117 views

Uploaded by api-21244456

- implementing new initiatives for mathematics
- Lesson Plan Fil-5
- raise the bar on problem solving
- 4.Participation
- Towards a Problematique for Research on Mathematics Teaching
- gr1 add subtract within 20 planning document
- math welcome newsletter
- Employability Skills What Are They
- 1-s2.0-S187704281503102X-main
- Educational Psychology
- thematicunit template-final
- M1L1-1
- 1
- year 6 t1 unit 3 mathematics term
- Mathematics 1
- cte anchor standards 2
- PPG Self Directed Workforce
- An Investigation of the Pre-Services Tea
- module three session one - ict to support teaching learning and assessment through numeracy
- Note on Consumer Decision Making Processes

You are on page 1of 10

Subtraction Word Problems

Gordon B. Willis and Karen C. Fuson

School of Education and Social Policy

Northwestern University

Two classes of second graders of average and above-average mathematics ability were taught to

use differing schematic drawings to represent differing categories of addition and subtraction

word problems. Children entered the three-digit numbers used in the problems into the schematic

drawings and then were to use the drawings to facilitate the choice of the solution procedure.

The children were able to make the correct drawing for a given category, usually inserted the

numbers from the problem into a schematic drawing correctly, and usually selected the correct

solution strategy for the problem. There was little support for the hypotheses that children use a

single part-part-whole schema to solve either all categories of problems or the more difficult

"Change" problems. The most difficult problems were those in which the underlying semantic

subtractive problem category ("Change-Get-Less" and "Compare") conflicted with the addition

solution strategy required to solve the problem. The good-to-excellent posttest performance on

most of the possible kinds of addition and subtraction word problems indicates that most of

these problems are within the zone of proximal development of second graders of average and

above-average mathematics ability. Thus American textbooks can include many of the more

difficult word problems, as do textbooks in the Soviet Union.

Solving addition and subtraction word problems involves reflect on that problem representation and modify it in some

at least three aspects: representing the word problem situation, way in order to select a solution strategy (Briars & Larkin,

selecting a solution strategy, and using the solution strategy 1984; Riley et al, 1983). A teaching method that helps

to find the answer. Initially, children solve word problems by children to represent the problem situation would be more

representing the problem with concrete objects and then using helpful than the prevalent solution sentence method.

these objects to carry out the solution strategy (Briars & In developing and testing such a method, we decided to

Larkin, 1984; Carpenter & Moser, 1984; Fuson, 1988; Riley, focus initially on the upper end of problem solving—that

Greeno, & Heller, 1983). Later, children solve problems by involving more difficult types of problems and multidigit

using more sophisticated counting strategies that also are numbers—because the need for representational support

directly derived from the representation of the problem situ- seems to be most crucial here: The more difficult problems

ation (Carpenter & Moser, 1984; Fuson, 1988). Finally, chil- require reflection on the representation, and the large num-

dren solve problems by choosing an arithmetic operation bers require a separate selection of the solution strategy. The

(addition or subtraction) and then using some particular teaching approach that we chose was to teach children to

method of adding or subtracting such as thinking strategies, represent word problems by making a schematic drawing that

known facts, or the multidigit addition or subtraction algo- models the semantic features of the problem situation; the

rithms (Carpenter & Moser, 1984; Fuson, 1988). Thus the numbers in the problem can then be written in this schematic

first two aspects of problem solving may be merged for small drawing and the drawing used to decide whether to add or

numbers or simple types of problems, but they are separate subtract to find the missing problem element.

for large numbers because these require the choice of an Understanding the schematic drawings used requires un-

algorithm. derstanding the different types of addition and subtraction

The most common method of teaching addition and sub- word problem situations. These are commonly divided into

traction word problems ignores children's need to represent four major categories (see Table 1). Two of these are basically

the problem situation and instead focuses only on the solution additive. In the Change-Get-More category, some initial

strategy: Children are taught to write a solution addition or quantity gets some more added to it; in the Put-Together

subtraction sentence (e.g., 8 + 5 = ? or 8 - 5 = ?) for a category, two separate quantities are put together to form one

problem and then are to write the answer for the sentence. combined quantity. The other two categories are basically

The disadvantage of this approach is particularly strong for subtractive. In the Change-Get-Less category, there is a quan-

the more complex kinds of word problems, for these require tity from which some quantity is taken; in the Compare

not only that children represent a problem but also that they category, two quantities are compared in order to find out

how much greater one quantity is than another. A fifth

This research was supported by a grant from the Amoco Founda- category, Equalize, is identified by some researchers; these

tion to the University of Chicago School Mathematics Project. problems are Compare problems that explicitly mention a

Correspondence concerning this article should be addressed to change needed to make the two original quantities equivalent.

Karen C. Fuson, School of Education and Social Policy, Northwestern Very little research has been done on this category, so we did

University, Evanston, Illinois 60208. not use it in this study. The names of the problem types in

192

TEACHING SCHEMATIC DRAWINGS 193

Table 1

Classification of Word Problems

Addition situations Subtraction situations

Change-Get-More Change-Get-Less

Missing end Missing end

Joe had 3 marbles. Joe had 8 marbles.

Then Tom gave him 5 more marbles. Then he gave 5 marbles to Tom.

How many marbles does Joe have now? How many marbles does Joe have now?

Missing change0 Missing change

Joe had 3 marbles. Joe had 8 marbles.

Then Tom gave him some more marbles. Then he gave some marbles to Tom.

Now Joe has 8 marbles. Now Joe has 3 marbles.

How many marbles did Tom give him? How many marbles did he give to Tom?

Missing start3 Missing start*

Joe had some marbles. Joe had some marbles.

Then Tom gave him 5 more marbles. Then he gave 5 marbles to Tom.

How many marbles did Joe have in the beginning? Now Joe has 3 marbles.

How many marbles did Joe have in the beginning?

Put-Together Compare

Missing all Missing difference

Joe has 3 marbles. Joe has 8 marbles.

Tom has 5 marbles. Tom has 5 marbles.

How many marbles do they have altogether? How many more marbles does Joe have than Tom?

a

Missing first part Missing big3

Joe and Tom have 8 marbles altogether. Joe has 3 marbles.

Tom has 3 marbles. Tom has 5 more marbles than Joe.

