ERIC ED567223: Structure Building Predicts Grades in College Psychology and Biology

2016·0.43 MB·English

by  ERIC

AppliedCognitivePsychology,Appl.Cognit.Psychol.30:454–459(2016)
Publishedonline7April2016inWileyOnlineLibrary(wileyonlinelibrary.com)DOI:10.1002/acp.3226
Structure Building Predicts Grades in College Psychology and Biology
KATHLEENM.ARNOLD1*,DAVIDB.DANIEL2,JAMIEL.JENSEN3,MARKA.MCDANIEL4and
ELIZABETH J. MARSH1
1DepartmentofPsychologyandNeuroscience,DukeUniversity,Durham,NC,USA
2DepartmentofPsychology,JamesMadisonUniversity,Harrisonburg,VA,USA
3DepartmentofBiology,BrighamYoungUniversity,Provo,UT,USA
4DepartmentofPsychologicalandBrainSciences,WashingtonUniversityinSt.Louis,St.Louis,MO,USA
Summary: Knowingwhatskillsunderliecollegesuccesscanallowstudents,teachers,anduniversitiestoidentifyandtohelp
at-riskstudents.Oneskillthatmayunderliesuccessacrossavarietyofsubjectareasisstructurebuilding,theabilitytocreatemental
representationsofnarratives(Gernsbacher,Varner,&Faust,1990).Wetestedifindividualdifferencesinstructure-buildingability
predictedsuccessintwocollegeclasses:introductorytopsychologyandintroductorybiology.Inbothcases,structurebuilding
predictedsuccess.Thiseffectwasrobust,withstructurebuildingexplainingvarianceincoursegradesevenafteraccounting
forhighschool GPA and SAT scores (in thepsychologycourse) ora measureof domain knowledge (inthe biology course).
Theresultssupporttheclaimthatstructurebuildingisanimportantindividualdifference,onethatisassociatedwithlearning
indifferentdomains.Copyright©2016JohnWiley&Sons,Ltd.
Success in college requires learning, retaining, and applying Anassumptionofthestructurebuildingframeworkisthat
largeamountsofinformation.Farfromstraightmemorization, narratives have a structure and that comprehensioninvolves
students must think critically and often need to go beyond buildingacoherentmentalmodelthatcapturesthisintended
given information, such as when they design experiments or structure.Thisframeworkarguestheprocessofbuildingthis
developarguments.Successlikelydependsuponmanyskills, model is similar whether the narrative is presented visually,
includingreadingcomprehension,writingproficiency,reason- auditorily,orpictorially.Evidenceforthisgeneralcomprehen-
ingability,andtheabilitytoworkwithnumbers,amongother sion ability stems from Gernsbacher’s Multimedia Compre-
things. Here we focus on one specific skill that may support hension Battery (MMCB; Gernsbacher & Varner, 1988).
academic success: structure-building, the ability to build a Initsentirety,thisbatteryconsistsofmultiplestories,each
“cohesive,mentalrepresentationor‘structure”’ofanyevent, presented in one of three modalities: written, auditory, or
whether read or encountered via some other modality pictorially, with comprehension questions following each
(Gernsbacher,Varner,&Faust,1990,p.431). story. Comprehension scores from the different modalities
To understand the concept of structure-building and how are highly correlated (rs range from .72 to .92), and a
it supports learning, consider the following text from a principle-component analysis showed that one factor, a
biological psychology textbook: general comprehension ability, underlies comprehension in all
modalities(Gernsbacheretal.,1990).Gernsbacherandhercol-
Working memory enables an individual to respond to a leaguesarguethatthisgeneralabilityisrelatedtohowwellone
stimulus that was heard orseena short while earlier,as cansuppressinformationirrelevanttoone’semergingrepresen-
tested in various forms of the delayed response task. tational structure. A deficit in suppression causes low
For example, an animal might see a light shine above comprehenders to frequently shift to new structures because
oneofseveral doors.Whenthelightgoesoff,theanimal theirrelevantinformationdoesnotmapontotheexistingstruc-
waitsthroughaspecifieddelay,andthenhastogotothe ture.Frequentshiftingcausesadisjointednarrativerepresenta-
door where it saw the light (Kalat, 1998). tion,therebyreducingcomprehension.
