Vous êtes sur la page 1sur 21

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/245280700

An Analysis of Arithmetic Problem Posing by


Middle School Students
ARTICLE in JOURNAL FOR RESEARCH IN MATHEMATICS EDUCATION NOVEMBER 1996
Impact Factor: 1.27 DOI: 10.2307/749846

CITATIONS

READS

140

175

2 AUTHORS:
Edward A. Silver

Jinfa Cai

University of Michigan

University of Delaware

95 PUBLICATIONS 2,001 CITATIONS

115 PUBLICATIONS 1,241 CITATIONS

SEE PROFILE

SEE PROFILE

Available from: Edward A. Silver


Retrieved on: 28 February 2016

An Analysis of Arithmetic Problem Posing by Middle School Students


Author(s): Edward A. Silver and Jinfa Cai
Reviewed work(s):
Source: Journal for Research in Mathematics Education, Vol. 27, No. 5 (Nov., 1996), pp. 521539
Published by: National Council of Teachers of Mathematics
Stable URL: http://www.jstor.org/stable/749846 .
Accessed: 01/03/2013 15:19
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

National Council of Teachers of Mathematics is collaborating with JSTOR to digitize, preserve and extend
access to Journal for Research in Mathematics Education.

http://www.jstor.org

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

Journalfor Researchin MathematicsEducation


1996, Vol. 27, No. 5, 521-539

AN ANALYSISOF ARITHMETICPROBLEM
POSINGBY MIDDLE SCHOOLSTUDENTS
EDWARD A. SILVER, Universityof Pittsburgh
JINFA CAI, Universityof Delaware
The mathematical
problemsgeneratedby 509 middleschoolstudents,who weregiven a briefwritten "story-problem"
descriptionandaskedto pose questionsthatcouldbe answeredusingtheinformation,were examinedfor solvability,linguisticandmathematicalcomplexity,andrelationships
within the sets of posed problems.It was found thatstudentsgenerateda largenumberof solvable mathematicalproblems,many of which were syntacticallyand semanticallycomplex, and
thatnearlyhalf the studentsgeneratedsets of relatedproblems.Subjectsalso solved eight fairly
andtheirprobperformance
complexproblems,andtherelationshipbetweentheirproblem-solving
lem posing was examinedto reveal that"good"problemsolvers generatedmore mathematical
problemsandmorecomplex problemsthan"poor"problemsolvers did. The multiple-stepdata
analysisschemedevelopedandusedhereinshouldbe usefulto teachersandotherresearchersinterested in evaluatingstudents'posing of arithmeticstoryproblems.

Recentrecommendations
for the reformof school mathematicssuggestan imporandEvaluation
tantroleforstudent-generated
problemposing.Forexample,theCurriculum
Standardsfor School Mathematics(NCTM, 1989) explicitly states that students
should"havesome experiencerecognizingandformulatingtheirown problems,an
theProfessional
(p. 138).Furthermore,
activitythatis attheheartof doingmathematics"
Standardsfor TeachingMathematics(NCTM, 1991) suggests the importanceof
teachers'providingopportunitiesfor studentsto pose theirown problems:"Students
shouldbe given opportunities
to formulateproblemsfromgiven situationsandcreate
new problemsby modifyingthe conditionsof a given problem"(p. 95). Thesedocumentsreflectan apparent
to makeprobhighlevel of interestamongmanypractitioners
lem posinga moreprominentfeatureof classroominstruction.Evidenceof thisinterest canalsobe inferredfromtherecentpublication
of a collectionof practitioner-oriented
articlesrelatedto problemposing(Brown& Walter,1993)andtheappearance
of numerous articlesin popularjournalswhose audienceis primarilyelementaryschoolteachers(e.g.,Silverman,Winograd,& Strohauer,1992;Maddon,1994).Infact,atthistime,
it appearsthatpractitionerinterestis runningfar aheadof the developmentof credible techniquesfor assessingmathematicalproblemposing andthe accumulationof
solid researchevidence regardingits nature.

of thisreportwas supportedin partby NationalScienceFoundationgrantMDRPreparation


8850580 and by a grant from the Ford Foundation for the QUASAR (Quantitative
Understanding:Amplifying Student Achievement and Reasoning) project. The opinions
expressedarethoseof the authorsanddo not necessarilyreflectthe views of eitherFoundation.
An earlierversionof this paperwas presentedat the 1993 annualmeetingof the American
EducationalResearchAssociation,Atlanta,GA. The authorsare gratefulto the editorandto
severalanonymousreviewerswho madevaluablecommentsconcerningan earlierversionof
this article,therebycontributingto its improvement.

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

522

ProblemPosing by MiddleSchool Students

Over the past severaldecades,considerableprogresshas been madein studying


many importantaspects of mathematicalproblem solving, including detailed
andstudiesrelatedto thelearningandteachanalysesof problem-solving
performance
ing of problemsolving (Charles& Silver, 1988; Silver, 1985; Schoenfeld, 1985).
Researchhas describedmany of the cognitive processes associatedwith the solution of mathematicalproblems that are posed by a source outside the solver.
Althoughcurrentinterestin mathematical
problemposingcanbe seen as representing
a new facet of a longstandinginterestin mathematicalproblemsolving (Stanic &
Kilpatrick,1988), far less is known aboutthe cognitive processes involved when
solvers generate their own problems (Kilpatrick,1987), or about instructional
strategiesthatcan effectively promoteproductiveproblemposing, althoughmore
progresshas been made on the latterfrontthanon the former.
Therehave been severalreportsof instructionalapproachesused to incorporate
problemposing into the mathematicsinstructionof studentsat a wide rangeof educationallevels in the U.S. (e. g., Healy, 1993; Keil, 1964/1965;Perez, 1985/1986;
Winograd,1990) and abroad(e.g., Hashimoto,1987; van den Brink, 1987). Some
reportshave also includedan examinationof the impacton studentsof experience
orformalinstruction
problemposing(e.g.,Keil, 1964/1965;
emphasizingmathematical
In
these
studieshave foundthathavingstuScott,
Perez, 1985/1986;
1977). general,
dentsengage in some kind of generativeactivityrelatedto problemposing--often
somethingas simpleas rewritinggiven storyproblems-has a positiveinfluenceon
achievement(Hashimoto,1987;Keil, 1964/1965;Perez,
theirword-problem-solving
or
their
attitude
towardmathematics
1985/1986;Scott,1977)
(Perez,1985/1986;
Winograd,
in these studiessuggeststhateven very
the
evidence
accumulated
1990/1991).Thus,
simpleexperienceswithmathematicalproblemposingcanhave a positiveimpacton
interventions
students.Nevertheless,despitethefactthatsomeaccountsof instructional
effects
of theintershown
the
positive
problemposinghave
emphasizingmathematical
not
ventionson studentachievementandattitude,these studieshave
directlyexamined mathematicalproblemposingitself. This priorresearchhas thereforeprovided
abouteithertheprocessesusedby studentsin problemgenrelativelylittleinformation
erationor the productsof students'problem-posingactivity.
A few researchershave examinedthe mathematicsproblemsposed by children
(e.g., Ellerton,1986; Silvermanet al., 1992), by prospectiveelementaryschool or
secondaryschool teachers(e.g., Leung, 1993; Silver, Mamona-Downs,Leung, &
Kenney, 1996), or by in-service middle school teachers(e.g., Silver et al., 1996).
Thusfar,researchon children'sproblemposing has tendedto focus on smallnumbers of subjects and to provide only a fairly superficial analysis of the posed
problems,if any analysis at all. For example,Ellerton(1986) comparedthe mathematicalproblemsgeneratedby eight high-abilityyoung childrenwith those generatedby eight low-abilityyoung childrenby asking each to pose a mathematical
problemthatwouldbe difficultfor a friendto solve. Ellertonreportedthatthe more
able studentsposed problemsthatwere morecomplex thanthose posed by the less
able students,but her criteriafor determiningproblemcomplexity were not well
specified. In anotherinvestigation,Silvermanet al. (1992) reportedthata class of

