Mathematical Tasks and Student Cognition: Classroom-Based Factors That Support and Inhibit

High-Level Mathematical Thinking and Reasoning

Author(s): Marjorie Henningsen and Mary Kay Stein
Source: Journal for Research in Mathematics Education, Vol. 28, No. 5 (Nov., 1997), pp. 524-
549
Published by: National Council of Teachers of Mathematics
Stable URL: [Link] .
Accessed: 26/09/2014 14:47

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
[Link]

.
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@[Link].

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

[Link]

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
Journalfor Researchin MathematicsEducation
1997, Vol. 28, No. 5, 524-549

Mathematical Tasks and Student

Cognition: Classroom-Based Factors
That Support and Inhibit High-Level
Mathematical Thinking and Reasoning

MarjorieHenningsen,Universityof Pittsburgh
MaryKay Stein, Universityof Pittsburgh

In orderto developstudents'capacitiesto "domathematics,"

classroomsmustbecomeenvironments
in whichstudentsareableto engageactivelyin rich,worthwhilemathematical [Link]
focuses on examiningandillustratinghow classroom-basedfactorscan shapestudents'engage-
ment with mathematicaltasks that were set up to encouragehigh-level mathematicalthinking
[Link] findingssuggestthatwhen students'engagementis successfullymaintained
at a high level, a large numberof supportfactorsarepresent.A decline in the level of students'
engagementhappensin differentways andfor a varietyof [Link]-
vide concreteillustrationsof the ways in which students'engagementin high-level cognitive
processes was found to continueor decline duringclassroomwork on tasks.

Duringthe pastdecade,muchdiscussionandconcernhave been focused on lim-

itationsin students'conceptualunderstanding as well as on theirthinking,reasoning,
andproblem-solvingskills in mathematics(Hiebert& Carpenter,1992;Lindquist&
Kouba,1989;NationalResearchCouncil,1989). In responseto these concerns,the
National Council of Teachersof Mathematics(NCTM) has publishedproposed
reformsof curriculum, evaluation,andteachingpracticescommonlyfoundin primary
andsecondaryschool mathematicsclassrooms(NCTM,1989, 1991, 1995).Among
the underlyinggoals of these reformeffortsareto enhancestudents'understanding
of mathematicsandto help thembecome bettermathematicaldoersandthinkers.
Whatdoes it meanto be a mathematicaldoerandthinker?Answersto this ques-
tion dependon one's view of the natureof mathematics.A view of mathematicsthat
hasgainedincreasing acceptanceoverpastyearsis onebasedon a dynamicandexploratory
stancetowardthediscipline(Romberg,1994).Thisdynamicstancetowardmathematics
requiresone to focuson theactive,generativeprocessesengagedin by doersandusers
of mathematics(Schoenfeld,1992), ratherthanview mathematicsas a static,struc-
turedsystemof facts,procedures,[Link]

The authorswish to acknowledgethe following individualsfor theirhelpfulcommentson

anearlierdraftof thispaper:EdwardSilver,CatherineBrown,ElizabethGeorge,BarbaraGrover,
MargaretSmith, Michele Saulis, and Marcia Seeley. We would also like to thank Gay
Kowalfor hergenerousassistancein preparingthe [Link] this paperwas sup-
portedby a grantfrom the FordFoundation(grantno. 890-0572) for the QUASAR project.
Any opinionsexpressedhereinarethoseof the authorsanddo not necessarilyreflectthe views
of the Ford Foundation.

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
MarjorieHenningsenand Mary Kay Stein 525

involve the use of mathematicaltools systematicallyto explorepatterns,frameprob-

lems, andjustify reasoningprocesses (Burton, 1984; NationalResearchCouncil,
1989; Romberg, 1992; Schoenfeld, 1992, 1994).
This moredynamicnotionof mathematical activityhasimplicationsfor ideasabout
what students need to learn and the kinds of activities in which students and
teachersshould engage duringclassroom [Link]'learningis seen
as the process of acquiringa "mathematicaldisposition"or a "mathematicalpoint
of view" (Schoenfeld, 1992, 1994), as well as acquiringmathematicalknowledge
andtools for workingwith [Link] mathematicaldis-
position is characterizedby such activitiesas looking for andexploringpatternsto
understandmathematicalstructuresandunderlyingrelationships;using available
resourceseffectively and appropriatelyto formulateand solve problems;making
sense of mathematicalideas, thinkingandreasoningin flexible ways:conjecturing,
generalizing,justifying, andcommunicatingone's mathematicalideas; anddecid-
ing on whethermathematicalresultsarereasonable(Schoenfeld,1992).Theseactiv-
ities have much in common with the active reasoning processes that Resnick
(1987) andothershave proposedas characteristicsof high-levelthinkingin a vari-
ety of [Link] studentsareto developthesecapacities,thenclassrooms
mustbecome environmentsin which they have frequentopportunitiesto engage in
dynamicmathematicalactivitythatis groundedin rich, worthwhilemathematical
tasks (NCTM, 1991; Schoenfeld, 1994).

ImportanceofMathematicalInstructionalTasks
Mathematicaltasksare centralto students'learningbecause "tasksconvey mes-
sages aboutwhatmathematicsis andwhatdoingmathematicsentails"(NCTM,1991,
p. 24). The tasksin which studentsengageprovidethe contextsin which they learn
to think about subject matter,and differenttasks may place differing cognitive
demandson students(Doyle, 1983;Marx& Walsh, 1988;Hiebert& Wearne,1993).
Thus,thenatureof taskscanpotentiallyinfluenceandstructuretheway studentsthink
andcan serveto limitor to broadentheirviews of the subjectmatterwithwhichthey
[Link] whatit meansto "domathematics" from
theiractualexperienceswith mathematics,andtheirprimaryopportunitiesto expe-
riencemathematicsas a disciplineareseatedin theclassroomactivitiesin whichthey
engage (Schoenfeld,1992, 1994).
How feasible is it to engage students consistently and successfully in high-
level tasks for "doingmathematics"in the classroom?Academic taskresearchers
(Doyle 1983, 1986, 1988) have noted thathigh-level tasks are often complex and
longerin durationthanmoreroutineclassroomactivitiesandarethusmoresusceptible
to various factors that could cause a decline in students' engagement to less
demandingthoughtprocesses. A previous study of mathematicaltasks in reform
classrooms at the middle-school level (Stein, Grover, & Henningsen, 1996)
obtainedresultsthatfurthersubstantiatedthe difficulty of maintaininghigh levels
of students'cognitive [Link] findings
suggest thatalthoughattentionto the natureof mathematicalinstructionaltasks is

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
526 Classroom-Based
Factors

important,attentionto the classroomprocesses surroundingmathematicaltasksis

[Link] been writtenabouttypes of mathematicaltasks
thataffordstudentsopportunitiesto do mathematics,therehave been fewer inves-
tigationsof the kindsof instructionalenvironmentsrequiredto supportthe imple-
mentationof tasksfor "doingmathematics." Thispaperaddressesthatgapby focusing
on classroom-basedfactorsthatinfluencethe ways in which studentsengage with
cognitively demandingmathematicstasks in real classroomsettings.

DifficultiesAssociated with ImplementingHigh-LevelTasks

Tasksthatareset up to engage studentsin cognitivelydemandingactivitiesoften
evolve into less demandingforms of cognitive activity (Doyle 1983, 1986, 1988).
Engagingin high-levelreasoningandproblemsolvinginvolvesmoreambiguityand
higherlevelsof personalriskfor studentsthando [Link]-
mentcan evoke in studentsa desirefor a reductionin taskcomplexitythat,in turn,
can lead them to pressureteachersto furtherspecify the proceduresfor complet-
ing the task or to relax accountabilityrequirements(Doyle, 1983). Theremay also
be a tendencyfor classroom-based workon tasksto driftawayfroma focus on mean-
ing and understandingtowardan emphasison accuracyand speed (Doyle, 1988).
Anotherfactorunderlyingunsuccessfultaskimplementationis a lack of alignment
betweentasksandstudents'priorknowledge,interests,andmotivation(Bennett&
Desforges, 1988). Such mismatchesmay cause studentsto fail to engage with the
task in ways thatwill maintaina high level of cognitive activity.
In general,a complexarrayof factorsis involvedin orchestrating classroomactiv-
ity and balancing classroom management needswith academicdemands. Factorscan
be rootedin the way classroomnormsareset up, in the motivationandlearningdis-
positionsof students,andin the generalclassroommanagementpracticesof teachers.
These factorsincludethe mannerin whichorderis establishedin the classroom,the
physicalorganization of theenvironment,theamountof timeallottedforvariousactiv-
ities,themannerin whichtransition periodsbetweentasksarehandled,theestablishment
of accountabilitystructures,andthe ways in which disciplineinterventionsarehan-
dled (Doyle, 1986).However,tasksthatbeginas cognitivelydemandingones do not
alwaysdecline,andit is equallyimportant to understandwaysin whichhigh-levelcog-
nitivedemandscan be maintainedas the tasksareimplementedin the classroom.