This framework is normally discussed in terms of
The less-skilled comprehender might represent the comprehending a single narrative from one source, but can
concepts of working memory and the delayed response beappliedtosituationswhereonemustintegratematerialfrom
task separately, failing to integrate the two concepts in a multiplesources.Considerthestudentlearningaboutworking
unified structure. In contrast, the skilled comprehender will memoryandthedelayedresponsetask;thestudentmighthear
map the new information about the delayed learning task alecturethatreliesontheoperation-spantaskasthemeasure
onto their existing representation of working memory, of working memory, and then later read a textbook passage
creating a unified structure. These representations have whereworkingmemoryismeasuredviathedelayedresponse
implications for understandingthe text; frequent shifts from task.Thelearnerdoesnotnecessarilywanttocreateseparate
existing to new structures reduce comprehension and may representations for these two sources; rather, the goal should
therefore reduce academic success across a variety of be to extract an integrated representation (structure) across
subject areas. sources.Inthissense,studentslikelyfaceamorechallenging
task than does the typical laboratory subject; in classes, stu-
dents often build mental models over time and across
sources,asopposedtolaboratoryexperimentsthattypically
*Correspondence to: Kathleen M. Arnold, Department of Psychology &
require building a representation of a single text. Frequent
Neuroscience,417ChapelDrive,DukeUniversity,Durham,NC27708,USA.
E-mail:[email protected] shifting may be more likely in classroom situations, given
Copyright ©2016 John Wiley & Sons, Ltd.
Structure building predicts grades in college 455
that learningis oftenspreadovertime andmayinvolve in- althoughminimal,isencouragingregardingourhypothesis
tegrationacrosslectures,textbooks,slides,discussions,prob- that structure-building ability might be related to learning
lemsets,graphs,andvideos.Theresultingdisjointedmodels in authentic academic contexts.
would then hinder students’ abilities to make inferences and Toextendtheexistingwork,weexaminedarangeofpre-
connectinformationpresented through multiple sources. In dictors of success in addition to structure building ability in
other words, the classroom may be a learningenvironment twoverydifferentcollegecourses:PsychologyandBiology.
for which structure-building skills are of great importance. At two different institutions, instructors1 administered the
Forthesereasons,webelieveitiscrucialtoinvestigatethere- MMCBtotheirclasses,andwelookedattherelationshipbe-
lationship between structure-building and classroom tweenMMCBandcoursegrades.Inbothcourses,additional
performance. measureswerecollectedthatallowedustoevaluatetherela-
In contrast to our approach, the literature is relatively tionship between MMCB and grades after controlling for
silentastowhichspecificskillspredictcollegesuccess(with morestandardpredictors.InthePsychologycourse,asubset
theexceptionsofworkingmemory andreadingcomprehen- of students provided their high school GPAs and standard-
sion,asdiscussedattheendofthispaper).Instead,thefocus izedtestscores,two‘goldstandard’measuresofpastperfor-
hasbeen onmeasures ofpast successes. Thebest predictors mance.IntheBiologycourse,asubsetofstudentscompleted
of college success include high school GPA (Cohn, Cohn, two measures of biology/science knowledge and skills, the
Balch,&Bradley,2004),highschoolclassrank(e.g.Baron Biology Concept Inventory (BCI) and the Lawson’s Class-
& Norman, 1992), and scores on standardized tests such as roomTestforScientificReasoning(LCTSR);bothmeasures
the ACT and the SAT (Hannon, 2014). Such measures are focusedonscience.Weaskedthreequestions:(i)Whatisthe
sufficient if the goal is prediction, but they provide little relationship between structure-building (as measured by
insight into the specific skills or knowledge to target for MMCB) and the typical predictors of academic success?
improvement. More targeted predictors focus on what To the extent that these measures are correlated, it would
knowledgeandskills a studentneedstosucceed inapartic- suggest that structure-building ability may be one of the
ular subject matter, such as the Force Concept Inventory skills underlying performance on those measures. (ii) Does
(FCI). This inventory measures student understanding of structure-building predict success in college, and is it simi-
forceandrelatedkinematics,andcorrelateswithperformance larlypredictiveinaSocialScience(Psychology)andNatural
in introductory physics classes at both the high school and Science Course (Biology)? (iii) Does structure building
university levels (Hestenes, Wells, & Swackhamer, 1992; predict success over and above the traditional predictors,
Savinainen & Scott, 2002; see also Cahill et al., 2016, ex- includingmeasuresofpastperformance(inthePsychology
tendingthesefindingstosecondsemesterintroductoryuni- course) and knowledge/skills specific to the class topic (in
versity courses, using the Brief Electricity and Magnetism the Biology course)?
Assessment[BEMA]asapredictorofcourseperformance).