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

EdwardA. Silver and Jinfa Cai

523

fifth-gradestudentswas able to generatestoryproblemsthatexceededin difficulty,


novelty, and interestthe wordproblemexercises foundin theirtextbooks,but their
criteriafor judging these qualitieswere likewise underspecified.If progressis to
be made in understandingthe natureof mathematicalproblemposing, or if rigorous attemptsareto be madeto studythe instructionalimpactof interventionsrelated
to mathematicalproblemposing,thenbetteranalytictechniquesmustbe developed
to study problemposing by elementaryschool and middle school students.
Some guidancefor the developmentof schemesto analyzechildren'smathematical
problemposing is providedby the approachused in those studiesof adultproblem
posingthathaveincludedmoreextensiveandrigorousanalyses.Forexample,Leung
(1993) successfullyuseda varietyof cognitiveanalysistools,suchas GeneralProblem
Solver (GPS) graphs (Newell & Simon, 1972) and arithmetic story problem
schemaanalysis(Marshall,1995),to examinetheproblem-posing
productsandprocesses
of about50 prospectiveelementaryschool teacherswho posed writtenarithmetic
story problemsin responseto writtenprompts.In anotherinvestigationinvolving
adultsubjects,Silver et al. (1996) developeda differentkind of scheme to analyze
the written-problem-posingproducts of about 80 preservice secondary school
teachersand in-servicemiddle school teacherswho posed mathematicalproblems
relatedto a complex taskenvironmentinvolvingthe hypothesizedpathof a billiard
ballon tablesof varioussizes andshapes.Theiranalyticschemeattendedto the nature
of posedproblemsin relationto the informationgiven in the taskenvironment.They
also examinedthe relationshipbetween subjects' problemposing and their solution of a specifiedmathematicalproblemin the sametaskenvironment.Aspects of
the analyticapproachesusedby Leung(1993) andby Silveret al. (1996) wereincorporatedinto the currentstudy.
A majorgoal of the studyreportedherewas to developanduse an analyticscheme
to examinethe problemposing of middleschool students.In particular,the scheme
developed and used in this study employed semanticcategoryanalysis and other
analytic tools borrowed and adapted from research on mathematicalproblem
solving. This analyticscheme providedthe basis for an examinationof the nature
and complexity of the arithmeticstory problems posed by middle school students.Anothergoal of this studywas to examinethe relationshipbetweenstudents'
problemposingandtheirproblemsolving.In this study,this goal was accomplished
by probingthe differencesbetween the problemposing of studentswho were successful problemsolvers and thatof studentswho were less successful.
Silver (1994) has noted that the term "problemposing" is generally appliedto
threequite distinctforms of mathematicalcognitive activity:(a) presolutionposing, in which one generatesoriginalproblemsfrom a presentedstimulussituation;
(b) within-solutionposing,in whichone reformulatesa problemas it is being solved;
and(c) postsolutionposing,in whichone modifiesthegoalsorconditionsof analready
solved problemto generatenew problems.It is a form of presolutionposing that
is examinedin this investigation.The decision to focus on arithmeticstory problems was based on the availability of an extensive researchbase on which to
of this type of activityfor middleschool students,and
build, on the appropriateness

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

ProblemPosing by MiddleSchool Students

524

on the fact thatpriorresearchhas shown the efficacy for elementaryand community college students of problem-posing experiences related to the writing or
rewritingof arithmeticstoryproblems.A greaterunderstandingof this particular
type of mathematicalproblemposingcanhavebotha practicalanda theoreticalpayoff. In particular,a scheme to analyzethe complexityof arithmeticstoryproblems
generatedby studentscouldbe useful bothto teachers,who mightwish to use such
orto measurestudentprogress,
a schemeto evaluatetheeffectivenessof theirinstruction
andto researchers,who mightuse it and/orthe resultsobtainedfromits use to help
them understandat least one form of a cognitive activity called problemposing.
METHOD
Subjects

middleschoolstudentsattendingschools
Subjectswere509 sixth-andseventh-grade
in four differentlow-income communitiesin urbanlocations in the United States.
The studentsattendedfour middle schools thatwere partof the QUASAR project
during the 1990-91 school year. QUASAR was intended to foster innovative
commathematicsinstructionin middleschoolsservingeconomicallydisadvantaged
in
the
munities(Silver& Stein,1996).Exceptfortheirinterestin participating QUASAR
project,the schools were typical of urbanmiddle schools in the U.S. The students
in the schools were also typical:An ethnically and linguisticallydiverse population (about50% of the studentswere AfricanAmerican,about20% were White,
about20%Latino,andabout10%AsianAmerican)who performedgenerallybelow
average on standardizedachievement tests. The sample was divided approximately equally between boys and girls.
Tasks and Administration

Each subjectcompleteda problem-posingtask andeight problem-solvingtasks


in a single class periodof approximately45 minutes.The taskswere administered
by the students'teachersduringmathematicsclass, as partof the biannual(fall and
spring) project testing at QUASAR schools. The eight open-ended problems,
together with the posing task, comprised one of four forms in the QUASAR
(QCAI)for grades6 and7 (Lane,1993).TheQCAI
CognitiveAssessmentInstrument
is an assessmentinstrumentdeveloped by the projectto measurestudents'mathematicalthinking,reasoning,andunderstanding
(Silver& Lane, 1993). QCAItasks
haveundergoneextensivescrutinyto ensuretheirqualityandfairness(Lane& Silver,
1995). The problem-posingtask andthe problem-solvingtasks in this form of the
QCAIhadbeen pilot tested severaltimes to ensurethatthe tasks assessed the cognitiveprocessesandcontentareastheyweredesignedto assess (Magone,Cai,Silver,
& Wang, 1994).
The eight problem-solvingtaskswere constructed-response
tasks,involvingvarnumbertheory,
ious mathematicscontentareas:fractions,geometry/measurement,
andstatistics.Fortheseproblem-solving
patternsandrelationships,ratio/proportion,

