sometimes assume that children need to be a certain age to learn certain mathematical ideas. However, a major research finding is that what is developmentally appropriate is largely contingent on prior opportunities to learn. Claims that children of particular ages cannot learn certain content because they are too young have consistently been shown to be wrong.”

Conclusion 9 from NMAP:“There are effective techniques — derived from scientific research on learning — that could be put to work in the classroom today to improve children’s mathematical knowledge.”

Funding agencies have generously supported research on learning principles and on small scale programs that implement these principles. As always, we need more research, but some of the research is now sufficiently advanced for broad implementation, at least on an experimental basis. The challenge for the field of mathematics education is how to use the programs and principles to improve educational practice.