However, while such measures can be extremely useful
within specific classes, they are limited in their ability to METHODS
generalize across subject areas.
Participants
There is initial evidence supporting the possibility that
structure-building is associated with success in college, and Psychology
thatitspredictivenesswillnotbelimitedtoaparticularsubject Nine hundred fifty-four undergraduate James Madison
matter.Severallaboratorystudieshave examinedtherelation Universitystudents(594females)enrolledinIntroduction
between structure building (as measured by MMCB scores) to Psychology participated in the study. Most students
and learning of classroom-like materials. Structure-building were between 18 and 21 years old (n=935), were in their
abilityissignificantlyrelatedtolearning(testperformance) first(n=614)orsecond(n=243)yearofcollege,andwere
afterreadingapsychologytextbookselection(Callender& native English speakers (n=925). SAT scores and high
McDaniel,2007)orlisteningtoalectureabouthowamechan- school GPAs for 148 1st semester students were obtained
ical device works (Bui & McDaniel, 2015). It is associated from university records.
with the accuracy of confidence judgments, with better
comprehenders more accurately assessing their relative test
Biology
performance(Maki,Jonas,&Kallod,1994),andtheefficiency
OnehundredfourundergraduateBrighamYoungUniversity
with which study time is regulated (Martin, Nguyen, &
students enrolled in an introduction to biology course for
McDaniel, 2015). In addition, one previous study examined
non-majors participated in the study. Students were in their
whetherstructure-buildingpredictsacademicsuccessinanac-
first (n=40), second (n=39), third (n=14), or fourth
tualclassroom(Maki&Maki,2002).Thisworkwasfocused
(n=10) year in college. No other demographics were col-
on whether course format (in-person vs. web-based lectures)
lected. In addition to the MMCB, 83 students completed
affected learning; the MMCB was used to examine whether
two additional individual difference measures: the LCTSR
the conclusions depended upon the comprehension skills of
and the BCI.2
thelearner.MMCBsignificantlypredictedexamperformances
for students in both the web course and the traditional class, 1Theinstructorswerethesecond(psychology)andthird(biology)authors.
andsignificantlypredictedperformanceinthewebcourseon 2Fifty-six biology students released their combined SAT and/or ACT
scores.Becauseofthisrelativelysmallsamplesizeformeasuringindividual
a post-course measure consisting of practice items from the
differencesandthelackofsubscores,standardizedtestscoreswerenotan-
Psychology GRE exam. Thus, the available evidence, alyzedinthiscourse.
Copyright©2016JohnWiley&Sons,Ltd. Appl.Cognit.Psychol.30:454–459(2016)
456 K. M. Arnold et al.
Materials consistency was low (Cronbach’s alpha=.38), possibly be-
causeofthemultidimensionalnatureoftheinventory.
MMCB
Following previous studies (Bui & McDaniel, 2015;
Callender & McDaniel, 2007; Callender & McDaniel, Design and procedure
2009;Martinet al.,2015), weusedonlythe written version
Psychology
of the MMCB, which is sufficient for measuring a general
StudentswereenrolledinoneoftwoIntroductoryPsychology
structurebuildingability.Inbothcourses,fourpassagesfrom
classeswithbothclassestaughtbythesameinstructor(the2nd
the Multi-Media Comprehension Battery (MMCB; author). On the first day of class, students completed the
Gernsbacher&Varner,1988)wereusedtomeasureindividual
MMCB. Final course grades were based on performance
differencesinstructurebuilding.Thefourpassagesranged in
on 12 quizzes and 4 exams, which were spaced out across
lengthfrom538to957words,andeachhad12correspond-
the semester. Quizzes were given at the beginning of class
ing multiple-choice questions about key details from the
and covered the previous reading assignment. Exams were
passage, for a total of 48 questions. The questions corre-
not cumulative and covered material from both lecture
sponding to each passage were answered immediately fol-
andreading.Questionsonboththequizzesandexamswere
lowing the reading of that passage, without access to the
multiple-choice and scored by scantron.
passage. Scores could range from 0 to 48.