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

EdwardA. Silver and Jinfa Cai

525

tasks, studentswere requirednot only to produceanswersbut also to justify their


solutionsor to explaintheirsolutionprocesses.The problem-posingtask,which is
shown in Figure 1, asked studentsto pose threequestionsthatcould be answered
on the basis of some given information.
To minimizethe effectof taskorder,QCAItasksweresystematicallyvariedacross
problembooklets. In particular,therewere threedistinctarrangementsof the nine
tasks within the test booklet, so that aboutone thirdof the sample completedthe
problem-posingtaskas the secondtaskin the booklet,one thirdcompletedthe problem-posing task as the fifth task in the booklet;and the otherone thirdcompleted
the problem-posing task as the eighth task in the booklet. About half of the
responsesconsideredin this studywere obtainedin fall 1990 andthe otherhalf in
spring 1991.
Writethreedifferentquestionsthatcan be answeredfromthe information
below.
Jerome, Elliot,and Arturotook turnsdrivinghome froma trip.Arturodrove80
milesmorethan Elliot.Elliotdrovetwice as manymilesas Jerome.Jerome
drove50 miles.
Question#1

Question#2

Question#3

Note: Inthe task booklet,studentswere givenmorespace inwhichto writetheir


responses.
task.
Figure1. Problem-posing

Data Coding

A summaryof thecodingschemedevelopedandusedin thisinvestigation


is provided
in Figure2. Eachstepin the codingprocessis explainedmorefully in this section.
Students' problem-posing responses were first categorized as mathematical
questions,nonmathematicalquestions,or statements.Those responsesgiven in the

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

ProblemPosing by MiddleSchool Students

526

form of mathematicalquestions,when takentogetherwith the informationgiven


in the task core, can be consideredto constitutea mathematicalproblem.Thus, it
was possibleto considerthe student-generated
questionsto be problemsandto analyze themas such.The next step involvedcategorizingthe mathematicalproblems
as solvableor not solvable.Problemswereconsideredto be notsolvableif theylacked
sufficientinformationor if they posed a goal thatwas incompatiblewith the given
information.Forexample,the response,"DidArturodrivefasterthanJerome?"was
considered to representa problem that was not solvable because information
regardingrelativedrivingspeedsor times was neithergiven in the tasknorsupplied
by the student.An exampleof an "impossible"problem-one in which the goal is
withtheconditions-is theresponse,"HowmanymilesmoredidJerome
incompatible
drive thanElliot?"

Responses

Nonmath
questions

Mathquestions

Statements

Solvable

Nonsolvable

Semantic
analysis

Linguistic
syntactic
analysis

Figure 2. Summaryof multiple-stepdatacoding scheme.

The laststepin the codingprocessinvolvedexaminingthe complexityof theposed


problems.One type of complexity was relatedto the linguistic or syntacticstructuresembeddedin the posed problems.In some priorresearch(e.g., Mayer,Lewis,
& Hegarty, 1992) the linguistic structureof mathematicsstoryproblemshas been
examinedby focusing on the presence of assignment,relational,and conditional
propositionsin problemstatements.An assignmentpropositionis a questionsuch

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

EdwardA. Silver and Jinfa Cai

527

as "Howmanymiles did they drivein all?"A relationalpropositionis a statement


suchas "Howmanymoremiles didArturodrivethanJerome?"A conditionalproposition is a question such as "If Arturodrove 80 miles more thanElliot, how many
miles did Arturodrive?"Mayeret al. (1992) foundthatproblem-solvingdifficulty
appearedto be relatedto linguistic complexity, in thatproblemswith conditional
andrelationalpropositionstendedto be moredifficultfor studentsto solve thanthose
containingonly assignmentpropositions.Thus,the presenceof conditionalor relational propositions can be taken as an indication of problem complexity. It is
importantto note thatthe task core (i.e., the informationpresentedto the students
fromwhichthey wereto generateproblems)containedtwo assignmentpropositions
("Jerome,Elliot,andArturotook turnsdrivinghomefroma trip"and"Jeromedrove
50 miles")andtwo relationalpropositions("Arturodrove80 miles morethanElliot"
and"Elliotdrovetwice as manymiles as Jerome").The taskcorethereforewas itself
quitecomplex froma linguisticperspective.Ouranalysisfocused on the additional
complexity contributedby the questionsposed by the students.Because this type
of complexity analysis was feasible for nonsolvable mathematicsproblems as
well as for those thatwere solvable,both types of responseswere consideredin the
analysis reportedhere.
Anothertype of complexity relatedto the mathematicalstructuresfound in the
posed problems.Because the posed problemscould be solved using some combinationof arithmeticoperations,one plausiblemeasureof mathematicalcomplexity wouldbe thenumberof operations,orthe numberof computational
steps,required
for solution.Leung(1993) successfullyanalyzedthe complexityof arithmeticproblems posed by preserviceelementaryschool teachersby using GPS graphsto deterusedin solvingtheposedproblems.Althoughthisapproach
minethenumberof operators
was shown by Leung to be very useful and powerful,it is not the only reasonable
way to measure mathematicalcomplexity. Counting the numberof steps in a
hypothesizedsolutionhas the advantageof assessingcomplexityin a straightforward
and reasonablemanner,but it also has the disadvantageof makingsome relatively
simplearithmeticstoryproblemsappearto be fairlycomplex.Forexample,if a problem's solutioninvolved the additionof five numbers,then it would be countedas
requiringfouroperationsteps for its solution.If one determinesproblemcomplexity by countingoperators,then this problemwould be seen as more complex than
anotherproblemthatrequireda multiplicationfollowed by a subtraction,because
this latterproblemrequiredonly two operationstepsfor solution.On the otherhand,
if one determinesproblemcomplexityby enumeratingdistinctsemanticrelations,
thenthelatterproblemwouldbe seento be morecomplexthanthefirstproblem,because
the latterproblemembodiestwo distinctsemanticrelationsandthe first embodies
only one. It is this second approachthatwas takenin the studyreportedhere;that
is, the numberof semanticrelationsratherthanthe numberof operatorswas used
to determinethe mathematicalcomplexityof the posed problems.
All mathematicallysolvable problems were subjected to semantic category
analysis using a classificationscheme of arithmeticword problemsdevelopedby
Marshall(1995). The posed arithmeticwordproblemswere classified on the basis

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

528

ProblemPosing by MiddleSchool Students

of theirunderlyingsemanticstructuralrelationsusing Marshall'sfive categories:


Change,Group,Compare,Restate,Vary. A mathematicalproblemthatwas classified as involvingN semanticstructuralrelationsin the classificationscheme was
designatedan N-relationproblem.If a mathematicalproblemcould be answered
directlyfromthe given information,it was designateda zero-relationproblem,and
in this analysis it would be said to involve zero semantic relations. Problems
involving a greaternumberof semanticrelationsareconsideredto be semantically
more complex thanthose involving fewer relations.
To examine interraterreliability, one person classified all students' problemposingresponses,afterwhicha secondpersonrandomlyselected50 students'responses
and independentlyclassified them. Interrateragreementfor the basic classification (mathematicalquestion, nonmathematicalquestion,or statement)was 93%.
Ratesof agreementon the classificationsof linguisticandmathematicalcomplexity
of students'posed mathematicalproblemswere also highly acceptable,93% for
linguistic complexity and 89%for mathematicalcomplexity. Because there was
substantial agreement between raters, the first person's classifications of all
students' problem-posingresponses were used in the subsequentanalyses.
The problem-solvingresponseswere evaluatedusing a focused, holistic scoring
method(Silver & Lane, 1993). A generalizedscoringrubricwith threeinterrelated
components (mathematical, conceptual, and procedural knowledge; strategic
knowledge;andcommunication)specifiedcriteriafor eachof five scorelevels (0-4)
and guided the developmentof a specific rubricfor each task (Lane, 1993). Each
of the students'responseswas scoredindependentlyby two middle school teachers, who were trainedto use the scoring rubric.Interrateragreementsfor each of
the eight tasks rangedfrom 75% to 89%,which was judged to be acceptable.
RESULTS
The resultsare presentedin two sections. The first section providesa summary
of students'problem-posingresponses,includingthe analyses of complexity and
relatedness;andthe secondpresentsan analysisof therelationshipbetweenstudents'
problemposing andtheirproblemsolving. In the analysesreportedhere, the sample is treatedas a whole ratherthanexaminedby gradelevel (6 or 7) or by testing
occasion (fall or spring).Data for this studywere collected duringthe first year of
QUASARprojectactivityat eachof the sampleschools.Duringthatyear,substantial
attentionwas devotedto thedesignof innovativeinstructional
programsandto enhancing teachers'knowledgeof contentandpedagogy;less changewas actuallyimplementedin classroominstructionon a day-to-daybasis. Thus,for the datafrom this
year,thereappearedto be no compellingreasonto separatethe sampleby response
occasion or by grade.
Problem-PosingResponses
Subjectsprovideda totalof 1465responses.Morethan70%of theresponseswere
classified as mathematicalquestions,about20% were statements,and 10%were

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

EdwardA. Silver and Jinfa Cai

529

nonmathematicalquestions. Approximately the same distributionof response


types was evident for each of the threeresponsescalled for in the task. Although
manycombinationsof responsetypes were theoreticallypossible, the dataindicate
that students tended to be consistent with respect to response types, because
approximately75% of the studentsgeneratedthreemathematicalquestions,three
nonmathematicalquestions,or three statements.
Nearly 80% of the studentsgeneratedat least one mathematicalquestion.More
thanhalf (about57%)of the studentsgeneratedthreemathematical
questions.In fact,
the studentswho generatedthreemathematicalquestionsaccountedfor over 80%
of the mathematicalquestionsgeneratedby all students.In otherwords,the remaining 40% of the studentsgeneratedless than20% of the mathematicalquestions.
The mathematicalquestionsposed by the studentswere of particularinterest,and
they were subjectedto furtheranalysesof mathematicalsolvability andlinguistic
and mathematicalcomplexity. The resultsof these analyses are discussed next.
MathematicalSolvability
Morethan90%of the mathematicalproblemsgenerated(i.e., the questionsposed
in thetaskcore)werejudgedto be mathematically
by studentsandthegiveninformation
solvable.Althoughthe solutionof some solvableproblemsmighthaverequiredinformationbeyond thatgiven in the task core and the posed question,the majorityof
the solvable problemscould be answeredon the basis of informationgiven in the
task core. Twelve studentseach generatedone mathematicallysolvable problem
thatcould be answeredon the basis of the given informationandnew information
suppliedby the studentin the posed question.An example of this kind of hypothesis-based mathematicalquestionis the following: "Howmany times would they
have to get gas if they got 160 miles each fill-up?"
LinguisticComplexity
The linguisticor syntacticcomplexityof the posed problemswas determinedby
examiningall posed mathematicalquestionsfor the presenceof assignment,relational, and conditionalpropositions.As mentionedearlier,the presenceof conditional or relationalpropositionsin the posed questionis takento be an indication
of problemcomplexity. In the responsesobtainedin this study,nearly60% of the
mathematicalquestionsinvolvedonly assignmentpropositions,about35%involved
relationalpropositions,andonly 5%involved conditionalpropositions.Almost all
students(80%)generatedat least one mathematicalquestioninvolving an assignment proposition.Althoughrelationalpropositionswere found in only aboutone
thirdof the responses, aboutone half of the studentsgeneratedat least one mathematicalquestioninvolving a relationalproposition.About 10%of studentsposed
at least one mathematicalquestionwith a conditionalproposition.
MathematicalComplexity
All mathematicallysolvableproblemswere examinedfor the presenceof the five
fundamental
semanticstructural
relations--Change,
Group,Compare,Restate,Vary--

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

530

Problem Posing by Middle School Students

or combinationsof these relations.Using this approach,over 90% of the solvable


mathematicalproblemscould be classifiedwith respectto semanticcomplexity,or
the numberof relationsrequiredfor solution.The numberof relationsin the posed
problemsrangedfrom 0 to 5. Examplesareprovidedin Table 1.
Table 1
Examplesof MathematicalProblemsand the CorrespondingNumberof SemanticRelations
Numberof relations
Zero
One
Two
Three
Four

Five

Examples
Did Arturodrive 80 miles more thanElliot?
[None]
How many miles did Elliot drive?
[Restate]
How many more miles did Elliot drive thanJerome?
[Compare/restate]
How many miles did the threeboys drive altogether?
[Group/restate/restate]
How many times would they have to get gas if they got 60
miles each fill up?
[Vary/group/restate/restate]
Did Arturodrive a longer time thanJeromeand Elliot drove
altogetherin the regularway?
[Compare/restate/group/restate/vary]

Most of the problemsposed were semanticallycomplex. In fact, about60% of


the solvable mathematicalproblemsinvolved two or morerelations,andthese are
mathematical
hereafterreferredto as multirelation
problems.Slightlymorethan20%
16%involved zero relations.The
and
about
one
of the problemsinvolved
relation,
generationof semanticallycomplexproblemswas fairlywell distributedacrossthe
sample.In fact,nearly70%of the studentsgeneratedat leastone multirelationmathematicalproblem,and a little less thanhalf of the studentsgeneratedat least two
multirelationproblems.Therewas a tendencyfor multirelationproblemsto appear
laterratherthanearlierin the responsesequence. Only 34% of the first responses
were multirelationproblems,whereas41%of the secondresponsesand49%of the
thirdresponseswere multirelationproblems.
The laterresponsesof studentstendedto be somewhatmore complex semantically thanwere the earlierresponses.Table2 shows the percentagesof studentswho
shifted or did not shift the complexity levels of theirposed mathematicallysolvableproblemsfromthefirstto thesecondresponsesandthefirstto thethirdresponses.
In particular,abouthalf of those studentsgave a morecomplexproblemas theirsecond responsethantheirfirst;whereas,the responsesof aboutone thirdof the studentsmoved in the oppositedirection.A matched-pairWilcoxontest indicatedthat
studentshad significantlymorecomplex mathematicallysolvable problemsin the
secondresponsesthanin theirfirstresponses(z = 2.71,p < .01). Similarresultswere
obtainedfor shifts from the first to the thirdresponses(z = 5.77, p < .001).