Waysof SupportingImplementationof High-LevelTasks

When students"do mathematics"in the classroom,the activityhas most likely
not occurredin a vacuum. Factors that contributeto the decline of high-level
demands,when consideredin the reverse,can point to ways of maintaininghigh-
level [Link],Doyle (1988) arguedthatteachersshouldbe especially
attentiveto the extentto which meaningis emphasizedandthe extentto which stu-
dentsareexplicitlyexpectedto demonstrate understandingof the mathematicsunder-
lying the activitiesin which they [Link] an emphasiscan be maintained
if explicit connections between the ideas
mathematical and the activitiesin which
[Link] whatstudentsalreadyknow

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
MarjorieHenningsenand Mary Kay Stein 527

andunderstandalso play an importantrolein engagingstudentsin high-levelthought

processes (Hiebert& Carpenter,1992). Some researchershave pointedout that if
cognitively demandingtasks are appropriatewith respect to students'levels and
kindsof priorknowledge,students'cognitiveprocessingduringtaskimplementation
standsa betterchanceof remainingat a highlevel (e.g., Bennett& Desforges,1988).
Also, structuringclassroom activity so that appropriateamounts of time are
devoted to tasks is important(Doyle, 1986).
Ina reviewof researchon classroominstruction forhigh-levelunderstanding,Anderson
(1989) supported these ideas as well as some [Link] noted the importance
of the Vygotskiannotionof scaffoldingin helpingstudentsto understandandmake
connections among importantideas. Scaffolding occurs when a student cannot
workthrougha taskon his or her own, anda teacheror morecapablepeerprovides
assistancethatenablesthe studentto completethetaskalone,butthatdoes notreduce
the overallcomplexityor cognitivedemandsof the [Link], teacherscan support
high-levelthinkingprocessesin studentsby explicitlymodeling(orby havinga stu-
dent model) such processes andthinkingstrategies(Anderson,1989). Finally,it is
importantto encouragestudentsto engage in self-monitoringor self-questioningas
they progressthrougha task(Anderson,1989; Schoenfeld,1983;Silver,Branca,&
Adams, 1980). Self-monitoringcan increasestudents'feelings of competenceand
controland,in turn,theirmotivationto remainengagedwith a task at a high level.
Thesefindingssuggestthatthe merepresenceof high-levelmathematicaltasksin
theclassroomwill notautomatically resultin students'engagement in doingmathematics.
Withoutengaging in such active processes duringclassroominstruction,students
cannotbe expected to develop the capacityto think,reason,andproblemsolve in
mathematicallyappropriateand powerful ways. Clearly, the ambientclassroom
environmentmustactivelysupportsuccessfulengagementof studentsin high-level
thinkingand reasoning.
This paperinvestigatesthe classroomfactorsthat eitherhinderor supportstu-
dents'engagementin high-level mathematicalthinking and reasoningfor doing
[Link] context for the presentinvestigationconsists of mathematics
classroomsthatareparticipatingin the QUASAR project',a nationaleducational
reformprojectaimed at fosteringand studyingthe developmentand implementa-
tion of enhancedmathematicsinstructionalprogramsfor studentsattendingmid-
dle schoolsin economicallydisadvantagedcommunities(Silver& Stein, 1996).The
projectis basedon the premisethatpriorfailuresof poorandminoritystudentsare
due to a lack of opportunitiesto participatein meaningfuland challenginglearn-
ing experiencesratherthanto a lack of abilitiesor [Link] the fall of 1990,
groupsof mathematicsteachersat six geographicallydispersedandethnicallydiverse
urbanmiddle schools have been working,in collaborationwith resourcepartners
from nearbyuniversities,to enhancetheir own local instructionand professional

'QUASAR (QuantitativeUnderstanding:AmplifyingStudentAchievementand Reasoning)is based

at theLearningResearchandDevelopmentCenteratthe Universityof Pittsburgh
andis directedby Edward
A. Silver.

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
 528 Classroom-BasedFactors

developmentprogramsin orderto providetheirstudentswithgoodmathematics instruc-

tionaimedatfosteringthinking,reasoning,[Link]
teachershave receiveda broadarrayof staff developmentsince the inceptionof the
project, the teachers' educationaland professionalbackgroundswere typical of
most middleschool mathematicsteachers(QUASARDocumentationTeam, 1993).
Comparedto the nationalprofile,QUASARteachersaremoreethnicallydiverse.

CONCEPTUALFRAMEWORK
The presentstudyis guidedby a conceptualframeworkbasedon the constructof
a mathematicalinstructional task(see Steinet al., 1996,for a moredetailedoverview
of thecontentanddesignof theframework). Theframework, shownin Figure1, defines
a mathematical taskas a classroomactivity,thepurposeof whichis to focus students'
attentionon a particularmathematicalconcept,idea, or skill.

Mathematical task
Mathematical task as implemented by
Mathematicaltask as set upby studentsin
as represented theteacher theclassroom
in
curricular/instructional intheclassroom Enactment of
materials * Taskfeatures taskfeatures
Cognitivedemands * Cognitive
processing
tudents'learning
outcomes

Factors Factors
setup
influencing influencingstudents'
implementation
Teachers' goals Classroom norms
Teachers' knowledge Taskconditions
of subjectmatter Teachers'instructional
Teachers'knowledge dispositions
of students Students'learning
dispositions

Figure 1. Relationshipsamong varioustask-relatedvariablesand students'learningoutcomes.

The shadedportionrepresentsthe areaprimarilyunderinvestigation.

Inthisframeworkmathematical taskspassthroughthreephases(represented by the

rectangularboxesin Figure1):as written
by curriculum as set
developers, upby theteacher
in theclassroom,andas implemented by [Link] fur-
therspecifiestwo dimensionsof [Link] firstdimensionis taskfea-
[Link] aspectsof tasksthatmathematics educatorshave identified

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
MarjorieHenningsenand Mary Kay Stein 529

as importantconsiderationsfor thedevelopmentof mathematicalunderstanding, rea-

soning, and sense making. These featuresinclude multiple solutionstrategies, mul-
tiplerepresentations,andmathematical communication. Duringtheset-upphase,these
featuresreferto the extentto whichthe taskas announcedby the teacherencourages
studentsto use more thanone strategy,to use multiplerepresentations,andto sup-
ply [Link] implementationphase,thesefeatures
referto the extentto which studentsuse the features.
The second dimension, cognitive demands, refers to the kind of thinking
processes entailedin solving the task as announcedby the teacher(duringthe set-
up phase) and the thinkingprocesses in which studentsengage (duringthe imple-
mentationphase).These thinkingprocessescan rangefrommemorizationto the use
of proceduresand algorithms(with or without attentionto concepts, understand-
ing, or meaning)to complex thinkingand reasoningstrategiesthatwould be typ-
ical of "doingmathematics" (e.g.,conjecturing,justifying,or interpreting).
Thepresent
investigation focused on this second dimension of cognitive demands and the
classroom-basedfactorsthatinfluencedthemas taskspassedfromthe set-upto the
implementationphase.
Accordingto the framework,the featuresandcognitive demandsof taskscan be
transformedbetween any two successive phases. For example,a task could be set
up to requirehigh-level cognitive activityby students,but duringthe implementa-
tion phaseit couldbe transformedin sucha way thatstudents'thinkingfocusesonly
on procedures,with no [Link] shadedcircle in Figure 1 rep-
resents the classroom-basedfactors that influence the ways in which students'
thinkingunfoldsduringthe task-implementation [Link] factorsincludeclass-
room norms,task conditions,and teachers'and students'[Link]
normsrefersto theestablishedexpectationsregardinghow academicworkgets done,
by whom,andwith whatdegreeof [Link]
to attributesof tasks as they relateto a particularset of students(e.g., the extent to
which tasks build on students' prior knowledge and the appropriatenessof the
amountof time thatis providedfor studentsto completetasks). Teacherandstudent
dispositionsrefersto relativelyenduringfeaturesof pedagogicalandlearningbehaviors
thattendto influencehow [Link]
includethe extentto whicha teacheris willingto let a studentstrugglewitha difficult
problem,thekindsof assistancethatteacherstypicallyprovidestudentswho arehaving
difficulties,andtheextentto whichstudentsarewillingto perseverein theirstruggleto
[Link],task,andteacherandstudentfactors,
taskscanbe shapedby the ambientclassroomculture.

PURPOSEOF THE STUDY

The purposeof the presentstudyis to identify,examine, andillustratethe ways
in which classroom-basedfactors shape students' engagement with high-level
[Link](Steinet al., 1996)has identifiedvariouspatterns
of studentengagementwith tasksthatwere set up to encouragedoing mathematics.

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
530 Classroom-BasedFactors

In some cases, studentsengagedactivelyin high-level cognitiveprocessescharac-

teristic of doing [Link] othercases they did not. In those cases in which
students'engagementwith the tasks did not exemplify doing mathematics,three
characteristictypes of student engagement were noted: cognitive activity that
focusedon themechanicaluse of procedures(withno connectionto underlyingmean-
ing), unsystematicexploration,and activity with no mathematicalfocus.
This studyidentifiesanddescribesboth those profiles of factorsthatwere asso-
ciatedwith maintaininghigh levels of cognitivedemandandthose with each of the
characteristicpatternsof decline. In addition,the studyincludesclassroom-based
illustrationsof the maintenanceanddeclinepatternsnotedaboveandthe factorpro-
files associatedwith them.

METHODOLOGY
Dataforthepresentstudyweredrawnfromanearlierinvestigationin whichinstruc-
tion in a representativesample of QUASAR classroomswas examined (Stein, et
al., 1996). This earlierstudyfocused on the natureof mathematicaltasks as vehi-
cles for buildingstudentcapacityfor mathematicalthinkingand reasoning.