How many marbles does Joe have? How many marbles does Tom have?

a

Missing second part Missing small

Joe and Tom have 8 marbles altogether. Joe has 8 marbles.

Joe has 3 marbles. He has 5 more marbles than Tom.

How many marbles does Tom have? How many marbles does Tom have?

Note. These examples are adopted from Riley, Greeno, and Heller's (1983) example problems. Compare problems can also be asked with the

words "less" and "fewer" rather than "more," and with Equalize questions containing either "less"/"fewer" or "more" (e.g., "How many more

marbles does Tom have to get to have as many as Joe?").

a

In these problems, there exists a conflict between the overall category situation as additive or subtractive and the operation required to solve

a particular problem subtype.

Table 1 were chosen to be comprehensible to the second- a final state. Therefore the same basic drawing was used for

grade children participating in this study. these two types, and the different nature of the change was

Because all problems involve three quantities and any of represented by a plus or minus symbol inserted by the child

these quantities can be unknown, there are three possible (see Figure 1). The Put-Together drawing was the part-part-

problem subtypes within each main problem type (see Table whole drawing used in the Developing Mathematical Proc-

1). Two of these require subtraction of the two given numbers esses series (Romberg, Harvey, Moser, & Montgomery, 1974).

in the problem and one requires addition of the two givens. However, the quantities were labeled part, part, and all (rather

Some of the problems created by particular unknowns may than whole) because often there is no actual "whole" that

be especially difficult because of the internal conflict between comprises the two given sets in a Put-Together problem. The

the basic semantic structure of the underlying situation as Compare drawing contained big and small quantities placed

additive or subtractive and the solution strategy (addition or adjacent to one another to facilitate their comparison; the

subtraction) required to solve a given problem with a partic- difference was enclosed by a broken line because that differ-

ular kind of unknown (DeCorte & Verschaffel, 1985a; Greer, ence is not actually a physically separate entity in a Compare

in press; Vergnaud, 1982; Willis & Fuson, 1985). For example, structure. The three quantities in each problem were given

in a Change-Get-More situation, a missing Start problem has mnemonic labels (see Figure 1). Children were to write the

the additive Change-Get-More underlying structure, but one letter for each label on the known or unknown quantity in

can solve it by subtracting the change number from the end the problem and then enter the known quantities into their

number. The conflict problems are footnoted in Table 1. correct place in the drawing. The drawing could then be used

Conflict problems have been found to be particularly difficult to determine the correct solution strategy. The choice of

for children (see Carpenter & Moser, 1983, and Riley et al., solution procedure was aided by the presence of the drawing

1983, for reviews). Therefore, of particular interest was either through the relations among the physical sizes of the

whether the schematic drawings would improve performance different parts or by the temporal ordering implied by the

on conflict problems. drawing.

Three different kinds of schematic drawings were used in The ultimate question, of course, is whether teaching sche-

the word problem instruction (see Figure 1). The Change- matic drawings improves children's ability to solve word

Get-More and Change-Get-Less problem structures are simi- problems. The prior question is whether children can even

lar; both problem types involve an initial state, a change, and learn this method. Because of the complexity of this method,

194 GORDON B. WILLIS AND KAREN C. FUSON

PUT-TOGETHER (COMBINE): missing SECOND PART problem types stem from difficulty in representing these prob-

(All) (Part) (Part) lems at a global level (indicated by children's making an

Jon and Bill have 814 toys altogether. Jon has 342 toys. How many toys inappropriate drawing), in understanding the specific relations

does Bill have? p p

among the three problem quantities (indicated by children's

342 inserting the three numbers into the drawing incorrectly), in

choosing a correct solution strategy (indicated by children's

adding instead of subtracting or vice versa), or in carrying out

814 the solution strategy correctly (indicated by children's making

A an addition or subtraction error).

Schematic drawings also provide an opportunity to obtain

CHANGE-GET-MORE; missing START information about how children represent problems, infor-

(Start) (Change) (End) mation that is not available when children solve problems by

Jon had some toys. Then Bill gave him J42 more toys. Now Jon has 814 toys. usual methods. This information is particularly pertinent to

How many toys did Jon have to start with? two hypotheses concerning such representations. Resnick

C (1983) proposed that children use a single part-part-whole

s £342 ) E schema, such as the Put-Together drawing, for the represen-

tation of all four kinds of addition and subtraction word

^ -^ 814 problems. Riley et al. (1983) and Kintsch and Greeno (1985)

hypothesized that children use a part-part-whole schema to

solve the more difficult kinds of Change problems. The former

proposal would be supported by a tendency for children to

HANG E-GET-LESS: missing CHANGE

make the Put-Together drawing for all types of problems and

(Start) (Change) (End)

Jon had 814 toys. Then he gave some toys to Bill. Now Jon has 342 toys. for this drawing to facilitate their solution. The latter proposal

How many toys did Jon give to Bill? would be supported by such a tendency for the more difficult

Change problems.

E A final focus of the study was to ascertain which of the 12

problem types are within the zone of proximal development

814 342 of second graders when multidigit numbers are used in the

problems (i.e., which types second graders can learn with

instructional support). There are at present no data on this

point. Children in the United States typically are infrequently

COMPARE: missing BIG exposed to the more difficult problem types with either small

(Small) (Difference) or large numbers (Stigler, Fuson, Ham, & Kim, 1986), and

Jon has 342 toys. Bill has 472 more toys than Jon has.

