1 Submitted to the American Journal of Physics
The impact of epistemology on learning: A case study from introductoryphysics
Laura Lising* and Andrew ElbyDepartment of Physics, University of Maryland, College Park, Maryland 20742
We discuss a case study of the influence of epistemology on learning for a student in anintroductory college physics course. An analysis of videotaped class work, written work,and interviews indicates that many of the students difficulties were epistemological innature. Our primary goal is to show instructors and curriculum developers that a studentsepistemological stance her ideas about knowledge and learning can have a direct,causal influence on her learning of physics. This influence exists even when research-based curriculum materials provide implicit epistemological support. For this reason,curriculum materials and teaching techniques could become more effective by explicitlyattending to students epistemologies.
In the past 15 years, physics education researchershave identified student difficulties in learning a broad rangeof physics concepts. Curricula targeting these difficultieshave produced dramatically improved conceptualunderstanding.1 In recent years, the physics educationresearch community also has begun to look at studentattitudes, expectations, and epistemologies (ideas aboutknowledge and learning).2,3,4 For instance, students maythink of physics knowledge as disconnected facts andformulas, or as interconnected concepts (often expressible asformulas). Students may think of learning physics asabsorbing information from authority or as building up theirown ideas.5 This discipline-specific epistemology researchbuilds on extensive research on more generalizedepistemology.6
The recent focus on epistemology in physics educationstems in part from two motivating ideas: (i) Studentsepistemologies may affect their science learning. In thatcase, attending to epistemology may help us explain thevariations in student learning outcomes with research-basedcurricula, create more effective curricula, and become betterphysics instructors. (ii) Fostering productive attitudes andepistemologies is in itself an important instructionaloutcome that could serve the students well beyond thecourse in question.
Our study addresses the first of these ideas and buildson previous research on college and pre-college learners.Most previous research has looked at correlations betweenepistemological measures and learning outcomes, findingthat specific clusters of epistemological beliefs correlatewith academic outcomes such as grade point average7 andmathematical text comprehension.8 In the physical sciences,one study found that certain epistemological beliefs correlatewith integrated conceptual understanding in middle school,9
while another found a correlation with ninth-graders abilityto reason on applied tasks.10 In college physics, May andEtkina found correlations between students gains onstandard conceptual measures and their epistemologies asinferred from weekly written reflections on their ownlearning.11
A few studies have gone beyond these correlationsto look at the causal influence of epistemology on studentslearning behavior. These studies, generally carried out byobserving students in the process of learning, have
attempted to describe not just whether, but how learning isaffected by epistemology and related factors. An excellentexample is Hogans thorough study on eighth-graders inwhich she observed relationships between studentspersonal frameworks for science learning and their socialand cognitive engagement patterns during group learning.12
Ryder and Leachs study found some correlations betweencollege students ideas about the nature of scientificknowledge and their self-reported activities duringinvestigative project work.13 Millar et al. observed that,among 9 to 14-year-olds, students interpretations ofclassroom inquiry tasks varied according to their perceptionsof the aims of scientific investigation.14 Taylor-Robertsonfound differences in cognitive strategies used by collegestudents according to their expectations of themeaningfulness of laboratory work,15 and Edmondson foundcorrelations between students reported learning strategiesand their epistemological stances as derived frominterviews.16 Dwecks work with students of varied agesshowed some dramatic differences in learning behavior inthe classroom which depended on students ideas about thenature of intelligence.17 And Hammers study on collegestudents described how students ideas about knowledge andlearning in physics affected how they solved physicshomework problems during think-aloud interviews.2 Takentogether, these studies suggest a causal link betweenepistemology and learning and also raise new questions andissues. One issue is the distinction between personal andpublic epistemologies. Public epistemology encompasses astudents ideas about the nature of knowledge and learningfor society as a whole - or for a disciplinary community.Personal epistemology concerns a students ideas about herown knowledge and learning. A students public andpersonal epistemologies can differ significantly. Forinstance, a student may doubt the possibility of coherencein her own knowledge (personal epistemology), but mayexpect scientists to seek and find coherence (publicepistemology). Some of the previous correlation studieshave looked at only one of these aspects of epistemology,while others have not made this distinction. Of the threepersonal frameworks for science learning in Hogansresearch, one aligns fairly closely with personalepistemology while another aligns with publicepistemology. She found that personal epistemology waslinked strongly to the students behavior, while publicepistemology showed almost no effect. Thus her results
2 Submitted to the American Journal of Physics
point toward personal epistemology as being much morerelevant to learning. For that reason, we focus on thepersonal epistemology of our student subject. This paperbuilds more on the work of Hogan,12 Hammer,2 and Mayand Etkina,11 which focused on personal epistemology, thanon the other studies mentioned above,3,4,7-10,13-16 whichlooked at public epistemology or a combination of personaland public epistemology and other attitude-related variables.
To build on this line of research, we have done anin-depth and naturalistic case study of a single student todistill and carefully describe the likely causal mechanisms.Of course, a case study cannot produce definitive,generalizable results about causality. But it can add depthand detail to the perceptive toolkit of the instructor andcurriculum developer by exploring specific causalmechanisms that might explain the correlations, and it cangenerate specific hypotheses about causal mechanisms forlater testing in controlled-intervention studies. Thefollowing hypothetical example illustrates this point.Suppose a correlation is found between how quickly peoplelearn rock climbing skills and how many safe exposures toheights they experienced as children. A possible causalmechanism underlying this correlation might be that lack ofsafe exposure to heights as children leads to a fear ofheights, which then leads to some learners making morecautious movements. Case studies of a few slow-learningnovice rock climbers might shed light on this hypothesis.As they first attempt new moves, do they give clues to theirfear of heights verbally or physiologically? Can we rule outother possible causes by watching their behavior in detail? Ifso, the next step toward establishing causal mechanismmight be a controlled-intervention study, safely exposingchildren to heights, enrolling them in a rock climbing class15 years later, and comparing their learning speed to acontrol group who received a different intervention aschildren (for example, reading about rock climbing). Ourgoal is to develop a plausible existence argument anddescriptive analysis for one particular causal mechanismbetween epistemology and learning, a mechanism that wehope will be tested in future controlled-interventionexperiments.
The various previous studies we have cited alsovary in the extent to which they disentangled studentspersonal epistemologies from their expectations aboutwhats rewarded in a particular course. It can be difficult todistinguish between what a student thinks is productive forher learning and what she perceives is required by theteacher or the curriculum. Yet these can be quite disparate attimes. Hammers work with one student illustrates anexample where a student ruefully and self-consciouslyabandoned her productive learning strategies to survive in amemorization-focused physics course.18 A 1999 study byElby gave some insight into the magnitude of theepistemology/expectations gap.19
Yet another issue arising in previous studies is thecontext-sensitivity of students epistemologies. Survey-based research on students epistemologies has establisheddifferences in approaches according to discipline, motivatingresearch that is discipline-specific (such as Ref. 3).However, studies that involved observations of learningbehaviors and studies with multiple epistemologicalassessments also uncovered a sensitivity of epistemology tocontext within a given discipline. Hogan, for example,
found that epistemologies assessed in interviews differedfrom the approaches students took in class. One mightexpect this difference between students tacit ideas and theirexplicitly articulated ones, but Hogans interview methodsincluded elicitation of tacit ideas through scenario-posing.12
Thus it has become clear that taking context-sensitivity intoaccount when designing studies and analyzing data is crucialin understanding epistemology and learning.
In our study, we look at a student, Jan and studyboth her person