Prior Investigation
Data sources.Trainedandknowledgeableobservers2wrotenarrativesummaries
of [Link] summariesformedthe primarybasis of the data
used in the [Link] fall of 1990to the springof 1993,
three3-day observationsessions (fall, winter,and spring)were conductedin three
representativeteachers'mathematicsclassroomsat each of fourprojectsites. The
classroom-based illustrationsdescribedin thisstudyweredrawnfromtheclassrooms
of fourprojectteachers:[Link],Ms. Capra,Ms. Hoffman,[Link].
The observertook detailed field notes on the mathematicsinstructionand stu-
dents'reactionsto the instruction;a cameraoperatorsimultaneouslyvideotapedthe
[Link] observations,the observerusedboththe videotapedlesson and
his or her field notes to completethe project'sClassroomObservationInstrument
(COI).3 As partof thatinstrument,the observerprovideddescriptionsandsketches
of the physical setting of the room, a chronology of instructionalevents, and

20Observers selected had strongbackgroundsin mathematicseducation,educationalpsychology, or

a relatedfield; a demonstratedcompetencein analyzinginstruction;priorexperienceobservingclass-
rooms; and an understandingof ethnic or multiculturalissues at the various sites. The observers
underwentextensivetraininganda sampleof theirwrite-upswere reviewedandfeedbackwas provided.
3Theinitialdraftof this instrumentdrewfromtwo mainsources:NCTM'sProfessionalStandardsfor
TeachingMathematics:WorkingDraft(1989), anda classroomobservationsystemused for the Stateof
Californiastudyof elementaryschool mathematics(Cohen,D. K., Peterson,P. L., Wilson, S., Ball, D.,
Putnam,R., Prawat,R., Heaton,R., Remillard,J., & Wiemers,M. (1990). Effectsof statelevel reformof
elementaryschool mathematicscurriculumon classroompractice[ElementarySubjectCenterSeriesNo.
25], EastLansing,MichiganStateUniversity,CenterfortheLearningandTeachingof ElementarySubjects
andthe NationalCenterfor Researchon TeacherEducation).The instrumenthasbeen pilot testedin sev-
eralmiddleschool mathematicsclassroomsandhas undergoneseveralroundsof critiqueandrevisions.

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
MarjorieHenningsenand Mary Kay Stein 531

responsesto questionsassociatedwith five themes:mathematicaltasks,classroom

discourse,the intellectualenvironment,managementandassessmentpractices,and
group work (if it occurred).
In the COI, a mathematicaltask is defined as a segmentof classroomwork that
is devotedto learningabouta particular mathematical [Link]
to segmenttheinstructional timeof eachobservedlessonaccordingto themainmath-
ematicaltaskswith [Link] artifactsassociatedwiththese
tasks were appendedto the [Link] two tasks thatoccupied the largestper-
centages of class time were designatedas Task A and Task B. In the mathemati-
cal taskssectionof the COI,the observerdescribedin detailthe natureof these two
tasks: their mathematicalcontent,the learninggoals of the teacherfor each task,
andthe behaviorsof the studentsas they engagedin these [Link] observeralso
describedthe extent to which each task focused students'attentionon procedural
stepswith or withoutconnectionsto underlyingconceptsandon doingmathematics
(e.g.,framingproblems,makingconjectures, justifying,andexplaining).In theremain-
ing three sections of the COI, the observer wrote aboutall activities thatoccurred
during the classroom lesson (not limited to Task A and Task B), referringspecif-
ically to the two main tasks when appropriate.
[Link] the initial study,a sampleof 144 COIswas selected for
codingwiththe goal of gaininga representative pictureof instructionacrossthefour
projectsitesandthefirst3 [Link] of eachobservationwas coded,
althoughthe entirenarrativesummaryfor the classroomobservationwas reviewed
and consideredin makingcoding decisions.4The COIswere coded using a system
based on the conceptualframeworkshown in Figure 1. The coding systemwas ini-
tiallydevelopedon the basis of a reviewof the literatureon academictasks(Bennett
& Desforges,1988;Doyle, 1983, 1988;Marx& Walsh,1988) andthe cognitivepsy-
chology of instruction(Anderson,1989),the literatureon mathematicalthinkingand
problemsolving (Grouws,1992;Silver, 1985), andmathematicsreformdocuments
(NCTM, 1989, 1991), as well as on knowledgeof the projectsites andtheirgoals.
Nineteen coding decisions, organizedinto four maincategories,were made for
each [Link] includedthe numberof minutesandpercentageof class
time devoted to the task, the type(s) of resource(s)that served as the basis for the
task,the mathematicaltopic of the task (conventionalmiddle-schooltopic, reform-
inspiredtopic, focus on mathematicalprocesses more thana particulartopic), the
context of the task, and whetheror not the task was set up as a collaborativeven-
tureamong students.
Set-up codes were assigned on the basis of a review of the task materials(pro-
vided as appendicesto the COI write-up)andthe task as specified by the teachers,
both duringtheirinitialannouncementsof what studentswere to do andduringthe

4Videotapesof observationsand additionalartifactsfrom an observationwere used as supplemental

data sources on a small numberof the tasks (8%). These sources were consultedwhen a coder deter-
mined thatthe writtendescriptionof the observationdid not provide sufficient informationon which
to base a decision.

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
532 Classroom-Based
Factors

task at any subsequentpointsat whichthe teachersunilaterallyprovidedadditional

specificationsto guide students'approachesto the task. Duringthe set-upphase,
codes wereassignedfor taskfeaturesandforthe cognitivedemandsof the task.(Only
cognitive-demandcodes were used in the presentstudy.) The cognitive demands
were classified into one of the following: memorization;the use of formulas,
algorithms,or procedureswithoutconnectionto concepts,understanding,or mean-
ing;theuse of formulas,algorithms,orprocedureswithconnectionto concepts,under-
standing,or meaning;and cognitive activity that can be characterizedas "doing
mathematics,"includingcomplex mathematicalthinkingand reasoningactivities
such as makingandtestingconjectures,framingproblems,andlookingfor patterns.
Implementation codes also were made for task features and cognitive
demands. These codes were assigned with reference to the ways in which stu-
dents went about working on the task. When coding the cognitive demands of
the task as implemented, coders were asked to make judgments about the
kinds of cognitive processes in which the majority of the students appearedto
be engaged. During this phase, coders independently recognized the need for
a new code to describe a frequently observed mannerof implementing doing-
mathematicstasks, one in which studentsexploredaroundthe edges of significant
mathematicalideas but failed to make systematicand sustainedprogressin devel-
oping mathematicalstrategiesor [Link] the presentarticle,this new
code is called unsystematic exploration.
The finalcategoryof codes includedjudgmentsaboutfactorsassociatedwithtask
implementation. Forhigh-leveltasksthatremainedso duringimplementation, coders
were instructedto identify as many as appliedfrom a list of possible factorsthat
could assistwith the maintenanceof tasksat high levels (e.g., the modelingof high-
level performanceby teachersor capablestudents;sustainedpressforjustification,
explanations,or meaningthroughteacherquestioning,comments,and feedback;
scaffolding[teachersor morecapablestudentssimplifyingthe task so thatit could
be solved while maintainingtask complexity];or the selection of tasks thatbuild
on students'priorknowledge).In the earlierstudy,high level was used to describe
tasks thatinvolved doing mathematicsor the use of formulas,algorithms,or pro-
cedureswithconnectionto concepts,understanding, or [Link]-leveltasks
thatdeclined, coders identifiedthe reasonsfor the decline from a list of possibili-
ties thatincludedthe routinizationof problematicaspectsof the tasks(studentspress
teacherto reducetask ambiguityor complexity by specifying explicit procedures
or the teacher"takesover"difficultpieces of the task);the shiftingof emphasisfrom
meaning,concepts,or understandingto the accuracyandcompletenessof answers;
the lack of sufficienttime for studentsto wrestlewith the demandingaspectsof the
tasks;andclassroommanagementproblemsthatpreventsustainedengagementin
high-level cognitive activities.
The authorsof thisstudy,alongwitha thirdindividual,servedas theprimarycoders
in theinitialstudy.A representativesubset(25%of the 144tasks)weredoublecoded.
Consensus was reachedby the two coders on all [Link]-
bility rangedfrom 53% to 100%,with an averageof 79%.

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
 andMaryKayStein
MarjorieHenningsen 533

Samplingfor the Present Study

In the initial study,58 of the 144 tasks were identifiedas being set up to encour-
age doing [Link] 58 tasksconstitutethe databasefor the presentinves-
tigation. During the implementation phases of these 58 tasks, students were
observedto engage activelyin doing mathematicsin 22 of the [Link] the remain-
ing 36 tasks,students'engagement withthetaskduringimplementation wasnotobserved
to exemplify doing [Link] 8 tasks, students'thinkingfocused on proce-
dureswithoutconnectionsto underlyingmeaning;in 11 tasks,studentsengagedin
unsystematicexploration;andin 10 tasks,students'thinkingwas perceivedto have
no mathematicalfocus. In the remaining7 tasks, students'forms of thinkingdur-
ing the implementationphase representeda variety of categories of cognitive
engagement,no one of which was well enough representedto justify its inclusion
in the presentreporting.