The biology sample had good reliability (Cronbach’s
alpha=.85), but reliability could not be calculated in the Biology
psychologysamplebecausescoresonindividualquestions Students were enrolled in one of two Introductory Biology
were not available. However, prior studies have consistently fornon-majorsclasses;bothclassesweretaughtbythesame
shown the MMCB to be reliable (Callender & McDaniel, instructor(thethirdauthor).StudentscompletedtheLCTSR
2009;Gernsbacheretal.,1990). and the BCI within the first two weeks of the semester and
theMMCBattheendofthesemesterinconjunctionwiththe
final exam. Final course grades were based on performance
SAT
on three unit exams, one final cumulative exam, 26 quizzes,
TheSATisastandardized aptitude testusedforcollege ad-
homework assignments, and class participation. Quizzes
missions. The Psychology instructor collected scores on the
consisted of four multiple-choice questions answered using
mathandverbalsubsections,eachofwhichhasamaximum
an interactive response device (clicker). Homework assign-
score of 800 for a combined total maximum of 1600 for
ments involved open-book quizzes on the assigned reading.
those two sections.
Unitexamswereopen-book(i.e.,studentscouldrefertotheir
textbookandnotes)andconsistedof75multiple-choiceques-
High school GPA tions. The final exam was closed-book and consisted of 80
High school GPA was collected through self-report in the
multiple-choicequestions.
Psychologycourse.GPAsthatwerereportedona100-point
scale were converted to a 4-point scale.
RESULTS
Scientific reasoning andbiological content knowledge
measures Table 1 presents average final course grade and MMCB
Two additional measures were collected in the Biology score for both classes. MMCB scores were similar across
course. First, science reasoning ability was measured using the two courses (t<1), were neither at ceiling nor floor,
the revised version of LCTSR (Lawson, 1978; Lawson and had sufficient variability to examine individual differ-
etal.,2000),whichisacontent-independenttestofbasicsci- ences. Table 1 also shows average SAT score and high
entificreasoningability.TheLCTSRconsists of12pairs of school GPA for those students who provided them, and
questions, for a total of 24 multiple-choice questions. The scores on the two science knowledge/ability measures from
first question in a pair requires the student to answer a the subset who took them.
science-reasoning problem and the second asks the student
to explain how the answer was derived. Scores can range Question 1: Is Structure-Building Related to Past
from0to24.TheLCTSRhadgoodreliabilityinoursample Academic Success and/or Domain-Specific Measures?
(Cronbach’salpha=.83). OurhypothesisimpliesthatMMCBscoresshouldbecorrelated
Second,knowledgeofbiologywasassessedwiththeBCI, withmeasuresofpastacademicsuccessinthesamewaytheyare
whichconsistsof30multiple-choicequestions(Klymkowsky, hypothesizedtoberelatedtocurrentacademicsuccess.Toeval-
Underwood,&Garvin-Doxas,2010).TheBCIwasdesigned uatethisidea,weexaminedthecorrelationsinthePsychology
to identify common student misconceptions with questions samplebetweentheMMCBandthetwomeasuresofpastsuc-
groupedbytopicssuchasmolecularpropertiesandfunctions, cess:SATscoresandhighschoolGPA.AsshowninTable2,
genetic behaviors, and evolutionary processes. Scores can higherMMCBscoreswereassociatedwithbetterhighschool
rangefrom0to30.Validityandreliabilityofthismeasureat grades (r=.33 p<.001) and with higher SAT Verbal scores
the individual question level have been previously reported (r=.16,p=.047);asmightbeexpected,asimilarrelationship
(Klymkowsky, Underwood, & Garvin-Doxas, 2010). How- betweenMMCBandSATMathdidnotreachsignificance.In
ever, to our knowledge overall internal consistency has not contrast,MMCBscoreswerenotcorrelatedwitheitherscience
been previously assessed. In our own sample, internal measure(LCTSR,BCI)inthebiologyclass.
Copyright©2016JohnWiley&Sons,Ltd. Appl.Cognit.Psychol.30:454–459(2016)
Structure building predicts grades in college 457
Table1. Meanscoresandstandarddeviations(inparentheses)onthemulti-mediacomprehensionbattery(MMCB),SAT—Verbal,SAT—Math,
HighSchoolGPA,Lawson’sClassroomTestforScientificReasoning(LCTSR),andtheBiologyConceptInventory(BCI)
Course Finalgrade MMCB SAT—Verbal SAT—Math HSGPA LCTSR BCI
Psychology 79.3(8.9) 35.5(5.4) 570.6(66.6) 563.7(73.9) 3.80(.34) — —
Biology 87.7(8.1) 35.5(6.8) — — — 17.5(4.4) 10.6(3.2)
Dashesindicatedatawerenotcollectedorwerenotanalyzed(becauseoflowNforthatmeasure).Possiblescoresforthefollowingmeasuresareasindicated:
MMCB(0–48),LSCTSR(0–24),andBCI(0–30).