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

EdwardA. Silver and Jinfa Cai

531

Table 2
Percent and Direction of Changes in Complexityof MathematicalProblemsAcross Responses
Directionof change in complexity
Less to more complex
Same
More to less complex

Responses
1st to 2nd
48%
20%
32%

1st to 3rd
55%
20%
25%

RelationshipsAmongPosed Responses
This analysisfocused on examiningpossible relationshipsamongthe responses
thatmightilluminatestrategicaspectsof the thinkingstudentsmay have done while
posing theirproblems.In particular,studentsmightpose theirsecondor thirdproblems in close associationwith theirfirst or second posed problems.Adaptingthe
analysisapproachdevelopedby Silveret al. (1996), we examinedtwo differenttypes
of relationshipsamongresponses:symmetricresponsesandchainedresponses.Sets
of symmetricresponseswere those thatreflectedrelationshipsbetween or among
given or imputedobjectsin the problemspace. Anotherkind of relatednessis evidentif the second or thirdposed responserequiresuse of informationderivedfrom
the solutionof an earlierposedproblem.Sets of questionshavingthis characterwere
consideredto be chainedresponses.Figure3 containsexamplesof symmetricand
chainedresponses.
Twenty-sevenpercentof the studentsgeneratedsymmetricresponses.Thesewere
usually in sets of threeresponses,but sometimestherewere only two responsesin
a symmetricset. For the studentswho gave symmetricresponses,their second or
thirdresponses appearedto be generatedby changingsome objects or conditions
froma priorposedresponse.Forexample,in thefirstexampleof thesymmetric
responses
in Figure3, the secondandthe thirdquestionswereposedjust by changingthename
"Jerome"in the first questionto "Elliot"and "Arturo,"respectively.Similarly,in
the second example of the symmetricresponsesin Figure3, the second and third
questionswereposed simplyby changingtherelationalterm"most"in the firstquestion to the relationalterms "least"and "middle,"respectively.Thirty-sixpercent
of the studentsgeneratedchainedresponses.Almost half of the chainedresponses
were found in sets of three;thatis, all the responsesgiven by some studentswere
relatedin this way.
About 45% of the students provided responses that were either symmetric
responses, chained responses, or both. Of those who provided such responses,
some studentsgave responsesthatwerebothsymmetricandchained,andsomeother
studentsgave only one type. Examiningthe firstresponsesof those who had symmetricand/orchainedresponsesrevealedthatabout40% of the studentsprovided
first responses involving Elliot, as in, "How many miles did Elliot drive?" or
"Elliotdrove 100 miles." The mileage thatElliot drove appearedto act as a primiformanystudentsin thehypothesized
tive,or"firstunknown,"
drivingsituationdescribed
in the given situation.

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

ProblemPosing by MiddleSchool Students

532

Symmetric Responses
HowmanymilesdidJeromedrive?
How manymilesdid Elliotdrive?
How manymilesdidArturodrive?
OR
Whodrovethe most miles?
Whodrovethe least?
Whodrovein the middle?

Chained Responses
Howmanymilesdid Elliotdrive?
HowmanymilesdidArturodrive?
Howmanymilesdidthe threeboys drivealtogether?
OR
Howmanymilesdid Elliotdrive?
Howmanymilesdid Elliotand Jeromedrivealtogether?
HowmanymoremilesdidArturodrivethanElliotand
Jeromedrovealtogether?
Figure 3. Examplesof responsesets classified as symmetricor chained.

Responses OtherThanMathematicalQuestions
About 10%of the students'responseswere classified as nonmathematicalquestions, of which aboutone of every five was the sortthatmight be asked in a reading comprehensionexercisebasedon the passage(e.g., "Whatarethe namesof the
threeboys on thetrip?""Whattripdidtheygo for?""Aretheytiredof the driving?").
Aboutone thirdof the nonmathematicalquestionsinvolvedquestioningthe underlying rationalefor the given informationor for the mathematicalrelationships(e.g.,
"Why[did] Arturodrive 80 miles morethanElliot?""Whydid Arturodrivemore
miles than Jerome?").Other nonmathematcalquestions involved other issues,
such as students'complaintsaboutbeing asked to pose problems(e.g., "Whyare
you asking me to do this?""My teacherdid not teach us how to do this.").
Nearly20%of the students'responseswere classifiedas statements.The majority of theseinvolvedrestatementsof the given information,suchas "Arturodrove80
or inferencesbasedon the given informamiles morethanElliot"or interpretations
tion,suchas "Elliotdrove100 miles."In some cases the interpretivestatementswere
incorrectandappearedto involvemistakinga relationalpropositionfor anassignment