Analysis Proceduresfor the Present Study

In orderto examinethe factorsthatwere associatedwith maintenanceor decline
of the doing-mathematicstasks, we first aggregatedand summarizedthe relevant
subsetof the factorsdatafromthe initialinvestigationaccordingto the maintenance
[Link] thesemaintenanceanddecline
categories,the numberof tasksfor which each factorwasjudgedto be an influence
was calculated. From this information, frequency graphs were constructedin
orderto be able to identify factorprofiles (i.e., the sets of factorsjudged to be pre-
dominantinfluences in the largestpercentageof tasks within each pattern).
After identifying the factor profiles, we returnedto our databaseof narrative
summariesandobservation videotapesto selectclassroomepisodesthatcouldbe devel-
oped into portraitsexemplifying(a) the maintenanceof high-leveltaskdemandsand
the factorsthatsupportthem and (b) each of the threeidentifiedpatternsof decline
and the factors that influence them. After selecting the classroom episodes, we
developeddetailedportraitsdescribingthe natureof the mathematicaltask in each
episode,how the taskwas set up by the teacher,how it was implementedby the stu-
dents, andhow the identifiedfactorsinfluencedthe implementationof the task.

RESULTS
Maintenanceof High-LevelCognitiveDemands
Figure2 shows the percentageof tasks in which each factorwas judgedto be an
influencein assistingstudentsto engage in doing mathematicsfor the 22 tasksthat
remainedat thatlevel duringthe [Link] numeralsat the top of
each barindicatethe numberandpercentageof tasksfor whichthe particularfactor
wasjudgedto be an influencein maintainingcognitivedemandsat the level of doing
mathematics wasjudgedto be aninfluencein 73%of thetasks).Typically,
(e.g.,scaffolding
threeto five factorsper task were believed by the codersto be influencesin assist-
ing studentsto remainengagedin doing mathematicsin particulartasks.

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
534 Classroom-BasedFactors

As shownin Figure2, five factorsappearedto be primeinfluencesassociatedwith

maintainingstudentengagementat the level of doing mathematics:task builds on
students'priorknowledge (82%), scaffolding (73%), appropriateamountof time
(77%),modeling of high-level performance(73%),and sustainedpressfor expla-
nationandmeaning(77%).Thesefindingssuggestthatwhentaskssuccessfullymain-
tain students'engagementin doing mathematics,these factorswould frequentlybe
expected to be in place supportingthathigh-level engagementin the tasks.

82%
90% (n = 18) 73% 77% 73% 77%
80% (n=16) (n= 17) (n=16) (n = 17)
70%
60%
50% 36%
40% (n= 8)
30% 14%
20% (n= 3) 5%
10% (n= 1)

s
~13~)
L ~
,c ~o~ 0O '
?9#j Pi~O

Figure 2. Percentageof tasks on which each factorwas judged to be an influence in assisting stu-
dentsto engage at high levels (totalnumberof tasks= 22; percentagestotalmorethan 100 becausetyp-
ically more thanone factorwas selected for each task)

Thesefindingsarein agreementwiththe moregeneralliteratureon academictasks

[Link] suggestedthattasksthatarelikely to maintainhigh-
level cognitivedemandsaretasksthatbuild on students'priorknowledge(Bennett
& Desforges, 1988) andareallottedan appropriateamountof time for the students
to engage at a high level, thatis, neithertoo little nortoo muchtime (Doyle, 1986).
Teachingbehaviorsthatwere found to supporthigh-level studentengagementin
this study,includingscaffolding,modelinghigh-levelperformance,andconsistently
pressingstudentsto providemeaningfulexplanations,have also been identifiedby
other researchers as importantinfluences in tasks that encourage students to
engage at high levels (Anderson, 1989; Doyle, 1988). These findings demon-
stratethateven thoughstudentswere activelyengagedduringthe tasks(as opposed
to being passive recipients),teachersstill had an importantrole to play in proac-
tively supportingstudents'[Link] otherfactorsthathavebeen
identifiedin the literatureas influentialin high-level engagement,students'self-
monitoring and frequent conceptual connections drawn by the teacher, were

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
 andMaryKayStein
MarjorieHenningsen 535

judged in this study to have less influence;thatis, they were factorsin only 36%
and 14%of the tasks, respectively.

Factor Profilesfor Specific Patternsof Decline

The factorsthatwere associatedwith each of the threetypes of decline areillus-
tratedin Figure3. In this section, we begin by describingthe characteristicfactor
profiles for each of the threetypes of decline. We then look across the threepro-
files of factorsto identify their similaritiesand differences.
Declineintousingprocedureswithoutconnectionto concepts,meaning,and under-
[Link] factorsmost frequentlyjudged to influencethose tasksin which stu-
dents' thinkingprocesses declined into the use of procedureswithoutconnection
to meaningor understandingwere the removalof challengingaspectsof the tasks,
shifts in focus from understandingto the correctness or completeness of the
answer,and inappropriateamountsof time allottedto the tasks. Of these, the fac-
tor most often cited was thatthe challengingaspectsof the taskwere removeddur-
ing the implementationphase,thusnecessitatinglower andless sustainedlevels of
thinking,effort,andreasoningby [Link]-leveltaskscanbe perceived
by students(andteachers)as ambiguous,risky,or both,thereis oftena "pull"toward
reducingtheircomplexityso as to managethe accompanyinganxiety(Doyle, 1988).
Reductionin complexitycan occurin severalways, includingthroughstudents'suc-
cessfully pressuringthe teacherto provideexplicit proceduresfor completingthe
taskor the teacher's"takingover"difficultpieces of the task andperformingthem
for the [Link] is done, however,the cognitive demandsof the task are
weakenedandstudents'cognitiveprocessing,in turn,becomeschanneledintomore
predictableand (often) mechanicalforms of thinking.
Anotherfrequentlycited factorwas a classroom-basedshift in focus away from
meaningandunderstandingtowardthe completenessor accuracyof [Link]
desiredoutcomeof the taskbecomesdefinedby the solutionratherthanby the think-
ing processesentailedin reachingthe [Link]
of both teachers and students often lead to such a narrowpreoccupationwith
solutions, at the expense of [Link] orientationcan easily overwhelm
tasksthatwere initiallyset up to encouragedoingmathematics,especiallyif a focus
on process leads to a slowed instructionalpace andlack of complete participation
by all. Finally,tasksthatdecline into proceduralformsof studentthinkingoften do
so because eithertoo much or too little time is devoted to them. In this situation,
studentshave too little time to grapplewith the importantmathematicalideas con-
tainedin the task. A quick pace gives the impressionof covering much groundin
an efficient mannerbut often robs studentsof the time neededto trulyengage with
the contentand to explore and thinkin ways characteristicof doing mathematics.
Declineintounsystematic exploration.5Thefactorsmostfrequentlyjudgedto influ-
ence the decline of tasks into unsatisfying forms of mathematicalexploration

"5Thereadershouldrecallthatunsystematicexplorationrefersto students'thinkingprocessescharacterized
by unsystematicexplorationandlack of sustainedprogressin developingmeaningor understanding.

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
 536 Classroom-Based
Factors
100%
100%(n = 8)
90% . 75% 75%
80% . (n = 6) (n= 6)
70%
Declinefromdoing mathematicsto
60%
procedureswithoutconnection
50% 38%
(totalnumberof tasks = 8) = 3)
40% 25%(n
40% -25%
30%
.(n=2) 13%
20% (n
10% 0%

C40
~ ~ t

100%
90%
80% 64%
64%
70% (n=7)
Declinefromdoing mathematics 55%
into unsystematicexploration 60% (n = 6) 45%
50%36% (n = 5)
(totalnumberof tasks = 11)
40% (n = 4)
30% 18% 18%
20% (n = 2) (n = 2)

10%0%

0q) S0N 'o Z\ o .

\.90 0060 • , .
, •, ,

100%
90% 80%
80% (n = 8) 70%
70%(n = 7)
Declinefromdoing mathematics
to no mathematicalactivity 60% 50%
50% (n = 5)
(totalnumberof tasks = 10) 40% 30%
(n = 3)
30%
20% 10% 10%
10% (n = 1) (n = 1)
10%0%

oo.,0o"

Figure 3. Percentageof tasks in which each factorwas judged to be an influence in decline.

Percentagestotal more than 100 because typically more thanone factor was selected for each task.