How many toys does Bill have? research on children's ability to solve the more difficult types

(Big)

S D has been limited to sums and differences to 18 (Carpenter &

Moser, 1983; Riley et al., 1983). However, a comparative

342 472 i analysis of mathematics textbooks suggests that second grad-

ers might be able to solve the easier types with large numbers

and could perhaps improve in solving some of the more

difficult types. The mathematics series used nationwide in the

Soviet Union includes for both first and second grades all 12

kinds of problems with a distribution that is almost equal

Figure 1. Examples of word problems with verbal labels applied across the 12 types (Stigler et al., 1986). The second-grade

and drawings as they would be filled before selection of a solution

procedure. text includes problems with two-digit numbers, and texts for

both grades include many two-step problems that involve two

different categories of problems (Stigler et al., 1986, and

this question breaks into several questions, each of which was unpublished data from this study). Because Soviet children

focused on in the data collection and analysis. Can children begin school 1 year later than children in the United States,1

make the schematic drawings? Do they choose the correct it seemed possible that the difficult problems with large num-

category of drawing for a problem (e.g., make Put-Together bers would be too difficult for low-ability second graders.

drawings for Put-Together problems)? If not, what are the Therefore this study was restricted to United States second

major kinds of substitutions of one category for another? Do graders who were of average and above-average ability in

they enter the numbers into a drawing so that it represents mathematics.

the particular problem correctly? Do children choose the It was possible to use multidigit numbers in this study

correct solution strategy? How are the problem representa- because the children in our sample had previously been taught

tions related to the choices of solution strategy? Is the solution

strategy carried out correctly—that is, does it lead to the 1

The Soviet Union has now started to admit first graders at age 6.

correct answer to the problem? The schematic drawings thus However, the text series containing all types of word problems was

enable us to pinpoint whether low rates of solving certain designed for 7-year-old first graders.

TEACHING SCHEMATIC DRAWINGS 195

algorithms for solving multidigit addition and subtraction drawings for typical Change and Compare stories are also provided

problems (Fuson, 1986a). The children used finger-pattern in Figure 1.

counting procedures for performing single-digit computations Worksheets and tests, Practice worksheets were constructed both

when an addition or subtraction fact was unknown (Fuson, for student practice and to provide data on problem difficulty. Pure

1986b; Fuson & Secada, 1986; Fuson & Willis, in press); they worksheets were those containing problems from a single semantic

were taught to count up rather than down for subtraction category. The Put-Together, Change-Get-More, and Change-Get-Less

problems (e.g., 1 4 - 8 was solved by counting up from 8 to worksheets each contained 18 problems equally distributed across

each of the three missing positions. The Compare worksheets con-

14). These children were in general quite competent at single- tained 27 problems distributed equally across the three missing posi-

digit and multidigit computation. tions and across problems in which the words more and fewer

appeared (e.g., "How many fewer problems does Joe have than

Tom?"). The mixed worksheets contained 24 intermixed problems

Method equally distributed across the four semantic categories and across the

three missing positions for each type. The equal distribution across

problem variables permitted analyses in which we compared the

Subjects relative difficulty of different problem types and different missing

positions. However, it was not an instructionally maximal treatment,

Subjects were second-grade children from two public schools in a

which would have concentrated on problems with which children

small city near Chicago. The schools in this city enroll children of

were having the most difficulty.

widely varying socioeconomic status. One of the classes contained 24

students and was categorized by the teachers as containing children The 10 most difficult problem subtypes were selected for inclusion

with high math ability (Class HA). The other class contained 19 in 10-problem pretests and posttests given to children before and after

children categorized by teachers as children of average math ability the teaching. These tests omitted the simple "Put-Together: missing

(Class AA). all" and the "Change-Get-More: missing end" problems because we

were concerned about fatigue and time limitations that would have

occurred with a longer test. The easiest problems were omitted

Procedure because we were concerned about ceiling effects on these problems.

In all of the Compare problems, the word more, rather than fewer,

Use of the schematic drawings. The categories of word problems was used. Problems were ordered so that no two problems from the

used in the study appear in Table 1 and the schematic drawings used identical semantic category (Put-Together, Change, Compare) were

for each category of word problems appear in Figure 1; both were placed sequentially. To control for problem ordering effects, we

discussed in the introduction. Children were first introduced to a constructed two forms in which problems were listed in the reverse

general category of word problems and shown the drawing for that order. All numbers in problems were of three digits, and one trade

category. For any given problem, they were taught first to write on (carrying or borrowing) was required to obtain the sum or difference.

the problem the letter for the verbal labels naming the three elements Children were given the test initially as a pretest and allowed to solve

in the story (the two givens and the unknown). The particular labels problems in any way. When the test was given as a posttest, children

in Figure 1 were substituted for Riley et al.'s (1983) labels because were told to make a schematic drawing for each problem. Each

they were more appropriate for second graders. problem was then evaluated with respect to whether the drawing

After applying the labels, the children were taught to make the made was of the correct problem category, the numbers were filled

schematic drawing that is appropriate for that general story category. into the drawing correctly, the correct solution strategy was chosen,

The drawing contained three parts, one corresponding to each of the and the solution strategy was carried out correctly.

labeled story elements. The two given numbers were then written in Organization of the instruction. Teaching was divided into units;

the appropriate parts of the drawing; these locations were determined one major problem category was taught within a unit. The order of

by the labels attached to the numbers. Last, the children chose the teaching was Put-Together, Change-Get-More, Change-Get-Less,

correct arithmetic solution procedure by determining how to obtain Compare, and mixed problem categories. The first unit taught to

the unknown (either by adding or subtracting), given the two filled Class HA focused first on small-number problems (sums and differ-

parts. The drawing facilitates this choice either by the relative sizes of ences to 18), and small-number worksheets were completed before

the sections of the drawing (for Put-Together and Compare problems, large-number problems were introduced. This was done because we

"small + small = big" and "big — small = small") or by the temporal were afraid that beginning with the large numbers would be too

ordering of events (for Change-Get-More and Change-Get-Less prob- difficult. For the first unit, children were also instructed to write a

lems, "some + some more = a larger result" and "some — some of number sentence that reflected the semantic structure of the problem

them = a smaller result"). (e.g., 6 + = 14 or + 342 = 629) before solution. It

An exampleof how this series of steps is applied to a Put-Together became evident, however, that the writing of the number sentence

problem is provided in the first part of Figure 1. The child identifies problem equation was a burdensome and much-resisted extra step at

the problem as a type from the Put-Together category and applies the both number sizes and that children tended to write the equation

appropriate verbal labels (part, part, and all) to the word problem, after solution or else omitted it entirely. Therefore, this step was

writing P, P, and A by the appropriate words. The first part (P) is bypassed in further units.

represented by Jon, the second and unknown (P) by Bill, and the all All teaching was done by one of the investigators. In one math

(A) by the given altogether. The next step is to produce the drawing, period, story problems of a particular category were presented, the

and the following step is to fill in the given elements. The child then classification and labeling were described, and the complete solution

analyzes the relations between the two known numbers in the drawing for different variants of the problem was illustrated by the instructor.