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

EdwardA. Silver and Jinfa Cai

533

proposition,which has been documentedas a common difficulty in the solving of


complex storyproblems(Cocking& Mestre, 1988). The two most frequentincorwere "Arturodrove80 miles"and"Elliotdrove70 miles."Some
rectinterpretations
of the given informationon the
otherstatementsappearedto involve interpretations
basis of the priorexperienceof the studentsratherthanon the informationexplicitly providedto them (e.g., "Jeromemustbe the youngest one since he only drove
50 miles.")
RelatingProblemPosing and ProblemSolving
In order to examine the relationshipbetween students' problem posing and
theirproblemsolving, two extremegroups(Hi and Lo) were formedon the basis
of problem-solvingperformance.Thenthe similaritiesanddifferencesbetweenthe
problemposing of the groupswere examined.
The Hi problem-solvinggroupwas comprisedof the 50 studentswith the highest mean score on the eight problem-solvingtasks;the Lo groupconsistedof the
50 studentswith the lowest mean scores. The two groupshad substantiallydifferent levels of problem-solvingsuccess-the Hi grouphada significantlyhighermean
problem-solvingscorethanthe Lo group(MeanHi= 3.14, MeanLo= 0.26; t = 50.20,
p < .001)-although the Hi groupitself had only a moderatedegree of success in
solving the problems.
The averagenumberof responsesgeneratedby the two groupswas identical,but
therewere intergroupdifferencesin the qualityof the responses.The Hi groupgenerateda significantlygreaterproportionof mathematicalquestionsthanthe Lo group
(Hi = .92, Lo = .47; Wilcoxon-Mann-Whitneytest, z = 5.67, p < .001). In contrast,
Lo groupstudentsgenerateda significantlygreaterproportionof statementsthan
Hi groupstudents(Hi = .06; Lo = .43; Wilcoxon-Mann-Whitneytest, z = 4.58, p
< .001). But a closer examination of the statements generatedby each group
revealed qualitativedifferences.In particular,78% (7 out of 9) of the statements
generatedby the Hi groupstudentswere correctinterpretationsof the given information(e.g., "Elliotdrove100 miles"),whereasonly 26%(16 out of 62) of the statements generatedby the Lo group studentswere correctand almost half the statements generatedby the Lo group studentsinvolved an incorrectinterpretationof
the given information(e.g., Arturodrove 80 miles). The Lo groupalso generated
a higherpercentageof nonmathematical
questions(10%)thandidthe Hi group(1%),
but the differencewas not statisticallysignificant.
Whenresponseswere examinedat the level of individualstudentsratherthancollectivelywithingroups,patternsof differencewerenotedsimilarto thosefoundbetween
the groupsas a whole. In particular,a significantlygreaterproportionof Hi group
studentswrote at least one mathematicalquestionthandid Lo groupstudents(Hi
= .96, Lo = .56; z = 4.68, p <.001). Moreover,a significantlygreaterproportionof
Hi groupstudentsgeneratedthreemathematicalquestionsthandid Lo groupstudents (Hi = .82, Lo = .36; z = 4.68, p < .001).

Differences were also detected in the complexity of the problems posed. On


average, Hi group studentsposed significantlymore multirelationmathematical

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

ProblemPosing by MiddleSchool Students

534

problemsthanLo groupstudents(MeanHi= 1.58,MeanLo= .68; t= 5.14,p < .001),


and a significantlygreaterproportionof the problemsgeneratedby Hi groupstudents were multirelationproblems(Hi = .61, Lo = .46; Wilcoxon-Mann-Whitney
test,z = 4.68, p < .001). Also, a significantlygreaterproportionof Hi groupstudents
generatedat least one multirelationmathematicalproblemthandid Lo groupstudents (Hi = .88, Lo = .42; z = 9.64, p < .001). Thus, studentsin the Hi groupgeneratednot only more mathematicalproblemsbut also more complex mathematical problemsthandid the studentsin the Lo group.
DISCUSSION
The middleschool studentsin this studywere able to generatea largenumberof
appropriatemathematicalquestionswhen presentedwith a textbook-likestorysituationas a stimulusfor questiongeneration.However,the findingthatabout25%
of the responseswere eitherstatementsor nonmathematical
questionssuggeststhat
the taskwas a novel one for manystudents.Nevertheless,even thoughthe taskwas
mathnovel, about80%of the studentswere ableto generateat leastone appropriate
ematical question and nearly 60% of the students generatedthree appropriate
mathematicalquestions.Moreover,a considerablenumberof studentswere able
to generatesyntacticallyand semanticallycomplex mathematicalproblems.
The findings of this study provide some insights into the natureof mathematical problem-posing processes and also into some aspects of the relationship
between problemposing andproblemsolving. Moreover,the study also offers an
approachto problemposing and an analyticscheme thatmay be of use to practitionersas well as to researchersinterestedin examiningtheposingof arithmeticstory
problems.Each of these contributionsis discussed next.
Problem-PosingProcesses
The analysesconductedin this studyprovidesome hints aboutthe natureof the
processesstudentsmay use when generatinga series of arithmeticstoryproblems.
Kilpatrick(1987) argued that one of the basic cognitive processes involved in
problemposing is association:"[Because]knowledgeis representedas a network
of associatedideas, thatnetworkcan be used to generateproblemsby takinga conceptnodein thenetworkandraisingquestionsaboutits associates"(p. 136).Thefindings of this study suggest anotherway in which associationappearsto play a role
in problemposing. In the task studiedhere, studentswere askedto generatethree
questionsfromthe given information,anda substantialportionof theresponsesgave
evidence of relationshipsor associationsbetween and amongresponses.In particular,morethanhalfof the studentsgeneratedthreemathematicalquestions,andgenproblems.Thus,once studentsbegangenerallythesewereall solvablemathematical
in nature,they tended to continue to
mathematical
that
were
eratingproblems
the
such
generate
problems.Moreover, findingthatnearlyhalfthe studentsgave symmetricresponsesor chainedresponsesis directlysupportiveof a view thatthe subjects in this studygeneratedproblemsusing a processof association.Manystudents

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

EdwardA. Silver and Jinfa Cai

535

appearedto generatetheirsecond andthirdresponsesby using theirfirst response


as a cue.
As noted, the complexity of posed problemstendedto increaseacrossresponse
occasions. Much of this increasedcomplexity is due both to the tendency of students to pose a first problemrelatedto Elliot-a problemthat tendedto be fairly
simple-and to thetendencyof manystudentsto generatesequencesof chainedproblems thatmoved from simplerto more complicatedrelatedproblems,with the latterproblemsbasedon theresultsof the simplerproblemsposedearlierin thesequence.
Althougha complex-to-simplesequenceof chainedproblemswas also theoretically
possible, it was rarelyobservedin the responsesobtainedin this study. Thus, the
findings of this study suggestlikely problem-posingprocesses,butthe datado not
allow for definitiveanalysesof the thinkingof subjectsas they generatedtheirproblem sequences.It wouldbe interestingto know, for example,if the studentshadthe
final problemin mind as the targetproblemandsimply wrotethe simplerproblems
first, perhapsas a resultof instructionalexperience,or if the more complex problems emergedfromtheirinitialgenerationof a simplerproblem.Moredetailedinvestigationof the processby which a studentgeneratesher or his firstposing response
(andthensubsequentresponses),in settingslike andunlikethe one usedin this study,
is warranted.
The findingthatso many studentsgeneratedrelatedproblemsis quiteinteresting
becausepriorresearchby Ellerton(1986) suggestedthatwell-plannedproblemgenerationwas a characteristic
of high-performing
mathematicsstudentsbutwas largely
absentin low-performingstudents.The subjectsof this studywere studentsattendcommunities,
ing schoolsin economicallydisadvantaged
settingsin whichmathematical
performancetendsto lag behindthatfoundin more affluentcommunities(Secada,
these studentswere not chosenfor this studybecausethey had
1992). Furthermore,
some specialmathematicalability.Consequently,the findingssuggestthata capacity for thoughtfulproblemposing may be moregenerallyavailableto studentsthan
mighthavebeeninferredfromotherresearchreports(e.g., Ellerton,1986;Krutetskii,
instruc1976).Becausethesedatawerecollectedduringthefirstyearof a mathematics
tionalreformprojectin the schools attendedby these students,it is possiblethatthe
findingsarethe result,in some part,of the influenceof thatproject.Yet, even if the
findingsarebecauseof the earlyinfluenceof a mathematicsinstructionalreforminitiativein the schools,theresultsareencouraging,in thattheysuggestthe likelyaccessibilityof problemposing for all middleschool students.
ProblemPosing and Problem Solving
The findingsof this studyalso contributeto an emergingunderstanding
of therelationshipbetweenproblemposing andproblemsolving. Students'problem-solving
was highlycorrelatedwiththeirproblem-posing
performance
performance.
Compared
to less successfulproblemsolvers,good problemsolversgeneratedmoremathematical
problems,and theirproblemswere more mathematicallycomplex. Moreover,the
of probleminformation(e.g., convertingrelationalpropositions
misrepresentations
to assignmentpropositions,as in rendering"Arturodrove80 miles morethanElliot"