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
MarjorieHenningsenand Mary Kay Stein 537

were inappropriateness of the taskfor the particulargroupof students,inappropriate

amountsof time allottedfor those tasks, andthe removalof challengingaspectsof
the [Link] inappropriateness of thetaskspansa varietyof reasons,includ-
ing low levels of motivation,lack of priorknowledge,andlack of suitablyspecific
[Link] theserelateto the appropriateness of thetaskfor a given group
of [Link] such, they suggest that an importantfactorin the success of high-
level tasks is the considerationof the relationshipbetween students and task;
teachersmustknow theirstudentswell in orderto make intelligentchoices regard-
ing the motivationalappeal,difficultylevel, anddegreeof taskexplicitnessneeded
to move studentsinto therightcognitiveandaffectivespaceso thathigh-levelthink-
ing can occur and progresscan be made on the task.
The second most frequentlycited factorwas inappropriateamountsof time. In
contrast to declines into proceduralizedactivity, for this type of decline (into
unsystematicexploration)the problemwas too much time in the majorityof tasks
for which this factorwas judged to be [Link] studentsareobservednot
to be makingdiscernibleheadwaytowardconstructingandunderstanding key ideas,
additionaltime by itself (i.e., withoutthe introductionof additionalsupportfactors)
appearsto exacerbatethe [Link],taskswereobservedto declineintounsys-
tematicexplorationbecause the challengingaspects of the task were [Link]
these cases, however,the removalof the challengewas less often due to the impo-
sition of a procedureandmore often due to the subtlealterationof the task in such
a way thatthe main point of the activity was lost or overshadowed.
Anotherfactorthatcontributedto the decline of tasks in this category,although
less so, was a lack of accountabilityfor high-levelproductsor [Link]-
ple, studentswere not expected to justify theirmethods,theirunclearor incorrect
explanationswere accepted,andthey were given the impressionthattheirworkon
these taskswould not "count."In such instances,studentscircumventedthe "real"
tasks and tendedto focus only on the work for which they received a grade.
Decline intono [Link] factorsmost frequentlyjudgedto influ-
ence the decline of tasks into activity with no mathematicalsubstancewere inap-
propriatenessof the task, classroom managementproblems, and inappropriate
amountsof time. Interestingly,classroommanagementproblemsappearedto play
a largerole when tasksdeclinedinto a completelack of mathematicalengagement
on the partof the [Link] suggests thatteacherswere strugglingwith keep-
ing studentsundercontrolin additionto keeping them focused on the mathemat-
ics (although the two may be subtly interrelated).Once again, inappropriate
amountsof time were cited, and, in this instance,the problemwas too much time
in the majorityof tasks in which this factorwas cited as influential.
Acrossthethreefactorprofiles,thedeclineintoprocedural thinkingappearsto be asso-
ciatedwiththemostclearlydiscernible,"crispest" pattern [Link]
of
of thethreemainfactors(comparedto therelativelyweakpresenceof theotherfactors)
suggestsa clearpictureof activityin classroomsin whichthesetypesof declineoccur.
Suchsharpdistinctionsamongthepredominance of thevariousfactorsarenotas read-
ily in
identifiable the declinesinto [Link] fact,thisprofileis the

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
538 Classroom-Based
Factors

least crispof all, with severalfactorscontributingmoderatelyto the declines. This

suggeststhatthereis a less readilyapparentset of influencesoperatingin thesecases.
Given thatdeclinesinto unsystematicexplorationwere not anticipatedin ourinitial
coding system,the lack of a crispprofilemay reflectinadequaciesin ourfactorcat-
egories with respectto theirabilityto capturethe kindsof classroomscenariosthat
lead to these types of decline. The factorprofilefor declines into no mathematical
activityis moresharplydifferentiated,butnot as crispas theprofilefor declinesinto
proceduralizedformsof students'[Link] most notablefeatureof this profile
was the strongpresenceof the factorof classroommanagementproblems.
The onemajorfactorthatoccurredacrossthethreetypesof declinewas inappropriate
amountsof time. Thus, it appearsas thoughplanningfor appropriateamountsof
timeandbeingflexiblewithtimingdecisionsas thetaskimplementation phaseunfolds
are extremely importantin order to avoid declines of all types. Major factors
appearingin two out of threeprofiles were the removalof challengingaspects of
the tasks(declinesinto proceduralizedthinkingandunsystematicexploration)and
inappropriatenessof the task for a particularset of students(declines into unsys-
tematicexplorationand into no mathematicalactivity). As mentionedearlier,the
factor"removalof challengingaspectsof tasks"had slightlydifferentformsin these
two types of declines. The factorof inappropriatenessof the task was very broad,
covering a variety of reasons for low mathematicalengagement, all of which
relatedto the appropriatenessof the task for a particulargroupof students.

QualitativePortraits
Our final objective is to provide illustrative,qualitativeportraitsof instruction
thatrepresentthe factorprofiles we found to be associatedwith the instructional
patternsalready described. In constructingthe portraits,we did not limit our-
selves to discussingthe specific mathematicaltasks from a task-analyticperspec-
tive. Instead, we were cognizant of the importanceof considering the overall
learningenvironment,includingthe actionsandinteractionsof the teacherandthe
studentspresentin the classroom.

Maintainingcognitive demandsat the level of doing [Link] overall

goal of this sequenceof lessons was to explorerelationshipsamongfractions,dec-
imals,[Link] thislesson, studentshadexperiencedmodelingfrac-
tions, decimals,andpercentagesusing multiplerepresentations,andthis particular
task representedan extension of thatwork. In the set-upof this task, each student
had a set of rectangulargrids of varioussizes (see Figure4) and was expected to
shade a specified portionof the [Link] portionsto be shadedwere
specified in a varietyof ways includinga percentageof the total, a fractionof the
total,a decimalfractionof the total,or a specific numberof [Link]
thenexpectedto providefor eachshadedregiona fraction,a decimal,or a percentage
thatrepresentedthe amountof the [Link], in a whole-classsetting,
studentswere expectedto be ableto explainone or moresolutionstrategiesfor each
[Link] set up,thetaskprovidedanopportunity to facilitatestudents'construction

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
MarjorieHenningsenand Mary Kay Stein 539

of connectionsamongthethreemodesof representingfractionalquantities(fractions,
decimals,andpercentages)in the contextof exploring,listeningto, interpreting,
jus-
tifying, andexplaininga varietyof solutionstrategiesfor the [Link],the
setupof thetaskwasclearlyorientedtowarddoingmathematics. Duringimplementation
of thetask,thehighcognitivedemandsweremaintainedanda varietyof factorscame
into play to supportstudents'high-levelengagementwith the task.

1. a. Shade .725 of the areaof this rectangle.

b. Whatfractionalpartof the area is shaded?
c. Whatpercentageof the area is shaded?

2. a. Shade 3/8 of the areaof this rectangle.

b. Whatpercentageof the area is shaded?
c. Whatdecimalpartof the area is shaded?

3. a. Shade six of the smallsquares in this rectangle.

b. Whatfractionalpartof the area is shaded?
c. Whatdecimalpartof the area is shaded?
d. Whatpercentageof the area is shaded?

Figure 4. Problemscomposing the mathematicaltask of exploringrelationsamong fractions,dec-

imals, and percentages(Bennett& Foreman,1990)

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
540 Classroom-BasedFactors

The foremostinfluentialfactorwas thatthe taskwas designedto build uponstu-

dents' prior knowledge and experiences with decimals, percentages, and frac-
tions representedin multipleways. Studentshad experienceda previouslesson on
relatingpercentagesto length and areausing a varietyof representations,includ-
ing rectangularareasof varioussizes. Thus studentswere not stymiedby the pre-
sentationof areasotherthantheusual 10 x 10 [Link],theteacher,often
directedstudents'attentionto [Link] example, as a strategy
for solving Problem2a, one studentpresenteda solutionin which he regroupedall
the squares into eight "piles" and shaded three of the piles. To help students
decide whatpercentageand decimalpartswere shaded(Problems2b and 2c), Mr.
Hernandezencouraged them to reason about some basic decimal conversions
they alreadyknew. This encouragementled studentsto use the facts that 1/4 = 2/8
= 25%andthereforethat1/8= 12.5%.So, 3/8wouldhaveto equal37.5%,whichwould
be writtenas .375. In the discussionof Problem3, Mr. Hemandezencouragedstu-
dents to use their priorknowledge to convert6/40 to a percentagewithoutusing a
[Link] studentchangedthe fractionto 12/80and reasonedthatin orderto
get to 100 from 80, he had to add 20, which was 1/4of 80, so he also added 1/4of
12 to the numeratorto obtaina fractionof 15/100 or 15%.Anotherstudentchanged
the fractionsaying, "Ifyou have 6 in 40, divide it in half, andyou'd have 3 in each
part;you'd have 3 in 20." He then multipliedthe numeratorand denominatorby
5 to obtainthe fraction15/100,or 15%.
Anotherkey factorin the students'successfulimplementationof the taskwas the
scaffoldingbehaviorexhibitedby the [Link] able to assiststu-
dents as they reasonedthroughthe problemswithoutreducingthe complexity of
the task at [Link] example,Mr. Hernandezcalled on Michelle to do Problem1
at the [Link] demonstratingProblem la, Michelle shadedin 72.5 of the 80
[Link]. Hernandezdid not immediatelycorrectMichelle's error;instead,he
askedherto explainherthinking,but she saidshe was [Link]
askedherto rereadthe problem,she realizedthatshe mighthave made an error;as
discussionof ProblemIc ensued,[Link] the class coulduse the
informationthattherewere a total 80 [Link] also askedthe studentsto think
aboutwhatwouldhappenif theytriedto distribute 100%acrossthe80 [Link]
thoughtaboutit andrepliedthatthey could find out what percentageeach square
representedandthateach squarewould have to be morethan 1%.Anotherstudent,
Cecily, thoughteach squarewouldbe worth1.25%.Michelleexplainedfurtherthat
therewouldbe 20 left overafterallotting1 to each squareandthat20 dividedamong
the 80 would give 1/4more for each [Link] was then able to show how
many squaresshouldbe shadedfor 72.5 %.Thus,[Link] able to direct
Michelle's (andthe class's) attentionto appropriateaspectsof the task thatwould
enable Michelle to succeed on her own.
At leastthreeotherimportantfactorsthatsupportedstudents'engagementin doing
mathematicswere [Link],Mr. Hernandezallowed an appropriateamountof
time for discussion of the problems,thus affordingstudentsopportunitiesto con-
sideranddiscuss multiplesolutionstrategiesfor the problemsgiven. Forexample,
following Michelle's explanation(describedabove)for Problem1, anotherstudent