and arrives at the subtraction solution to obtain the unknown (the Then the children spent between 2 and 4 days completing the practice

number that is to fill the empty box). Thus the physically represented worksheets for that category. Children received little systematic indi-

"part + part = all" structure is potentially capable of inducing the vidualized feedback on the worksheets because these were kept for

proper solution strategy as long as the child has filled in the sections purposes of analysis. As much as possible, particular difficulties that

of the drawing correctly. Examples of the use of the labels and children had were addressed initially as the instructor aided them

196 GORDON B. WILLIS AND KAREN C. FUSON

individually on worksheets and later when additional brief classroom problems (i.e., on the Class HA mixed worksheets), the per-

instruction was used to rectify general problems that became apparent centages of correct drawing selection were fairly similar across

through scrutiny of childrens' behavior in solving worksheet prob- the four problem categories: The correct drawing was made

lems. The initial teaching of each problem type was restricted to the

story-problem category under current consideration. When a new

roughly three fourths of the time. The most frequent mistake

category was introduced, the differences rather than similarities be- was making the Compare drawing for Put-Together problems.

tween it and previous categories were emphasized. All of these were made for the "Put-Together: missing-part"

After all four categories of problems had been introduced and problems; none were for the "missing all" problems. Thus

practiced, a short unit was given on problems of mixed categories. some HA children seem to interpret a Put-Together problem

For these problems, children had to decide which major category of with a known whole and known part as a comparison between

problem a particular story represented and then make that drawing. the whole and the part. Also somewhat frequent were Put-

Again, the particular defining features of each of the main categories Together drawings made for Change-Get-More problems.

were stressed. These indicate that the Compare drawings made for Put-

For the AA class, the Change-Get-More and Change-Get-Less units Together problems were not simply a result of an overall

were aggregated because transfer from a Get-More to a Get-Less preference for making the Compare drawing.

structure was very rapid. No instruction on small numbers was given

because the HA children had coped quite well with the larger num-

The HA class posttest drawing selections showed improve-

bers. Unfortunately, the second class had to be taught near the end ment over those made for the worksheets in every category

of the school year when unexpected activities intruded. Consequently, except Put-Together. The percentages of correct selections of

insufficient time was available for practicing the separate units and Change and Compare drawings were very high. Because the

especially for completing the worksheets of mixed problem categories. Put-Together and Compare drawings are so similar, an alter-

These children completed on the average only half of the mixed native scoring gave credit for Compare drawings made for

problems and had no time to receive a review that stressed distinctions Put-Together problems and vice versa. When this was done

among the problem types. (see the PT/Cm percentages in Table 2), the score for Put-

Together problems increased to the high level of the other

Results problem categories. As on the worksheets, this tendency to-

ward mislabeling was asymmetric: Compare drawings were

Selection of the Appropriate Drawing made for Put-Together problems much more than the reverse

(24% vs. 6%). All of the Put-Together problems on the posttest

Children were able to make all of the schematic drawings were of the "missing part" type, and the percentage of Com-

used in the study. They also were fairly good at selecting the pare substitutions for Put-Together problems almost doubled

appropriate drawing (i.e., at making the drawing that matched from the worksheets to the posttest, from 13% to 24%. Thus

the word problem category). The percentages of correct draw- the tendency for these children to view "Put-Together: missing

ing selection are given in Table 2 for the posttests and for the part" problems as "Compare: missing difference" problems

Class HA mixed worksheets (the Class AA mixed worksheets certainly did not decrease with practice.

were not analyzed because so few children had time to com- The overall level of posttest correct drawing selection was

plete them). When the HA children werefirstpracticing mixed considerably lower for Class AA than for Class HA. When

Table 2

Percentage of Correct Performance by Problem Category on Measures of Problem Representation

Drawing selection

Drawing Overall Solution

Measure Correct PT/Cm Fill-in representation strategy

Put-Together

Class HA mixed worksheets 75 88 72 84 92

Class HA posttest 69 93 88 93 95

Class AA posttest 71 93 79 84 92

Change-Get-More

Class HA mixed worksheets 72 65 81 84

Class HA posttest 90 83 93 95

Class AA posttest 79 67 79 83

Change-Get-Less

Class HA mixed worksheets 83 72 79 81

Class HA posttest 92 84 87 83

Class AA posttest 61 61 81 69

Compare

Class HA mixed worksheets 83 89 56 71 73

Class HA posttest 89 95 74 78 87

Class AA posttest 72 94 61 69 69

Note. Drawing fill-in is the percentage of correct drawings filled in correctly; it includes Put-Together and Compare substitutions for each other.

Overall representation includes any drawing in which the placement of numbers would lead to the correct solution strategy. HA = high ability;

AA = average ability; PT = Put-Together; Cm = Compare.

a

These percentages include those cases in which Compare drawings were selected for Put-Together problems and vice versa.

TEACHING SCHEMATIC DRAWINGS 197

Put-Together drawings made for Compare problems and vice Analysis of the Class HA worksheets by missing position

versa were scored as correct, selection of a correct drawing indicated that four problem subtypes were particularly diffi-

rose above 90% for these two categories. The substitutions of cult for children to represent in this initial learning period.

these two types for each other were symmetric for this class, The "Change-Get-More: missing end," "Change-Get-Less:

suggesting a more general confusion between these similar missing start," "Compare: missing big," and "Compare: miss-

drawings. Correct drawings were made for Change-Get-Less ing small" problems had more than twice as many incorrect

problems markedly less often than for the other three problem overall representations as did the other subtypes within these

categories; Put-Together drawings were often made for these main types. On the HA posttest, "Compare: missing big" and

problems. "Compare: missing small" were the most difficult to represent;

"Compare: missing difference," "Change-Get-More: missing

start," and "Change-Get-Less: missing start" were somewhat

Filling in the Drawing more difficult than the remainder of the problem subtypes.