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

536

ProblemPosing by MiddleSchool Students

as "Arturodrove80 miles")evidentin the statementsandnonmathematical


questions
generatedby the poorerproblemsolverssuggestat leastone information-processing
deficiencythatmaycontributeto theirgenerallypoorerproblem-solving
performance.
Althoughtheseresultsarenot surprising,they lend empiricalsupportto the theoretical argumentofferedby Kilpatrick(1987) thatthe qualityof the questionsstudents
posemightserveas anindexof how well theycansolveproblems.Althoughthis study
has providedcredibleevidence of a directlink betweenproblemposing andproblem solving, understandingthe depthandnuancesof the relationshipbetweenposing and solving certainlydeserves more researchattention.
Do studentspose only problemsthatthey can solve, or thatthey thinkthey can
solve? If so, then one would expect a stronglink betweenposing and solving, and
therewould be furthersupportfor the argumentthatposing could be taken as an
index of solving. Because subjectsin this studywere not askedto solve the problems they posed, thereis no directevidence to answerthis question.Indirectsupportfor an affirmativeansweris provided,however,by the findingthatthe majority of the questionsstudentsposed could be solved using the informationgiven in
the situation. Nevertheless, mathematicianscertainly pose mathematicalproblems or conjectures that they are not certain they can solve (e.g., Goldbach's
Conjecture),andresearchwith adultsubjectshas foundthatthey often pose mathematicalproblemsthatthey couldnot solve on theirown (Silveret al., 1996). Thus,
furtherinvestigationof this relationshipis warranted,with special attentionbeing
paid to the circumstancesunderwhich the problemposing occurs.
Potential Contributions of the Task and the Analysis Scheme

The simple taskused in this studywas successfulin evoking considerablearithmetic problem-posingactivityon the partof middle school students.Furthermore,
the applicationof the semanticcategoryandlinguisticcomplexityanalysesconducted
in thisstudy,andtheexaminationof problemrelatedness,
provedusefulin illuminating
mathematical
the
students'
of
problemposing.Nevertheless,it is cermanyaspects
used in this study. Othermeato
the
scheme
are
limitations
true
that
there
tainly
sures could have been used to assess the linguistic and mathematicalcomplexity
of the posed problems.For example,linguisticcomplexitycould have been determinedby measuringtextcoherenceratherthanby examininglinguisticpropositional
structures.Similarly,it can be arguedthat semanticcategoryanalysis is only one
way to examine the mathematicalcomplexity of the posed problems;the method
developedby Leung (1993) used GPS graphsinvolving associatedcountsof problem objects and operators,is a valid alternativeand may have yielded somewhat
differentresults.Moreover,the schemeis certainlylimitedto arithmeticstoryproblems, andthis is only one type of problemthatmay be of interestto researchersand
teachers.Despitethese limitations,however,the schemedevelopedandused in this
thatarithmeticstoryproblemsgeneratedby middleschool
studyclearlydemonstrates
studentscan be analyzedand not merely described.
The analyticschemeused in this studyoffers teachersandresearchersa credible
meansby whichto examinethe complexityandsophisticationof the arithmeticstory

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

EdwardA. Silver and Jinfa Cai

537

problemsthatstudentspose. Fromtheperspectiveof educationalpractice,the scheme


might be used or adaptedfor use by teachers in orderto evaluate the effectiveness of their posing-orientedinstructionor to measurethe progress of their students in problemposing over time. From the perspective of research,it is likely
thatthis scheme could be useful not only in subsequentstudiesof arithmeticproblem posing but also as a prototypefor analysesof problemposing in domainsother
thanthatof arithmeticstory problems.In general,the demonstratedfeasibility of
adapting some of the techniques proven useful in the study of problem solving
to the analysisof problemposing shouldencouragefurtheradaptationand analysis in the studyof problemposing in a broadrangeof mathematicaldomains.This
would, in turn,resultin a betterunderstandingof the cognitive processesinvolved
in mathematicalproblem posing.
Anothercontributionof this studymay also be foundin the marginsof the analyses reportedhere.About25%of theresponsesweregiven in a formotherthanmathematicalquestions.The dataanalysis schemes used in this studytreatedthe statements and nonmathematicalquestions as being of marginal interest, yet these
responsesactuallyconstitutea potentiallyinterestingdatasourcefor some general
issues relatedto mathematicalproblemposing.Some studentsmay haveposed nonmathematicalquestionsor statementsbecause of the novelty of the task and their
lack of experiencewith problem-posingactivities or because of ambiguityin the
directions,which simply askedthemto pose "questions,"but some may have done
so as a legitimateresponseto what they perceivedto be the task demandsor as a
way of queryingimportantaspectsof the quantitativerelationshipsin the given situation.Forexample,some studentsappearto have treatedthetaskas a readingcomprehension exercise and generated statementscontaining interpretationsof, or
inferences from, the given informationin the situation (e.g., Elliot drove 100
miles).Yet otherstudentsposednonmathematical
questionsinquiringabouttheunderrationale
for
the
information
or
lying
given
askingfor supplementalinformation(e.g.,
Wherewerethesepeoplegoing?).The responsesof some otherstudentssuggestthat
issuesrelatedto theirpersonalcommitmentsandvalues(e.g., morality,justice,human
to themas issuesof formalmathematics.
For
mayhavebeenas important
relationships)
some
studentsrevealedan apparentconcernabouttheequitabledistribution
example,
of drivingresponsibilitieswhentheyposedthefollowingkindsof questions:"If they
each drivean equalamount,how manymiles wouldeachpersondrive?""Whydoes
Arturodrive so long?""Whydid Elliot drive twice as far as Jerome?"
Although it is not possible to know precisely the underlyingreasonsfor these
unexpectedresponses,theirappearancesuggeststhatan open-endedproblem-posing
task,whichinvitesstudentsto expresstheirown questions,mayleadto outcomes(i.e.,
posedproblemsorquestions)differentfromtheonesa teacherorresearcher
mighthave
in mind.Theseresponsessuggestthepowerof problemposing,even a simpletasklike
the one used in this study,as an experiencein whichpeopleexpressthemselveswith
situationsorideas(Silver,1994).Andtheseresponsesalsosugrespectto mathematical
gest the complexeducationalandresearchchallengesconnectedwith understanding
whatthe posed problemsthemselvesrepresentas productsof humanactivity.