Introduction to Psychology and the Introductory Biology
Table2. Correlationsbetweenpredictorvariablesineachcourse courses.Toconfirmtheseeffects,separateregressionanaly-
ses with MMCB as the only predictor were computed for
Course
each course to determine if structure-building scores pre-
Psychology(n=148) Biology(n=83) dicted final course grades. The results mirror the figures;
SAT— SAT— structure building ability significantly predicted final grades
Measure MMCB Verbal Math MMCB LCTSR in both courses (Psychology: β=.22, p<.001; Biology:
β=.29, p=.002).3 That is, one’s ability to form a mental
SAT—Verbal .16* model of a narrative is related to one’s outcomes in college
SAT—Math .08 .21*
classes.
HSGPA .33*** .19* .12
LCTSR .15 Question3:DoesStructure-BuildingExplainAdditional
BCI .06 .27* Variance after Accounting for Traditional Predictors?
First,weexaminedwhether MMCB scores predicted college
MMCB=multi-mediacomprehensionbattery.LSCTSR=Lawson’sClassroom
TestforScientificReasoning.BCI=BiologyConceptInventory. performanceoverandabovethegoldstandardofpastperfor-
*p<.05. manceinhighschool.Weansweredthisquestionwiththepsy-
***p<.001.
chology sample, which included both SAT scores and High
School GPA for 148 first semester students. These variables
(SATVerbal,SATMath,andhighschoolGPA)wereentered
intothefirststepoftheregression,withMMCBscoresadded
in the second step. Table 3 shows the expected relation be-
tweenpastperformance(asmeasuredbySATandhighschool
GPA)andperformanceinthepresentcourse:pastperformance
was highly predictive of psychology grades. However, and
moreimportantlyforpresentpurposes,addingtheMMCBin
Step 2 increased the model’s predictive value (ΔR2=.04,
p=.002). Differences in structure building significantly pre-
dictedfinalgradesinpsychology(β=.21,p=.002),evenafter
accountingforhighschoolGPAandSATscores.
Figure1. FinalgradesintheIntroductiontoPsychologycourseas
DidMMCBscorespredictperformanceincollegeBiology
afunctionofscoresontheMMCB.Theregressionlineisderived
overandabovethemeasuresofscientificreasoning(LCSTR)
fromthemodelwithMMCBastheonlypredictor
andknowledge(BCI)?Aswiththeprioranalysis,thescience
variables were entered into the first step of the regression
model,andMMCBscoreswereaddedinthesecondstep.As
Table4shows,science-reasoningskills,butnotpriorbiology
knowledge, predicted final course grades. More importantly,
adding MMCB scores to the model increased its predictive
value (ΔR2=.04, p=.048). After accounting for science-
reasoningskillsandpriorbiologyknowledge,structurebuild-
inguniquelypredictedfinalgrades(β=.20,p=.048).
DISCUSSION
Student ability to construct coherent mentalmodels (as mea-
Figure2. FinalgradesintheIntroductoryBiologyfornon-majors
suredbytheMMCB)predictedgradesinbothcollege-levelin-
courseasafunctionofMMCBscores.Theregressionlineis
troductorypsychologyandintroductorybiologycourses.This
derivedfromthemodelwithMMCBastheonlypredictor
work moves the field beyondrelying onpastperformance to
predict future performance, and instead identifies a specific
Question2:CanStructure-BuildingPredictGradesin
College? Figures 1 and 2 show that students with greater
structure building ability performed better in both the 3Psychology:adjustedR2=.05,Biology:adjustedR2=.08
Copyright©2016JohnWiley&Sons,Ltd. Appl.Cognit.Psychol.30:454–459(2016)
458 K. M. Arnold et al.
Table 3. Regressionresultspredictingfinalcoursegradeinthepsy- reading a textbook (Callender & McDaniel, 2007) or when
chologycourseusingpastperformance(step1)andMMCB(step2)
forcedtoconsidertherelationshipsbetweenpartsofthetext
Psychology(n=148) (e.g.,byunscramblingit;McDaniel,Hines,&Guynn,2002).