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
MarjorieHenningsenand MaryKay Stein 541

multiplied.725 by 80 to get 58 andexplainedthathe obtaineda fractionof 58/80and

reducedit to 29/[Link] dividedthe 80 squaresinto 10 equal
columnsof 8 squareseachandthenshadedin 7 columnsand2 moresquares(because
2 is 1/4Of1/10[of 80], or 1/4of 8 [in this case], which equals .025) for a total of 58
[Link] to use a calculatorto find the solutions.
Anotherfactorevidentin the lesson was modelingof [Link]
factor is well-illustratedby the examples describedabove in Problem 1. The dis-
cussion of multiple solution strategies at the overhead projector provided an
opportunityfor [Link],as well as severalstudents,to modelhigh-levelper-
formanceand to make theirthoughtprocesses explicit. Finally,throughoutall the
discussion of the problems,Mr. Hernandezpressedstudentsto explain theirsolu-
tion [Link] consistentlyrequiredstudentsto attachmeaningto the symbols
andrepresentations withinthe contextof the problemsthey were [Link]-
dents arrivedat a numericalanswer,Mr. Hernandezwould ask such questionsas,
"Canyou explain what thatnumberrefers to?"
This scenarioillustratesthevarietyof classroom,task,teacher,andstudentfactors
thatsupportedstudents'thinkingat the level of doing mathematicsduringthe task.
As the taskunfolded,Mr. Hernandezwas able to orchestratesuccessfullythe com-
plex arrayof factorsneededto maintainthehigh-levelcognitivedemandsof thetask.
Decline toproceduralthinkingwithoutconnectionto [Link]
taskwas embeddedin a seriesof lessonsthatfocusedon [Link]
hadbeenintroducedto P61ya's four-stepproblem-solvingprocessandhadworkedon
at leastone othernonroutineproblembeforeencounteringthe following:
ForMother'sDay, Davie, my littlebrother,Kathy,my youngersister,andI all
contributedmoney to buy a presentfor Mom. Davie had saved 80 pennies, 2
nickels,and 1 [Link] me 3 half-dollarsthatshe hadsaved,andI con-
tributedthe rest. Actually, with what Davie and Kathygave me, the 17 coins
in my bank were just enough to make up the total cost of $8.12. What coins
were in my bank?(Meyer & Sallee, 1983, p. 335)
Studentsbegan workingon this taskin small groupsnearthe end of one lesson.
There appearedto be a clear expectationthat the studentswould work collabora-
tively to solve the [Link] the teacher,Ms. Capra,set up the task, she pro-
videdeachgroupwithone copy of theproblemandone copy of a formlistingP6lya's
[Link] referto P61ya'sstepsas they
solved the problem and to document their work accordingly. The task set-up
encouragedcognitiveprocessesthatareconsistentwith [Link]
the task demandedcomplex and sustainedmathematicalthinkingand reasoning,
because it could not be solved with well-rehearsed,easily accessible formulasor
[Link] the P61yaproblem-solvingsteps also may have encour-
aged self-monitoringandregulationof students'own thoughtprocesses,a hallmark
of high-level thinking.
Duringtaskimplementation, students'cognitiveprocessesdeclinedintoprocedural
thinkingthatmadelittle if any connectionto understandingor [Link]'

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
542 Classroom-BasedFactors

failureto engagein high-levelcognitiveprocesseswas influencedby a varietyof fac-

tors,theforemostbeingtheremovalof challengingaspectsof [Link],
theteacher'sorchestration of students'engagementwiththetaskin a mannerthatstrictly
followedP6lya'sfour-stepprocesschanneledtheirthinkingintopredetermined path-
waysattheexpenseof theirgrapplingwiththemathematical ideasandprocessesembed-
ded in the task. Throughouttask implementation,Ms. Capraregulatedstudents'
thinkingprocessesby partitioning thetaskintostepsthatmatchedeachstepof theP61lya
process. First,students were instructed to workon the firstP61lyastep(understanding
the problem)in theirsmallgroupsandthento discussit in a whole-classdiscussion.
Thentheyweretoldto do the samewiththe secondP61lya step(devisinga plan).It was
only aftercompleting these two
initial steps thatstudentsbeganto workon solvingthe
problem within theirsmall groups. anygivenpoint time,students'workwas con-
At in
strainedby the productsassociatedwith a particularstepin the P6lyaprocess.
Withineach of these steps, therewas evidence that studentswere dealing with
P6lya's ideas at a superficiallevel. Forexample,while Ms. Capraandher students
discussedStep 1, understandingthe problem,they focused on listingthe mainfacts
of the problem(e.g., Davie contributed80 pennies, 2 nickels, and 1 dime). They
did not advanceto a discussionof the mainmathematicalfeaturesof the problem,
the structureof the relationshipsamongthose features,or the goal of the problem.
Similarly,duringdiscussion of Step 2, a label of a strategywas stated(guess and
check), but no discussion took place regardingwhat the strategyentailed,why it
was an appropriateapproachfor the presentsituation,or how to use the strategy.
The declineinto mechanicalformsof thinkingwas also influencedby a focus on
[Link] this case, the answerthatwasjudged as corrector incorrectwas
notthemathematical solutionto theproblem,butratherthestudents'responsesto P61ya' s
four-stepprocess. The "correctness" of students'responses to eachof thesestepsbecame
the focus, ratherthanthe validityof theirmathematicalapproachto the problem.
Finally,the decline of students'engagementwith high-levelcognitive processes
was also influencedby lack of [Link] very quickly,pro-
gressingfrom one step to the next with very little time spenton any one aspectof
the problem-solvingprocess. At each step, the students'small-groupdiscussions
were haltedprematurelyas specific studentswere askedto displaytheirresponses
to the entireclass. Studentshad only 10 minutesto work on solving the Mother's
Day problem(Step 3) in their small [Link] Ms. Caprasaid it was time to
stop, many groupsappearedto need and to desire more time.
Overall,students'work on this potentiallyrich task did not progressinto mean-
ingful engagementwith the mathematicalcontentandprocesses embeddedin the
[Link] fragmentingandchannelingstudents'thinking,the use of P6lya's steps
(alongwiththerushednatureof theclass)in thiscase,led to a mechanical,as opposed
to substantiveand creative,engagementwith the problem.
Declineto unsystematic Theoverallgoalof thissequenceof lessonswas
exploration.
to havestudentsexploremetricareameasurement, as well as relationships betweenlin-
earandareameasurement. Studentshadpreviouslyexperienced a seriesof lessonsinvolv-
inglinearandareameasurement unitsandwerejustbeginningto explore
withnonstandard

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
MarjorieHenningsenand Mary Kay Stein 543

[Link] focus of the previouslessons hadbeen on the relationships

amongdifferentunitsof measurein the contextof linearversus areameasurement.
In thetasksetup,eachgroupof threeorfourstudentsreceiveda teacher-made activ-
ity sheet6explainingthattheirfirsttaskwas to builda [Link] were also
to constructthe squaremeteras efficientlyas possibleusinganyof thefollowingmate-
rials(butno scissors):paper,tape,rulers,andbase-tenpieces to [Link]-
structing thesquaremeter,studentswereto modela squaredecimeter, a squarecentimeter,
and a squaremillimeterin one cornerof theirsquare-metermodel (placingsmaller
squaresinsidethe largersquares).By constructingthe squaremeterthemselves,stu-
dents could gain a good sense of the size of the squaremeter and how its size
relatedto [Link] addition,studentscoulduse theirmodelsto explore
otherrectangleswiththe samearea(by cuttingup andrearranging theirsquare-meter
models). At set-up,the task encouragedstudents'engagementin doing mathemat-
ics. In executingthe task,studentswererequiredto rely on knowledgegainedfrom
theirpriormeasurementexperiencesandto decide whatunitsof measureandmea-
suringtools to use, given the limitationsset by theirteacher,Ms. [Link], stu-
dentshadto makeconjectures abouthowtobuildthesquaremeterefficiently andaccurately
andhow to show the relationshipsamongthe different-sizedsquares.
As implementationof the taskunfolded,it becameclearthateven thoughthe stu-
dentswere on-task,the cognitivedemandsof the taskwere not [Link] evi-
denced by theirconstructionmethods,many groupsappearedto be viewing their
squaressolely in termsof lineardimensions,withno focuson [Link]
built their models by first constructingfour strings of paper each of length 1
meter, then forminga square,and finally filling in the empty space in the middle
(essentiallybuildingthe entireperimeterfirst).Also, the students'lack of engage-
mentin the intendedhigh-levelprocesseswas evidentin the way manygroupsinter-
pretedthe instructionsfor modelingthe [Link] have
missedthepoint,drawingthe smallersquaresas isolatedmodels,one insidetheother.
Ms. Hoffmanintendedfor studentsto drawthe smallersquaresso thateach square
sharedboth areaand partiallineardimensionswith the squareslargerand smaller
thanitself (see Figure5 for an illustration),makingvisually apparentthe relation-
ships among the differentmetriclinearand areameasures.
One reason studentsfailed to engage with the intendedprocesses was the inap-
propriatenessof the task with respectto the clarityand specificityof task expecta-
tions, as evidencedby the largeamountof time most groupsspentcompletingtheir
constructionsandby the factthatstudentsseemedto needmoreguidanceaboutwhat
was expectedin termsof how to situatethe smallersquareswithintheirsquaremeter
[Link]. Hoffmancouldhaveinstructedstudentsto placethe smallersquareareas
withinthe largerareaso thatthe relationshipsamongthe linearand areameasures
would have been [Link] instructionwould have provideda little moredirec-
tion,while maintainingthe cognitivedemandsof the taskat [Link] additional

6Theteacheradaptedandexpandedsuggestionsin TeacherActivity109Bfoundin BennettandForeman

(1990).