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

538

Problem Posing by Middle School Students

REFERENCES
Brown,S. I., & Walter,M. I. (1993).Problemposing:Reflectionsandapplications.Hillsdale,NJ:Erlbaum.
Charles,R. I., & Silver,E. A. (Eds.).(1988). Theteachingand assessing of mathematicalproblemsolving. Reston, VA: NationalCouncil of Teachersof Mathematics.
Cocking,R. R., & Mestre,J. P. (Eds.).(1988).Linguisticand culturalinfluenceson learningmathematics.
Hillsdale, NJ: Erlbaum.
Ellerton,N. F. (1986). Children'smadeup mathematicsproblems:A new perspectiveon talentedmathematicians.EducationalStudiesin Mathematics,17, 261-271.
Hashimoto,Y. (1987). Classroompracticeof problemsolving in Japaneseelementaryschools. In J. P.
Becker & T. Miwa (Eds.), Proceedings of the U.S.-Japan Seminar on Mathematical Problem
Solving (pp. 94-119). Columbus,OH: ERIC/SMEACClearinghouse(ED 304-315).
Healy, C. C. (1993). Creating miracles: A story of student discovery. Berkeley, CA: Key
CurriculumPress.
Keil, G. E. (1965). Writingand solving originalproblemsas a means of improvingverbal arithmetic
IndianaUniversity,1964).Dissertation
Abstracts
problemsolvingability(Doctoraldissertation,
International,
25(12), 7109.
Kilpatrick,J. (1987). Problemformulating:Wheredo good problemscome from?In A. H. Schoenfeld
(Ed.), Cognitivescience and mathematicseducation(p.123-147). Hillsdale, NJ: Erlbaum.
Krutetskii, V. A. (1976). The psychology of mathematicalabilities in school children. Chicago:
Universityof Chicago Press.
Lane, S. (1993). The conceptualframeworkfor the developmentof a mathematicspeformanceassessment instrument.EducationalMeasurement:Issues and Practice, 12(2), 16-23.
Lane, S., & Silver, E. A. (1995). Equityand validity considerationsin the design and implementation
of a mathematicsperformanceassessment:The experienceof the QUASAR project.In M. Nettles,
& A. L. Nettles (Eds.), Equityand excellence in educationaltestingand assessment(pp. 185-219).
Boston, MA: Kluwer.
Leung, S. S. (1993). The relationof mathematicalknowledge and creativethinkingto the mathematical problemposing of prospectiveelementaryschool teacherson tasks differingin numericalinformation content (Doctoral dissertation, University of Pittsburgh, 1993). Dissertation Abstracts
International,54(06), 2082.
Maddon,P. J. (1994). Making storyproblemsrelevant.ArithmeticTeacher,41, 526-527.
Magone, M., Cai, J., Silver E. A., & Wang, N. (1994). Validatingthe cognitive complexity and content quality of a mathematics performance assessment. International Journal of Educational
Research,21(3), 317-340.
Marshall,S. P. (1995). Schemasin problem solving. New York:CambridgeUniversityPress.
Mayer,R. E., Lewis, A. B., & Hegarty,M. (1992). Mathematicalmisunderstandings:Qualitativereasoningaboutquantitativeproblems.In J. I. D. Campbell(Ed.),Thenatureand originsof mathematical
skills (pp. 137-154). Amsterdam:Elsevier.
National Council of Teachers of Mathematics. (1989). Curriculumand evaluation standardsfor
school mathematics.Reston, VA: Author.
NationalCouncilof Teachersof Mathematics.(1991). Professionalstandardsfor teachingmathematics.
Reston, VA: Author.
Newell, A., & Simon, H. A. (1972). Humanproblemsolving. Englewood Cliffs, NJ: PrenticeHall.
Perez,J. A. (1986). Effects of student-generatedproblemson problemsolving performance(Doctoral
dissertation,ColumbiaUniversity, 1985). DissertationAbstractsInternational,46(10), 2954.
Schoenfeld, A. H. (1985). Mathematicalproblemsolving. Orlando,FL: Academic Press.
Scott, N. C. (1977). Inquiry strategy, cognitive style, and mathematicsachievement. Journalfor
Research in MathematicsEducation,8, 132-143.
Secada, W. G. (1992). Race, ethnicity,social class, language,and achievementin mathematics.In D.
A. Grouws(Ed.),Handbookof researchon mathematicsteachingand learning(pp. 623-660). New
York:Macmillan.
Silver,E. A. (1985). Teachingand learningmathematical
problemsolving:Multipleresearchperspectives.
Hillsdale, NJ: Erlbaum.

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

EdwardA. Silver and Jinfa Cai

539

Silver,E. A. (1994). On mathematicalproblemposing. For theLearningofMathematics,14 (1), 19-28.


Silver,E. A., & Lane,S. (1993). Assessmentin the contextof mathematicsinstructionreform:The design
of assessmentin the QUASAR project.In M. Niss (Ed.),Assessmentin mathematicseducationand
its effects (pp. 59-70). London:Kluwer.
Silver, E. A., Mamona-Downs,J., Leung, S., & Kenney,P. A. (1996). Posing mathematicalproblems:
An exploratorystudy.Journalfor Research in MathematicsEducation,27, 293-309.
of thepossible"in mathematics
Silver,E. A., & Stein,M. K. (1996).TheQUASARproject:The"revolution
instructionalreformin urbanmiddleschools. UrbanEducation,30(4), 476-521.
Silverman,F. L., Winograd,K., & Strohauer,D. ( 1992). Student-generated
storyproblems.Arithmetic
Teacher,39, 6-12.
Stanic, G. A., & Kilpatrick,J. (1988). Historicalperspectiveson problemsolving in the mathematics
curriculum.In R. I. Charles& E. A. Silver (Eds.), Theteachingand assessing of mathematicalproblem solving (pp. 1-22). Reston, VA: NationalCouncilof Teachersof Mathematics.
van den Brink,J. F. (1987). Childrenas arithmeticbook authors.For the LearningofMathematics,7,
44-48.
Winograd,K. (1991). Writing,solving and sharingoriginalmath storyproblems:Case studies in the
cognitive behaviorof fifth-gradechildren(Doctoraldissertation,Universityof NorthernColorado,
1990). DissertationAbstractsInternational,51(10), 3324.

AUTHORS
EDWARD A. SILVER, Professor of Cognitive Studies and Mathematics Education, School of
Education; and Senior Scientist, Learning Research and Development Center; University of
Pittsburgh,Pittsburgh,PA 15260
JINFA CAI, Assistant Professor, Departmentof MathematicalSciences, University of Delaware,
Newark,DE 19716

This content downloaded on Fri, 1 Mar 2013 15:19:21 PM


All use subject to JSTOR Terms and Conditions

Vous aimerez peut-être aussi