On the AA posttest, differences in representation were spread

Drawings were scored as correctly filled in if the correct fairly equally across missing positions and main categories,

semantic type was drawn and the numbers were placed cor- though "Compare: missing big" and "Compare: missing dif-

rectly within the drawings, with the exception that substitu- ference" were the problem subtypes represented least well.

tions of Compare and Put-Together drawings were allowed Thus there was some difficulty in representing all of the

because of their considerable similarity. Percentages of correct Compare missing positions, but the "missing big" problem

drawing fill-in are given in Table 2. For Class HA, the overall was particularly difficult to represent.

average correct fill-in was 66% on the worksheets. This rose

to 82% correct on the posttest, indicating considerable learn- Use of Correct Solution Strategy

ing from the worksheet practice and the review. On both

worksheets and the posttest, performance on the Compare A solution strategy was scored as addition if the answer was

problems was somewhat lower than for the other three prob- greater than the two numbers given in the problem; it was

lem categories. The results for Class AA posttests generally scored as subtraction if the answer was less than the larger

paralleled those of Class HA, but were overall about 15% number. The percentages of problems on which children

lower. It is not clear (a) how much of the decrement in this chose the correct solution strategy are given in Table 2. On

class in comparison with the first is due to difference in ability the posttests both classes used correct solution strategies more

and (b) how much was produced by the less-than-adequate frequently on the additive Put-Together and Change-Get-

learning and review time. More situations than on the subtractive Change-Get-Less and

Compare situations, even though the easiest problems in each

of the additive categories were omitted from the posttest. This

Overall Adequate Problem Representation difference between overall additive and subtractive categories

was greater in Class AA than in Class HA. Overall, Class HA

A second analysis of problem representation considered had an impressively high correct choice of solution strategy

only whether the drawing produced by the child adequately (89%), and Class AA had a respectable 77%.

represented the relationship among the problem elements, The pure worksheets completed by Class HA permitted an

independently of the drawing initially made by the child. analysis by missing position within each main category with-

Thus if one would select the correct solution strategy from a out the distractions of other problem types. Put-Together and

filled-in drawing, the drawing was credited as a correct overall Change-Get-More problems showed no effect of missing po-

representation (see Table 2). This analysis was done because sition on correct strategy. Analysis of two pure Change-Get-

one can assimilate a given problem into a different problem Less worksheets revealed significant effects of missing position

structure while still remaining faithful to the relationships in on performance for both, Fs(2, 46) = 15.17 (MSE = 0.15) and

the story; for example, one can think of a Change-Get-More 4.50 {MSE - 0.24), ps < .02, respectively; strategy scores were

problem as a Put-Together problem in which the starting much lower when the missing element was the first one (the

quantity and the change are the "parts" and the end quantity start quantity). Averaged across the two worksheets, the mean

is the "all." We wished to ensure that children could have this performance levels were 63%, 94%, and 90% correct for the

freedom of individual interpretation; the use of drawings was start, change, and end quantities, respectively. Compare work-

intended to be facilitative, not prescriptive. On the posttest, sheets also showed a significant effect of missing position,

only one problem category had more than three instances of Fs(2, 42) = 8.30 (MSB = 0.21) and 5.77 (MSB = 0.19), ps <

a correct overall representation that differed from the problem .03, respectively; the missing big form (68% correct across

category.2 For Class AA there were six instances of a Put- worksheets) was much more difficult than the others (96%

and 91 %, respectively, for missing difference and small). Thus

Together drawing filled in correctly for a Change-Get-Less

problem; these instances were distributed across all three

missing positions. Though the number of correctly filled-in 2

Because the drawing fill-in percentages already include Put-To-

noncanonical drawings was fairly small, such drawings did gether and Compare substitutions for each other, the correct overall

raise the percentages of correct overall representation to 79, representation percentages of course also include a substantial num-

87, and 78 for the HA mixed worksheets, the HA posttest, ber of Compare drawings correctly representing "Put-Together: miss-

and the AA posttest, respectively. ing part" problems.

198 GORDON B. WILLIS AND KAREN C. FUSON

for both Change and Compare problems, it appears that particular problem category: 50% versus 91% for the Class

children have difficulty interpreting these "subtractive" types HA posttest, 50% versus 80% for the Class AA posttest, and

as ones that require an addition, even with the help of the 65% versus 87% for the Class HA mixed worksheets. The

schematic drawing. more specific proposal of Riley et al. (1983) and of Kintsch

and Greeno (1985) that children solve the more difficult

Relationship Between Overall Representation and Change problems by using a Combine (Put-Together) schema

also received little support. Class HA used Put-Together draw-

Correct Solution Strategy ings on only 3% of the Change posttest problems. Class AA

On the posttest, the fit between a correct overall represen- used them more frequently, on 17% of the Change-Get-More

tation and a correct solution strategy was quite good. Most of problems and on 33% of the Change-Get-Less problems.

the time a correct overall representation led to the choice of However, the Put-Together drawing was much less effective

a correct solution strategy: on only 5% and 1% of the prob- than the Change drawing for Change problems. The Change

lems for Class HA and Class AA, respectively, was there a drawing was filled in correctly 93% of the time and led to a

correct solution procedure 85% of the time, whereas the Put-

correct overall representation but an incorrect solution strat-

Together drawings used for Change problems were filled in

egy. Of these, 84% were on conflict problems, those in which

correctly only 50% of the time and led to a correct solution

the underlying semantic problem category conflicts with the only 54% of the time.

solution strategy. Thus the semantic elements in a conflict

problem may stimulate an incorrect solution strategy even if

a proper drawing has been made. Furthermore, an incorrect Pre-Post Improvement

overall representation only rarely, on 8% and 6% of the

posttest problems for Class HA and Class AA, respectively, Mean performance scores on the 10-problem pretests and

led to a correct strategy. For Class AA these cases were posttests for correct strategy and correct answer measures are

distributed across all problem categories. For Class HA, over given by missing position in Table 3. For Class HA, there was

half of such cases were Compare problems, distributed across a significant overall pre-post strategy improvement from 79%

all three missing positions. to 89%, tilff) = 3.09, p < .01. The percentage of correct

answers also improved significantly, from 56% to 72%, t{2§)

Use of the Put-Together Drawings for Other = 2.99,/»<.01. The fairly sizable discrepancy between correct

Problem Categories strategy and answer measures for this class on the posttest was

mainly due to the tendency of many children to attempt to

There was little evidence to support Resnick's (1983) pro- perform the three-digit addition or subtraction straight from

posal that children use a part-whole schema like the Put- the drawing or from the story without first copying the