Amoretractabletechniquefortheclassroomthathasshown
Predictorvariables F ΔR2 β Partr
successinthelaboratoryistoprovideaidsduringlecturethat
Step1 32.20*** .40*** helpscaffoldconstructionofamentalmodelofthesystemor
SAT—Verbal .39*** .37*** processthatthelectureistargetingbyprovidingillustrativedi-
SAT—Math .16* .15* agramstothe learner(Bui& McDaniel,2015).However,no
HSGPA .37*** .36***
studieshaveyetexaminedhowtosupportlowcomprehenders
Step2 27.97*** .04**
SAT—Verbal .37*** .35*** whenthesetoolsareunavailableorwhenthecontentdoesnot
SAT—Math .15* .15* afford a visual diagram. The implications of an intervention
HSGPA .31*** .28*** couldbebroad,asthesimilarityofresultsacrossPsychology
MMCB .21** .19** and Biology suggests that finding ways to compensate for
lowcomprehensionhasthepotentialtohelpstudentsacross
MMCB=multi-media comprehension battery. Part r is equivalent to
semipartial r and indicates the unique contribution of each independent subject areas.
variable. Findingawaytosupportstructure-building skillsmaybe
*p<.05.
**p<.01. especially important when courses require students to make
***p<.001. inferences or transfer knowledge to new situations. Infer-
ences require learners to connect disparate ideas, a task that
Table 4. Hierarchical regression results predicting final course may be especially difficult when those ideas are modeled
gradesinthebiologycourse in separate structures rather than in one unified structure
(Gernsbacher, 1997). Similarly, a unified structure would
Biology(n=83)
enhance one’s ability to recognize similar patterns across
Predictorvariables F ΔR2 β Partr disparate environments. Recognizing such similarities is a
necessary prerequisite for transferring old knowledge to
Step1 8.83*** .18***
solve new problems (Gentner, 1983).
LCTSR .36** .35**
BCI .15 .14 We close by emphasizing that we are not arguing that
Step2 7.45*** .04* structure-building is the only cognitive skill that matters for
LCTSR .33** .32** academic success. Clearly, structure-building is just part of
BCI .14 .14
the cognitive toolkit the student brings to the classroom.
MMCB .20* .20*
For example, consider the most-researched individual differ-
LCTSR=Lawson’sClassroomTestforScientificReasoning.BCI=Biology ence in cognitive psychology, working memory. The ability
Concept Inventory. MMCB=multi-media comprehension battery. Part r is
to hold and manipulate information in memory is correlated
equivalenttosemipartialrandindicatestheuniquecontributionofeachinde-
pendentvariable. withmanyskills,includingmeasuresofacademicsuccesssuch
*p<.05. asself-reporteduniversityGPA(Gropper&Tannock,2009),
**p<.01.
high school grades and ACT scores (Cowan et al., 2005),
***p<.001.
and SAT scores (Hannon & McNaughton-Cassill, 2011).
skillassociatedwithacademicsuccess.Furthermore,thisskill Structure-building itself is modestly correlated with working
predicted performance in two disparate subjects, Psychology memory (r=.13; Martin et al., 2015). However, working
and Biology, suggesting structure building is a general memoryisalessflexibletoolinthatmostattemptstotrain
comprehensionskillthatunderliesacademicsuccessacross working memory have been relatively unsuccessful, when
domains. These findings extend Gernsbacher’s structure success is defined as transfer to other tasks (Unsworth,
building model (Gernsbacher, 1997) from explaining only Redick, McMillan, Hambrick, Kane, & Engle, 2015).
narrativecomprehensiontoalsoexplainingimportantaspects Structure-buildingmaybeabettercandidatefor an interven-
of learning complex academic (non-narrative) information. tion,totheextentthatstructure-buildingcouldbeimproved
Thus, these findings reinforce the preliminary evidence that (perhapswithself-explanationreadingtraining;McNamara,
structurebuildingisacoreskillincomprehension,regardless 2004; although Gernsbacher’s, 1997, suppression-deficit
ofthenatureoftheeventsormaterialsthatthelearneristrying modelmaydisfavoratrainingapproach).