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
544 Classroom-BasedFactors

reason, related to the inappropriatenessof the task for this group of students,
appearedto be theirlackof priorknowledgeneededto attacktheproblemefficiently.
For example, many studentslackedknowledge aboutrelationshipsamong differ-
ent-sized lineardimensionsand especially abouthow lineardimensionscould be
relatedto area.

An arrangementthat illustrateshow
the lengthsand areas are related

The type of arrangementconstructed

by manyof the students

Figure 5. "Builda squaremeter"task (Note. Figuresare not drawnto scale)

Anotherfactorthatinfluencedthe declinein cognitivedemandswas thatstudents

mayhavebeengiventoo muchtimeto [Link]
the entireperiodto completethe [Link] therehadbeen a tightertime limit, some
groupsmay have been motivatedto [Link], since manyof the
groupsappearedto be flounderingwith this partof the task,it may have been help-
ful for the teacherto intervenewith these groupsearly in orderto help them adopt
a more efficient plan of action.
A final factorthatinfluencedthe decline of this task was thatsome of the math-
ematicallychallengingaspectsof the taskwere removedor becameovershadowed
in the lengthyprocessof buildingthe [Link] groupsthatreachedthe part
of the taskin which they were supposedto drawin the different-sizedmetricareas
did not seem to be frustratedby it. The ways in which manygroupsdid this (shown
in Figure5), however,indicatedthattheydidnotrealizethe significanceof theplace-
mentof these [Link],this partof the taskbecamea nonproblemfor the stu-
dents,simply a matterof drawingsome different-sizedsquaresinside one another.
Despite the fact thatthe majorityof studentswere activelyengagedin the lesson

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
MarjorieHenningsenand Mary Kay Stein 545

overall, they failed to focus on the importantmathematicalideas containedwithin

this complex and interesting task. Students earnestly tried to engage in doing
mathematics,andthe teacherearnestlytriedto supportthe maintenanceof the task
at a high level;however,for the majorityof students,theseeffortswereunsuccessful.
The difficultiesdidnot lie in classroommanagementproblems,andtheteachernever
shifted the focus to one correct way of doing the task. Instead, the cognitive
demandsdeclined primarilybecause of several factorsrelatedto the appropriate-
ness of the task for the studentsand the level and kind of guidancethey neededto
engage at the level of doing mathematics.
Declineto no [Link] of thissequenceof thelessons
was to encouragestudentsto use patternfinding to discover propertiesof two-
dimensionalgeometricfiguresratherthanto [Link] partic-
ulartaskfocusedon anglesandhow theycanbe usedto definedifferenttypesof triangles.
In the task setup,Mr. Kingsley,the teacher,gave each pairof studentsa tangram
puzzle consisting of five triangles,one square,and one nonrectangularparallelo-
gramarrangedin such a way thatthe shapescoveredthe entireareaof a [Link]
studentswere also given an activitysheet askingthemto systematicallyexplorethe
similaritiesand differencesin the angles of the pieces. Studentswere expected to
(a) identifythe two pieces thatwere the same as (congruentwith) otherpieces and
eliminatethem, (b) recordthe type of each angle in the remainingfive pieces, and
(c) examine and recordsimilaritiesand differencesin the angles across those five
pieces. This systematicexplorationwas ultimatelyto resultin students'recogni-
tion thatthe triangulartangrampieces wererightisosceles trianglesandthatthe tri-
angles'acuteangleswerecongruentwiththe acuteanglesin the parallelogram piece.
As set up, the task encouragedstudentsto engage in cognitive processes that are
consistentwithdoingmathematics,suchas discoveringimportantmathematical ideas
through hands-on of and a
exploration patterns,establishing implementing systematic
methodof recordingthe resultsof one's explorations,andmakingobservationsof
similarities and differences across the angles of different geometric figures to
deepen understandingsof connectionsamong variousgeometricshapes.
Studentsdidnotmakemuch,if any,progresson [Link]
with the task when Mr. Kingsley was at theirtable assistingthem andaskingthem
[Link] restof the time,they were only half-interested,satidle, played
with the tangrampieces in nonmathematical ways, or talkedto otherstudentsin the
class aboutnonmathematicaltopics.
The studentsfailed to engage in the intendedhigh-level cognitive processes for
a varietyof [Link] foremostreasonwas the inappropriateness of the taskwith
respectto the clarityandspecificityof the [Link]
not specific enough to guide studentstowarddiscoveringthe relevantmathemat-
ical [Link] lack of specificity was especiallyproblematicbecausethe stu-
dents lacked the relevantpriorknowledge needed to make effective comparisons
and [Link] example, students'inabilityto distinguishacute, obtuse,
and right angles hindered their efforts to record systematically and generalize
[Link] a result,most studentsplayedaroundwith a few comparisonsbut

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
546 Classroom-BasedFactors

failed to makeprogresssystematicallyin discoveringthe ways in which all the tri-

angles were similar.
Anotherfactorthatled to a declinein the cognitive demandson the studentswas
[Link] the beginningof the implementationphase,
some of the studentsneeded pencils, othersneeded recordingsheets to recordthe
informationthey discovered,and still otherswere workingor playingwith the tan-
grampieces. Throughoutthe task, a significantsubsetof studentswanderedfreely
aroundthe room, visiting with [Link] studentsappearedto be engag-
ing in off-taskbehaviorsbecause they were unsurehow to proceedwith the task.
The final factorthatcontributedto the decline in the level of students'engage-
ment with this task duringthe implementationphase was the amountof time that
studentswere allowed to [Link] the fact thatlittle progresswas being
made,studentswereallowedto continueworkon thistaskfor38 [Link] amount
of off-taskbehaviorincreasedsteadilyduringthis time as studentsappearedto reach
the conclusion thatthey could not work effectively on the task.
Overall,the decline into lower levels of studentengagementwith this potentially
richtaskwas representativeof the ways in which otherhigh-leveltasksin ourdata-
base declinedintono [Link] this particularcase, classroomman-
agementandtimingproblemsappearedto be closely intertwinedwith the problem
of mismatchbetweenthe cognitive demandsof the taskand students'priorknowl-
edge. Althoughboththe lack of attentionto appropriate tasksandthe recordingsys-
tems thatcontributedto the managementproblemsappearedfrom the outsetof the
task, it was difficult to discernwhich of these problemscame first.

SUMMARY AND CONCLUSIONS

At the outsetof the study,we soughtto addressthreeareas:(a) to providea pro-
file of factors associatedwith tasks that were set up to engage studentsin cogni-
tiveprocessesatthelevelof doingmathematics andthatdidengagestudentsin high-level
thinking and reasoning, (b) to describefactorsinfluencingthe threecharacteristic
patternsof declinein students'engagementwithhigh-levelcognitiveprocesses,and
(c) to providedetailedqualitativeportraitsfrom our databaseto illustrateall four
patternsand the profiles of factorsassociatedwith them.
Ourfindingssuggestthattherewas a discernibleset of factorsinfluentialin assist-
ing studentsto engageat high levels. These includedfactorsrelatedto the appropri-
atenessof thetaskforthestudents andto supportive suchas scaffolding
actionsby teachers,
andconsistentlypressingstudentsto providemeaningfulexplanationsormakemean-
[Link] fortheroleof theteacherin reform
classrooms,in whichstudentsareexpectedto be activelyengagedin doingmathematics.
Notonlymusttheteacherselectandappropriately setupworthwhilemathematical tasks,
buttheteachermustalsoproactivelyandconsistentlysupportstudents'cognitiveactiv-
ity withoutreducingthe complexityandcognitivedemandsof thetask.
Students' engagement with tasks that declined to lower levels of cognitive
activity happenedin different ways and for different reasons. For each of the
threepatternsof decline, we were able to identifya set of predominantclassroom-
basedfactorsthatcontributedto the decline in the cognitive demandsof the tasks;