Together drawing to solve all of the word problem types. On numbers in a vertical form necessary for appropriate use of

the posttest, Class HA showed considerably more use of the the computation algorithms. Large variations in individual

Compare drawing for Put-Together problems than of Put- problem difficulty were observed for this class. Increases in

Together drawings for any non-Put-Together type. When a performance for individual problem subtypes were tested with

child did choose a Put-Together drawing for a non-Put- McNemar's (1954) test of correlated proportions; results are

Together problem, these drawings led to a correct solution shown in Table 3. The strategy scores generally improved

strategy much less frequently than did the drawings for the significantly on all problems for which there was sufficient

Table 3

Percentage of Correct Strategy Use and Correct Answers on the Pretest ("Pre") and Posttest ("Post")

Strategy Answer

Class HA Class AA Class HA Class AA

Problem Pre Post Pre Post Pre Post Pre Post

Put-Together

Missing first part 86 95 67 83 43 86* iS 67*

Missing second part 95 95 58 100* 71 67 58 92

Change-Get-More

Missing start 81 90 67 83 48 81* 58 83

Missing change 57 100* 58 83 33 86* 42 83*

Change-Get-Less

Missing start 52 62 58 42 52 57 42 42

Missing change 95 100 67 83 76 81 6/ 83

Missing end 86 86 67 83 62 71 58 67

Compare

Missing difference 100 100 75 75 67 67 6/ 58

Missing small 86 86 67 92 62 67 58 75

Missing big 52 76* 50 42 43 57 42 42

TEACHING SCHEMATIC DRAWINGS 199

room for improvement except on the "Change-Get-Less: representation and in correct solution strategy than was use

missing start" problem. of the drawings matching the problem category. Thus using

Class AA also made significant overall improvements in three different kinds of drawings seems to be more effective

strategy and answer variables: from 63% to 77% for correct than using just one kind, a result consistent with that of

strategy and from 53% to 70% for correct answer (both ps < Zwerts (cited in DeCorte & Verschaffel, 1985b, and available

.05). The two problem subtypes that required addition showed only after our study was completed).

a small decrease in strategy scores, raising the possibility that Both classes showed improvement from the pretest to the

some children within this class simply tended to subtract on posttest in selecting the correct solution strategy and in finding

all problems at the posttest and therefore improved overall the correct answer. Particular improvement was shown for

on posttest score only because 8 of the 10 problems required those conflict problems with an additive semantic structure

subtraction. This possibility was partly supported by the find- (i.e., the Put-Together and Change-Get-More problems re-

ing that 4 of the 12 children for whom complete posttest and quiring subtraction). However, on worksheets and on the

worksheet data were available did in fact subtract all problems posttest, children showed particular difficulty with those con-

at the posttest, although they did not do so on the worksheets. flict problem types that had an underlying subtractive struc-

An examination of the pretest-posttest strategy scores of the ture (Change-Get-Less and Compare problems) but required

remaining children indicated that there was considerable im- addition as the solution strategy. Attempts to target specific

provement on the three problems that showed the Largest teaching toward such problems therefore seem justified. Over-

increases in Table 3: "Put-Together: missing second part," all, however, the posttest scores indicate that all 12 of these

"Change-Get-More: missing change," and "Compare: missing problem types are well within the "zone of proximal devel-

small." Thus the improvement in Class AA would best be opment" of American second graders of average and above-

considered as being limited to these three types of problems. average ability; that is, these children can profit from instruc-

No child in Class HA used the same operation on all problems tion on and practice with these problems. Because these

on any worksheet or test. problems all contained three-digit numbers that involved

The two problem types that showed the lowest posttest trading (borrowing and carrying), problems with smaller num-

performance for both classes were the conflict problems for bers might well be within the range of below-average second

the subtractive problem categories: the Change-Get-Less and graders or above-average first graders. The almost total omis-

sion of the more difficult subtypes of addition and subtraction

the Compare problems that require addition. All four prob-

word problems from United States textbooks thus does not

lems in the additive problem categories (Put-Together and

seem warranted.

Change-Get-More) showed fairly high posttest performance,

and they are conflict problems: They require subtraction for There are several ways in which the teaching unit presented

solution. Therefore, the subtractive conflict problems seem to here might be improved in an attempt to achieve more sizable

be particularly difficult to learn to solve. increases in performance: (a) spacing the learning of word

problem solutions throughout the school year for children to

Discussion benefit from distributed practice on mixed types of problems;

(b) relating the word problem instruction to instruction on

Second graders of average and above-average mathematics symbolic addition and subtraction so that children do not

ability can make the schematic drawings used in this study, learn only a single interpretation of addition or subtraction

can reliably distinguish among the different semantic word that later must be unlearned; (c) always giving a mix of

problem categories and make the correct drawing for a prob- problems that will require some addition and some subtrac-

lem from the category, can usually insert the numbers from tion so that children never work word problems mindlessly

the problem into the drawing correctly, and can select the by simply combining the numbers in all problems in the same

correct solution strategy for most problems. There was a way; (d) giving feedback to individual students about their

strong relation between a correctly filled-in drawing and a performance; and (e) targeting more teaching and practice

correct solution strategy; few correctly filled-in drawings led effort toward the most difficult semantic subtypes. The order

to an incorrect solution strategy, and few incorrectly filled-in of teaching of Put-Together, Change, and Compare that was

drawings led to a correct strategy. On about a fourth of the used does seem to have been appropriate. Put-Together prob-

lems are easy to represent and are also generally easy to solve,

posttest Put-Together problems with a missing part, the

and therefore serve as a good introduction, whereas Compare

above-average children made a Compare drawing; these draw-

stories and their representations are difficult and should be

ings led to correct solution strategies and may reflect some presented later.

children's view that these problems primarily involve a com-

parison between the small "part" and the big "all." The Our work suggested two improvements in the schematic

average children substituted Put-Together and Compare drawings. First, the drawings for the Put-Together and Com-

drawings for each other on about a fifth of each kind of pare problems should be inverted so that the large box is

problem, seemingly indicating a general confusion between placed above the small ones. This enables the child, in moving

these similar drawings. There was little evidence that children from the drawing to a vertical subtraction problem, to retain

successfully used Put-Together drawings for all problems types the two given numbers in the same spatial relationship. Be-

(cf. Resnick, 1983) or for the more difficult Change problems cause problems requiring addition have drawings with the

(cf. Riley et al., 1983, and Kintsch & Greeno, 1985). Such numbers to be added beside each other, this change makes

use was infrequent or was considerably less accurate in overall no difference for such problems. Second, the Compare draw-

200 GORDON B. WILLIS AND KAREN C. FUSON

PUT-TOGETHER ing might be altered so that the two smaller boxes correspond-

ing to the small and difference elements are physically sepa-

A rated from the box containing the big element, thus percep-

tually differentiating the Compare drawing from the similar

Put-Together drawing. Examples of the updated drawings for

use in later studies are illustrated in Figure 2.