tounderstand. Structurebuildingisalsomoderatelycorrelatedwithreading
Structurebuildingisclearlyanimportantskillthatisasso- comprehension (r=.46; Maki et al., 1994), another important
ciated with learning. However, because of the correlational skill that has been associated with academic success (Feldt,
natureofthisstudy,itisnotclearifthisskilldirectlyimpacts 1988). Tests for reading comprehension, such as the popular
learning or if training in structure-building would result in Nelson–Dennyreadingtest(Brown,Nelson,&Denny,1973),
highergrades.Forexample,itispossiblethatathirdunmea- andthe MMCB share surface feature similarities (e.g.reading
suredvariable isdrivingbothscores. However, wedothink short passages, answering multiple-choice questions about the
itisnoteworthythatpriorlaboratorystudieshaveshownthat passages),andthemoderatecorrelationbetweenthetestsindi-
certain interventions can help less-able structure-builders cate that they measure some of the same variance. However,
compensatefortheirpoorstructurebuildingskills.Less-able the Nelson–Denny and the MMCB target somewhat different
structure-builders show improved learning and memory for skills, as is reflected in different patterns of results for text-
texts when required to answer embedded questions while presentation manipulations across low-skilled reading
Copyright©2016JohnWiley&Sons,Ltd. Appl.Cognit.Psychol.30:454–459(2016)
Structure building predicts grades in college 459
(Nelson–Denny) comprehenders and low-skilled structure Cohn,E.,Cohn,S.,Balch,D.C.,&Bradley,J.(2004).Determinantsof
builders. McDaniel et al. (2002) found that increasing the undergraduateGPAs:SATscores,high-schoolGPAandhigh-schoolrank.
difficulty of processing the words in a story (by deleting EconomicsofEducationReview,23(6),577–586.
Cowan, N., Elliott, E. M., Saults, J. S., Morey, C. C., Mattox, S.,
18% of the letters throughout the text, which the reader
Hismjatullina,A.,&Conway,A.R.(2005).Onthecapacityofattention:
had to fill in) decreased recall of the story for low reading Its estimation anditsrole inworkingmemoryand cognitiveaptitudes.
comprehenders relative to an intact-text control, but did not CognitivePsychology,51(1),42–100.
compromiserecallforlowstructure-builders(relativetoanin- Feldt,R.C.(1988).Predictingacademicperformance:Nelson–DennyRead-
ingTestandmeasuresofcollegestudents’studyofexpositoryprose.Psy-
tacttextcontrol).Incontrast,whenthestorywasmademore
chologicalReports,63,579–582.
difficult by presenting the sentences in random order and re-
Gentner,D.(1983).Structure-mapping:Atheoreticalframeworkforanalogy.
quiringreaderstorearrangethesentencesintoacoherentstory, CognitiveScience,7(2),155–170.
recallofthestoryimprovedforlow-skilledstructure-builders Gernsbacher,M.A.(1997).Twodecadesofstructurebuilding.Discourse
(relativetotheirrelativelylowperformanceintheintacttext Processes,23(3),265–304.
condition), whereas low-reading comprehenders’ recall did Gernsbacher,M.A.,&Varner,K.R.(1988).Themulti-mediacomprehension
battery.Eugene,OR:InstituteofCognitiveandDecisionSciences.
not improve relative to the intact text condition, a condition
Gernsbacher,M.A.,Varner,K.R.,&Faust,M.E.(1990).Investigating
inwhichtheirperformancewasrelativelyhigh(theideaisthat differences in general comprehension skill. Journal of Experimental
the sentence arranging intervention stimulated low-structure Psychology.Learning,Memory,andCognition,16(3),430.
builders to create a more organized representation then they Gropper,R.J.,&Tannock,R.(2009).Apilotstudyofworkingmemoryand
academicachievementincollegestudentswithADHD.JournalofAttention
wouldotherwise).Theseexperimentaldissociationsconverge
Disorders,12(6),574–581.
withthetheoreticalassumptionsthatstructurebuildingreflects
Hannon, B. (2014). Predicting college success: The relative contribu-
skill in creating coherent mental models and not necessarily tions of five social/personality factors, five cognitive/learning fac-
front-end reading processes (such as lexical decoding), tors, and SAT scores. Journal of Education and Training Studies,
whereasstandardreadingtestsarehighlysensitivetotheskills 2(4), 46–58.
Hannon,B.,&McNaughton-Cassill,M.(2011).SATperformance:Under-
for front-end reading processes (Mason, 1978, 1980; Petros,
standing the contributions of cognitive/learning and social/personality
Bentz,Hammes,&Zehr,1990).
factors.AppliedCognitivePsychology,25,528–535.
Insum,structurebuildinggoesbeyondreadingbymeasur-
Hestenes,D.,Wells,M.,&Swackhamer,G.(1992).Forceconceptinventory.
ing one’s general ability to build a mental model of events ThePhysicsTeacher,30,141–158.
anddiscourse,regardlessofmodality.Inaclassroom,students Kalat, J. W. (1998).Biologicalpsychology(6th edn). Pacific Grove, CA:
areexpectedtolearnfromnotonlyreadings,butalsolistening Brooks/ColePublishingCompany.