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
MarjorieHenningsenand Mary Kay Stein 547

however,therewas variationin how sharplydistinguishablethe factorprofileswere

across decline patterns.
The least readilyidentifiablefactorprofilewas associatedwith decline to unsys-
tematicexploration,a patternthatwas not anticipatedin ourinitialcoding system.
In this pattern,the studentsearnestlyattemptedto remainfaithful to the setup of
the task at the level of doing mathematicsand teachersmade attemptsto support
high-level engagement,but studentswere ultimatelyunsuccessfulwith respectto
performingat a high level and engaging with the importantmathematicalideas in
the [Link] we pointedout earlier,the lack of a crispfactorprofileassociatedwith
this patternmay be due to inadequaciesin our [Link] the other
two patternsthatreflecteda declinefromdoing mathematicsto some otherspecific
level of students'engagementwith cognitive processes,this patterndeclinesfrom
doing mathematics,but does not reflect a decline to a readilyidentifiablelevel of
students'[Link] speculatethatas teachersandstudentsbecomemore
confidentandmore willing to takerisks with the kindsof tasksthataim to engage
studentsin doingmathematics,thistypeof declinepatternmightbecomemorepreva-
lentthanothertypesof declinein students'[Link] an empir-
ical questionthatmay warrantfurtherinvestigation.
Across all threepatternstherewas one factor,the appropriateness of the amount
of time (eithertoo little or too much) allottedfor the task, that appearedas a pre-
dominantinfluence;however, this factorappearedto functiondifferentlyin each
of the threedecline [Link] agreementwith researchon students'engagement
with academictasks,these findings suggest thatplanningfor appropriateamounts
of time and flexibility with timing decisions may play an importantrole in avoid-
ing decreasesin the level of cognitiveactivityengagedin by studentsas tasksunfold
in the classroom(Doyle, 1986). Two otherfactors,removalof challengingaspects
of the task and inappropriatenessof the task for a varietyof reasons(e.g., lack of
interest,motivation,knowledge, or uncleartask expectations),were each judged
to be predominantinfluences in two of the threedecline [Link] prevalence
of these factorsas influencesin students'decliningcognitiveactivityhas also been
found by otherresearchers(Bennett& Desforges, 1988; Doyle, 1983, 1986).
The resultsenabledus to examine also factorsthat were notjudged to be influ-
entialin thethreecharacteristic patternsof declinein students'[Link] found
thatclassroommanagementproblemswere a predominantinfluencein only one of
the threecharacteristicpatterns:decline to no mathematicalactivity. In the other
two patternscombined,classroommanagementwas an influentialfactorin thedecline
of only one task,a findingthatseems to be at odds with the moregeneralliterature
on academictasks;this literaturesuggests thatclassroom managementproblems
oftenaffecttheimplementation of high-leveltasks(Doyle, 1988).Futureresearchcould
investigate more deeply the variationsin how classroommanagementdid or did not
influencethe differentpatternsof decline in the tasksavailablein ourdatabase.
The use of realclassroom-basedscenariosto illustrateempiricallygeneratedpat-
ternsof students'engagementcan be seen as similarto the use of cases to illumi-
nate [Link] generally,within the field of researchon teaching,
Shulman(1986) has discussed the need for the developmentof specific cases that

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
548 Classroom-BasedFactors

areconnectedto moregeneralprinciplesof [Link] Shulman,such

cases are useful because they connect to the complex world of everydaypractice
and also to a largerset of ideas [Link] largerideas have the dis-
advantageof being abstract,but at the same time, they have the advantageof being
more [Link] cases are selected or developed to illustrateprinciples
orideas,theresultant producthastheadvantageof beingnotonlyaninteresting account
of practice,but also a case of a [Link] being relatedto largersets
of ideas, the cases become more meaningful,more "connectable"to otherimpor-
tantideas, and morepowerful as a guide for futureresearchor practice.
In this study,the patternsof studentengagementandprofiles of influentialfac-
tors, althoughnot principles,did provide a coherentframeworkwithin which the
qualitativeportraitscouldbe [Link]'engagementpatternsandthe fac-
torprofileshadcertainqualitiesof abstractionandgeneralizabilitybecausetheyhad
been suggestedby patternsof findingsfroman [Link]
portraitsareplacedinto this largerconceptualspace, theirmeaningsbecome more
[Link] teachersandstudentscanbe interpreted within
a frame of reference that includes more general notions of pedagogy and the
developmentof mathematicalunderstandingsby students.
REFERENCES
Anderson,L. M. (1989). [Link] M. C. Reynolds(Ed.),Knowledgebasefor the begin-
ningteacher(pp. 101-115).Washington,DC:AmericanAssociationof CollegesforTeacherEducation.
Bennett, A., & Foreman,L. (1990). Visual mathematicscourse guide (Vol. 2). Portland,OR: Math
LearningCenter.
Bennett,N., & Desforges, C. (1988). Matchingclassroomtasks to students'[Link]
School Journal,88, 221-234.
Burton, L. (1984). Mathematical thinking: The struggle for meaning. Journal for Research in
MathematicsEducation,15, 35-49.
Doyle, W. (1983). Academic [Link] of EducationalResearch,53(2), 159-199.
Doyle,W. (1986).Classroomorganization andmanagement. In M. C. Wittrock(Ed.),Handbookof research
on teaching (3rd ed.; pp. 392-431). New York:Macmillan.
Doyle, W. (1988). Workin mathematicsclasses: The context of students'thinkingduringinstruction.
EducationalPsychologist, 23, 167-180.
Grouws, D. A. (Ed.). (1992). Handbook of research on mathematics teaching and learning. New
York: Macmillan.
Hiebert,J., & Carpenter,T. P. (1992). Learningandteachingwith understanding. In D. A. Grouws(Ed.),
Handbookof researchon mathematicsteachingand learning (pp. 65-97). New York:Macmillan.
Hiebert,J., & Wearne,D. (1993).Instructionaltasks,classroomdiscourse,andstudents'learningin second-
[Link],30, 393-425.
Lindquist,M. M., & Kouba,V. L. (1989).Measurement. In M. M. Lindquist(Ed.),Resultsfrom thefourth
mathematicsassessmentof the NationalAssessmentof EducationalProgress (pp. 35-43). Reston,
VA: NationalCouncil of Teachersof Mathematics.
Marx,R. W., & Walsh,J. (1988).[Link],88, 207-219.
Meyer, C., & Sallee, T. (1983). Make it [Link] Park,CA: Addison-Wesley.
National Council of Teachers of Mathematics. (1989). Curriculumand evaluation standardsfor
school [Link], VA: Author.
NationalCouncilof Teachersof Mathematics.(1991). Professionalstandardsfor teachingmathematics.
Reston, VA: Author.
NationalCouncil of Teachersof Mathematics.(1995). Assessmentstandardsfor school mathematics.
Reston, VA: Author.

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions
Marjorie Henningsen and Mary Kay Stein 549

NationalResearchCouncil. (1989). Everybodycounts. Washington,DC: NationalAcademy Press.

QUASARDocumentationTeam.(1993, March).TheQUASARteacherbackgroundquestionnairesum-
mary report on selected items. Pittsburgh, PA: Learning Research and Development Center,
Universityof Pittsburgh.
Resnick, L. B. (1987). Educationand learning to [Link],DC: NationalAcademy Press.
Romberg,T. A. (1992).Perspectiveson scholarship [Link]. A. Grouws(Ed.),Handbook
of research on mathematicsteaching and learning (pp. 49-64). New York:Macmillan.
Romberg,T. A. (1994). Classroominstructionthatfostersmathematicalthinkingandproblemsolving:
[Link] A. H. Schoenfeld(Ed.),Mathematicalthinkingandprob-
lem solving (pp. 287-304). Hillsdale, NJ: Erlbaum.
Schoenfeld,A. H. (1983). Beyond the purelycognitive:Belief systems,social cognitions,andmetacog-
nitions as drivingforces in [Link], 7, 329-363.
Schoenfeld,A. H. (1992). Learningto thinkmathematically:Problemsolving, metacognition,andsense
makingin [Link] D. A. Grouws(Ed.),Handbookof researchon mathematicsteachingand
learning (pp. 334-371). New York:Macmillan.
Schoenfeld,A. H. (1994). Reflections on doing and [Link] A. H. Schoenfeld(Ed.),
Mathematicalthinkingand problemsolving (pp. 53-70). Hillsdale, NJ: Erlbaum.
Shulman,L. S. (1986). Those who understand:Knowledgegrowthin [Link],
15(2), 4-14.
Silver, E. A. (Ed.). (1985). Teachingand learning mathematicalproblem solving: Multipleresearch
perspectives. Hillsdale, NJ: Erlbaum.
Silver, E. A., Branca,N., & Adams, V. (1980). Metacognition:The missing link in problemsolving?
InR. Karplus(Ed.),Proceedingsof theFourthInternational for thePsychologyof Mathematics
Conference
Education(pp. 213-221). Berkeley, CA: Universityof California.
Silver,E. A., & Stein,M. K. (1996).TheQUASARproject:The"revolution of thepossible"in mathematics
instructionalreformin urbanmiddleschools. UrbanEducation,30, 476-522.
Stein,M. K., Grover,B. W., & Henningsen,M. (1996). Buildingstudentcapacityfor mathematicalthink-
ing and reasoning: An analysis of mathematical tasks used in reform classrooms. American
EducationalResearchJournal, 33, 455-488.

Authors

Marjorie Henningsen, ResearchSpecialist,LearningResearchand DevelopmentCenter,University

of Pittsburgh,3939 O'HaraStreet,#711, Pittsburgh,PA 15260; marjh@[Link]
Mary Kay Stein,ResearchAssociate,LearningResearchandDevelopmentCenter,Universityof Pittsburgh,
3939 O'HaraStreet,#828, Pittsburgh,PA 15260; mkstein@[Link]

This content downloaded from [Link] on Fri, 26 Sep 2014 14:47:54 PM

All use subject to JSTOR Terms and Conditions