The use of equations that semantically model a problem as

intermediate representations during problem solution seems

to be superfluous in conjunction with the schematic drawings.

Furthermore, because many young children consider the

equal sign in equations to mean "results in" or "gives" (Bar-

oody & Ginsburg, 1983; Behr, Erlwanger, & Nichols, 1980;

Kieran, 1980; Labinowicz, 1982), equations seem to be more

appropriate for young children to use for Change problems

than for the other types of problems. Thus equations may not

be an effective substitute for the drawings for Put-Together

and Compare problems. Interestingly, DeCorte and Verschaf-

CHANGE-GET-MORE fel (1985a) reported that children often have trouble writing

equations, especially for problems in which there is a conflict

C between the semantic structure of a problem and the solution

procedure. However, this problem may be acute only for the

Put-Together and Compare problems, insofar as Bebout

E (1986) showed that children can readily learn to write number

sentences that directly model Change problems with various

missing positions.

The results also suggest that the use of labels such as Put-

Together, Change, and Compare may be useful in helping

children distinguish among the different semantic types of

word problems and, ultimately, in solving the difficult conflict

problems. At the beginning of instruction, when asked "What

kind of problem is this?", children called problems simply

CHANGE-GET-LESS "addition" or "subtraction" problems. This classification per-

mits confusion between the semantic structure of a problem

C as additive or subtractive and the required arithmetic solution

procedure (addition or subtraction). The problem category

O E

labels and schematic drawings allow distinctions to be made

between these two different aspects and provide an organiza-

tion of the problem elements that may facilitate the decision

concerning the arithmetic solution procedure. Thus the draw-

ings may provide a more general and more flexible basis for

the learning of the relationships between concrete situations

and the numerical problems that describe and enable one to

solve those situations.

COMPARE References

Baroody, A. J., & Ginsburg, H. P. (1983). The effects of instruction

B on children's understanding of the "equals" sign. Elementary

School Journal, 84, 199-212.

Bebout, H. C. (1986). Teaching children to symbolically represent

addition and subtraction verbal problems with number sentences

that represent problem structure. Unpublished doctoral dissertation,

University of Wisconsin, Department of Curriculum and Instruc-

tion.

Behr, M., Erlwanger, S., & Nichols, E. (1980). How children view the

equals sign. Mathematics Teaching, 92, 13-15.

Briars, D. J., & Larkin, J. H. (1984). An integrated model of skills in

solving elementary word problems. Cognition and Instruction, I,

D 245-296.

Carpenter, T. P., & Moser, J. M. (1983). The acquisition of addition

Figure 2. Modified versions of schematic drawings. and subtraction concepts. In R. Lesh & M. Landau (Eds.), Acqui-

TEACHING SCHEMATIC DRAWINGS 201

sition of mathematical concepts and processes (pp. 7-44). New of Mathematics Education.

York: Academic Press. Kintsch, W., & Greeno, J. (1985). Understanding and solving word

Carpenter, T. P., & Moser, J. M. (1984). The acquisition of addition arithmetic problems. Psychological Review, 92, 109-129.

and subtraction concepts in grades one through three. Journal for Labinowicz, E. (1982, March). Children's intellectual autonomy and

Research in Mathematics Education, 15, 179-202. the construction of bi-directional equality. Paper presented at the

DeCorte, E., & Verschaffel, L. (1985a). Writing number sentences to Annual Meeting of the American Educational Research Associa-

represent addition and subtraction verbal problems. In S. K. Da- tion, New York City.

marin & M. Shelton (Eds.), Proceedings of the Seventh Annual McNemar, Q. (1954). Psychological statistics (2nd ed.). New York:

Meeting of the North American Chapter of the International Group Wiley.

for the Psychology of Mathematics Education (pp. 50-56). Colum- Resnick, L. B. (1983). A developmental theory of number under-

bus, OH: International Group for the Psychology of Mathematics standing. In H. P. Ginsburg (Ed.), The development of mathemat-

Education. ical thinking (pp. 109-151). New York: Academic Press,

DeCorte, E., & Verschaffel, L. (1985b). Working with simple word Riley, M. S., Greeno, J. G., & Heller, J. I. (1983). Development of

problems in early mathematics instruction. In L. Streefland (Ed.), children's problem-solving ability in arithmetic. In H. P. Ginsburg

Proceedings of the Ninth International Conference for the Psychol- (Ed.), The development of mathematical thinking (pp. 153-196).

ogy of Mathematics Education (pp. 304-309). Utrecht, The Neth- New York: Academic Press.

erlands: State University of Utrecht. Romberg, T. A., Harvey, J. G., Moser, J., & Montgomery, M. (1974).

Fuson, K. G (1986a). Roles of representation and verbalization in Developing mathematical processes. Chicago: Rand-McNally.

the teaching of multi-digit addition and subtraction. European Stigler, J. W., Fuson, K. C, Ham, M., & Kim, M. S. (1986). An

Journal of Psychology of Education, I, 35-56. analysis of addition and subtraction word problems in American

Fuson, K. C. (1986b). Teaching children to subtract by counting up. and Soviet elementary mathematics textbooks. Cognition and In-

Journal of Research in Mathematics Education, 17, 172-189. struction, 3, 153-171.

Fuson, K. C. (1988). Children's counting and concepts of number. Vergnaud, G. (1982). A classification of cognitive tasks and operations

New York: Springer-Verlag. of thought involved in addition and subtraction problems. In T. P.

Fuson, K. C, & Secada, W. G. (1986). Teaching children to add by Carpenter, J. M. Moser, & T. A. Romberg (Eds.), Addition and

counting on with finger patterns. Cognition and Instruction, 3, subtraction: A cognitive perspective (pp. 39-59). Hillsdale, NJ:

229-260. Erlbaum.