Klymkowsky,M.W.,Underwood,S.M.,&Garvin-Doxas,R.K.(2010).
tolectures,viewinggraphsandvideos,participatingindiscus-
Biological Concepts Instrument (BCI): A diagnostic tool for revealing
sions, and many other tasks. Because of the multi-modality
studentthinking.arXiv:1012.4501.
nature of educational contexts, a skill like structure-building Lawson,A.E.(1978).Thedevelopmentandvalidationofaclassroom
that crosses modalities is likely a better target than reading test of formal reasoning. Journal of Research in Science Teaching,
comprehension for interventions, a possibility that should be 15(1),11–24.
Lawson,A.E.,Alkhoury,S.,Benford,R.,Clark,B.R.,&Falconer,K.A.
exploredinfutureresearch.
(2000). What kinds of scientific concepts exist? Concept construction
andintellectualdevelopmentincollegebiology.JournalofResearchin
ScienceTeaching,37(9),996–1018.
ACKNOWLEDGEMENTS
Maki, W. S., & Maki, R. H. (2002). Multimedia comprehension skill
predictsdifferentialoutcomesofweb-basedandlecturecourses.Journal
Thisresearch wassupported inpartbygrant R305A130535 ofExperimentalPsychology:Applied,8,85–98.
toDukeUniversityfromtheInstituteofEducationSciences, Maki, R. H., Jonas, D., & Kallod, M. (1994). The relationship between
comprehensionandmetacomprehensionability.PsychonomicBulletin
US Department of Education. The opinions expressed are
&Review,1(1),126–129.
those of the authors and do not represent the views of the
Martin,M., Nguyen, K., &McDaniel,M.A.(2015).Structure building
Institute or the US Department of Education. differencesinfluencelearningfromeducationaltext:Effectsonencoding,
retention,andmetacognitivecontrol.UnderReview.
Mason,M.(1978).Fromprinttosoundinmaturereadersasafunctionof
REFERENCES
readerabilityandtwoformsoforthographicregularity.Memory&Cognition,
6,568–581.
Baron,J.,&Norman,M.F.(1992).SATs,achievementtests,andhigh-school Mason,M.(1980).Readingabilityandtheencodingofitemandlocation
classrankaspredictorsofcollegeperformance.EducationalandPsychological information.JournalofExperimentalPsychology:HumanPerception
Measurement,52,1047–1055. andPerformance,6,89–98.
Brown,J.I.,Nelson,M.J.B.,&Denny,E.C.(1973).TheNelson–Denny McDaniel, M. A., Hines, R. J., & Guynn, M. J. (2002). When text
readingtest.Boston,MA:Houghton-Mifflin. difficultybenefitsless-skilledreaders.JournalofMemoryandLanguage,
Bui,D.C.,&McDaniel,M.A.(2015).Enhancinglearningduringlecture 46(3),544–561.
note-takingusingoutlinesandillustrativediagrams.JournalofApplied McNamara, D. S. (2004). SERT: Self-explanation reading training. Dis-
ResearchinMemoryandCognition,4(2),129–135. courseProcesses,38,1–30.
Cahill, M. J., McDaniel, M. A., Frey, R. F., Hynes, K. M., Repice, M., Petros,T.V.,Bentz,B.,Hammes,K.,&Zehr,H.D.(1990).Thecompo-
JiuqingZhao,J.,&Trousil,R.(2016).Understandingtherelationship nentsof textthatinfluencereadingtimesand recallin skilledand less
betweenstudentattitudesandstudentlearning.UnderReview. skilledcollegereaders.DiscourseProcesses,13,387–400.
Callender, A. A., & McDaniel, M. A. (2007). The benefits of embedded Savinainen,A.,&Scott,P.(2002).TheForceConceptInventory:Atoolfor
questionadjunctsforlowandhighstructurebuilders.JournalofEducational monitoringstudentlearning.PhysicsEducation,37(1),45–52.
Psychology,99(2),339–348. Unsworth,N.,Redick,T.S.,McMillan,B.D.,Hambrick,D.Z.,Kane,M.J.,&
Callender,A.A.,&McDaniel,M.A.(2009).Thelimitedbenefitsofrereading Engle,R.W.(2015).Isplayingvideogamesrelatedtocognitiveabilities?
educationaltexts.ContemporaryEducationalPsychology,34(1),30–41. PsychologicalScience,26,759–774.
Copyright©2016JohnWiley&Sons,Ltd. Appl.Cognit.Psychol.30:454–459(2016)