Fuson, K. C, & Willis, G. B. (in press). Subtracting by counting up: Willis, G. B., & Fuson, K. F. (1985). Teaching representational

More evidence. Journal for Research in Mathematics Education. schemes for the more difficult addition and subtraction verbal

Greer, B. (in press). Understanding of arithmetical operations as problems. In S. K. Damarin & M. Shelton (Eds.), Proceedings of

models of situations. In J. Sloboda & D, Rogers (Eds.), Cognitive the Seventh Annual Meeting of the North American Chapter of the

processes in mathematics. Oxford, England: Oxford University International Group for the Psychology of Mathematics Education

Press. (pp. 288-293). Columbus, OH: International Group for the Psy-

Kieran, C. (1980). The interpretation of the equals sign: Symbol for chology of Mathematics Education.

equivalence relations versus an operator symbol. In R. Karplus

(Ed.), Proceedings of the Fourth Annual Meeting of the Interna- Received February 3, 1986

tional Group for the Psychology of Mathematics Education (pp. Revision received November 25, 1987

163-169). Berkeley, CA: International Group for the Psychology Accepted December 13, 1987 •

- implementing new initiatives for mathematicsUploaded byapi-199094940
- Lesson Plan Fil-5Uploaded bykamiiya
- raise the bar on problem solvingUploaded byapi-21940065
- 4.ParticipationUploaded byphirlayayadil
- Towards a Problematique for Research on Mathematics TeachingUploaded byGreySi
- gr1 add subtract within 20 planning documentUploaded byapi-260343680
- math welcome newsletterUploaded byapi-328209963
- Employability Skills What Are TheyUploaded byMariam Oyindamola Campbell
- 1-s2.0-S187704281503102X-mainUploaded byAttar Raha
- Educational PsychologyUploaded byIoana Basarab
- thematicunit template-finalUploaded byapi-335047731
- M1L1-1Uploaded byeva
- 1Uploaded byHmanal Hdr
- year 6 t1 unit 3 mathematics termUploaded byapi-267136654
- Mathematics 1Uploaded byRebecca Brown
- cte anchor standards 2Uploaded byapi-298799918
- PPG Self Directed WorkforceUploaded byKrish
- An Investigation of the Pre-Services TeaUploaded byGrupo Sterling
- module three session one - ict to support teaching learning and assessment through numeracyUploaded byapi-299424782
- Note on Consumer Decision Making ProcessesUploaded byMauricio Samayoa
- assessmentcommentaryUploaded byapi-347025977
- Frogs_and_Flies_1i.pdfUploaded bySujairi Amhari
- BehaviouralCharacteristicsChecklistOfMathematicallyGiftedStudentsAppI&II 2Uploaded byZaheer Khan
- mathsUploaded byHaris Rao
- Math Individual LessonUploaded bychelseam122791
- Learning and Teaching Strategies » American ScientistUploaded byMike Likesbikes
- Rubrik, Panduan Pelajar & GuruUploaded byMohd Zaki
- castillero green sheet 2016Uploaded byapi-325331094
- Company Portfolio - RecruitingUploaded bydisman32
- freshman report cardUploaded byapi-289723424

- Algebra Learning: Insights From Cognitive PsychologyUploaded byapi-21244456
- Algebra Learning: Insights From Cognitive PsychologyUploaded byapi-21244456
- Structure in Elementary MathematicsUploaded byapi-21244456
- MLCA Post-Test Problem Data % Correct Thai1Uploaded byapi-21244456
- What Community College Developmental Mathematics Students UnderstandUploaded byapi-21244456
- Benny Goes to College: Long-Term Consequences OfUploaded byapi-21244456
- Habits of Mind: An Organizing Principle ForUploaded byapi-21244456
- Carnegie Learner Study Clinical InterviewUploaded byapi-21244456

- Topic4-BallscrewCalculationsUploaded byvairam44
- Hsslive XII Chem QN Bank 2 SolutionsUploaded byZaibunisa Mehaboob
- Ai²TS-4(XI)_SET - AUploaded byAmey Kale
- BN44-00201A.pdfUploaded byEgnaldo Pieretti
- 62388726 Module 6 MV Dry Type Power TransformerUploaded byoadipphone7031
- A Level Physics 13 June2015Uploaded byAnthony
- SLM04 Physics 1 - Timezone 2Uploaded byElz
- Units_ MUploaded byApurva Chouksey
- MIT6_013S09_chap02.pdfUploaded byMoeed Sheikh
- psexUploaded byMarco Alfanta
- ANSI C29.1-1998 (R2002) - Test methods for electrical power insulators.pdfUploaded byj_maldonado_86
- Chapter 5.2CUploaded bykoolro
- Micromotion F series Documentation(EN).PDFUploaded by劉彥君
- Chapter 1.HO GasesUploaded bySutiono Yono
- Exp 5.docUploaded bysju65
- MCQs General Unit IUploaded byN1234m
- DYNAMICS-LAB-MANUAL-ME2307.pdfUploaded byHarish Ram
- Xi Lecture SheetUploaded byaman
- Tugas Termodinamika 2 (Viodita Rizki Tek.kimia S1 C - 14)Uploaded byViodita Rizki
- Lubricant Properties CalculatorUploaded byAlberico Muratori
- Solution_Assignment_Laws of Thermodynamics_3rd Week (1)Uploaded byM.Saravana Kumar..M.E
- 3c3-AppsOf2ndOrders_StuUploaded byMauro Gomez Sandoval
- psphystestju02Uploaded byST
- CHM346PS5(S2008)keyUploaded byMC Badlon
- adiazUploaded bywilda
- Motion 1.docxUploaded byACrazyNakedMan
- Measurement and Modelling of Partial Discharge Behaviour in a Spherical Cavity Within a Solid Dielectric Material as a Function of Cavity DiameterUploaded byreply2amit1986
- rr222104-elctrical-and-electronics-engineeringUploaded bySRINIVASA RAO GANTA
- Theoretical 1 MonopoleUploaded byPangeranAndareasPanggabean
- Thermal Re Sis Trance and Colling SystemUploaded byaksaltaf9137