JOURNAL OF RESEARCH IN SCIENCE TEACHING

VOL. 37, NO. 2, PP. 109138 (2000)

Learning in Science: A Comparison of Deep and Surface Approaches

Christine Chin,1 David E. Brown2

1School

of Science, National Institute of Education, Nanyang Technological University,

469 Bukit Timah Road, Singapore 259756

2Department

of Curriculum and Instruction, College of Education, University of Illinois

at Urbana-Champaign, 1310 S. Sixth St., Champaign, Illinois 61820
Received 7 April 1998; accepted 9 September 1999

Abstract: The purpose of this study was to explore in greater depth what has been called by previous
researchers, a deep versus surface approach to learning science. Six Grade 8 students judged as typically
using learning approaches ranging from deep to surface were observed and taped during class group laboratory activities in a chemistry unit. They were also interviewed individually before and after instruction
about related science concepts. On analysis of the students discourse and actions during the activities and
their interview responses, several differences in learning approaches seemed apparent. These differences
fell into five emergent categories: generative thinking, nature of explanations, asking questions, metacognitive activity, and approach to tasks. When students used a deep approach, they ventured their ideas more
spontaneously; gave more elaborate explanations which described mechanisms and causeeffect relationships or referred to personal experiences; asked questions which focused on explanations and causes, predictions, or resolving discrepancies in knowledge; and engaged in on-line theorizing. Students using a
surface approach gave explanations that were reformulations of the questions, a black box variety which
did not refer to a mechanism, or macroscopic descriptions which referred only to what was visible. Their
questions also referred to more basic factual or procedural information. The findings also suggest that to
encourage a deep learning approach, teachers could provide prompts and contextualized scaffolding and
encourage students to ask questions, predict, and explain during activities. 2000 John Wiley & Sons,
Inc. J Res Sci Teach 37: 109138, 2000

Some students are more successful than others in learning science. This may be due to differences in the way students learnwhether it is meaningful or rote learning (Ausubel, 1968).
Meaningful learning requires relevant prior knowledge, meaningful learning tasks, and a meaningful learning set (Novak, 1988). In contrast, rote learning is arbitrary, verbatim, and not related to experience with events or objects, and lacks affective commitment on the part of the
learner to relate new and prior knowledge. The nature of students learningthat is, meaningful or roteis related to the construct approaches to learning.

Correspondence to: C. Chin

2000 John Wiley & Sons, Inc.

110

CHIN AND BROWN

Approaches to Learning
Approaches to learning or learning approaches refer to the ways in which students go about
their academic tasks, thereby affecting the nature of the learning outcome (Biggs, 1994). Research on approaches to learning derives much from the seminal work of Marton and Saljo
(1976) on reading from text using phenomenographic methods, where learning is studied from
the perspective of the learner, based on qualitative analysis of interview data and descriptive
analyses of differences between the learning behaviors of small numbers of students. These authors distinguished between deep and surface approaches to learning, and this distinction appears to be a powerful form of categorization for differences in learning strategies (Entwistle
& Ramsden, 1983, p. 17).
The general framework and defining features of the deep and surface approaches were described by Biggs (1987) and Marton (1983). In essence, the deep approach is associated with
intrinsic motivation and interest in the content of the task, a focus on understanding the meaning of the learning material, an attempt to relate parts to each other, new ideas to previous knowledge, and concepts to everyday experiences. There is an internal emphasis where the learner
personalizes the task, making it meaningful to his or her own experience and to the real world.
In contrast, the surface approach is based on extrinsic or instrumental motivation. The learner
who uses a surface approach perceives the task as a demand to be met, tends to memorize discrete facts, reproduces terms and procedures through rote learning, and views a particular task
in isolation from other tasks and from real life as a whole.
In his 3P (presage, process, product) (Biggs, 1987) or systems (Biggs, 1994) model of student learning, Biggs described both personal and situational factors as presage factors that exist before and which interact to determine the learning approach adopted in a learning situation.
Personal factors include ability, personality, locus of control, cognitive style, motivation, values, attitudes, prior knowledge, conceptions of learning, and general experiences, while situational factors comprise nature of task, time pressures, the context in which it is performed,
method of teaching, assessment, and perceptions of institutional requirements. These presage
factors affect the quality of performance (product) or learning outcomes via the learning approach (process) adopted. The learning approach comprises the motive for undertaking the task
and the congruent strategies used.
Marton (1983) and Ramsden (1988) stressed the importance of the particular context in determining whether a deep or surface approach to learning is used, and emphasized that the approach to learning should not be seen as a style-like, stable trait of the student, but as a response
to a situation. However, Entwistle (1981) and Schmeck (1988) believed that students have a
predilection to adopt deep or surface approaches that persist over different situations and refer
to this consistency as a learning orientation. Biggs (1987) also shared this viewpoint and argued
that although students can change their approaches to learning according to the demands of each
situation, the extent to which this change occurs is affected by the students predisposition to
change, which is in turn influenced by personal characteristics such as ability.
Learning in Science
Learning is a generative process requiring effort in which learners actively construct their
own meanings that are consistent with their prior ideas rather than passively acquire knowledge
transmitted to them (Osborne & Wittrock, 1983, 1985). Understanding occurs through both individual constructions and interactions with the natural and physical world, as well as the social process of communicating with others in the scientific community (Driver, Asoko, Leach,
Mortimer, & Scott, 1994). The development of rich, domain-specific conceptual knowledge

LEARNING IN SCIENCE: A COMPARISON

111

depends on the successful integration of the learners prior knowledge with the domain knowledge. If prior knowledge and disciplinary knowledge do not connect and intertwine, learning of
scientific concepts is reduced to rote memorization of facts (Roth, 1990). Furthermore, if no
learning occurs, this could be because few links are generated between constructed meanings
and existing ideas, learners have little or no motivation to change their existing ideas, or they
are unable or unwilling to restructure earlier ideas (Osborne & Wittrock, 1983, 1985).
Understanding in science can be facilitated by learners generating analogies, metaphors,
problems, models, and related devices that build meaningful relations between new information
and past experience or among the parts of the new information (Wittrock, 1994). For example,
studies by Wong (1993a, 1993b) showed that self-generated analogies are tools by which individuals can generate, evaluate, and modify their own explanations, and advance their conceptual understanding in science. The spontaneous generation of analogies in scientific problem
solving was also documented by Clement (1988).
The role of explanations in science learning is potentially important, since explanation and
the understanding of how and why something happens are major aims of science as a whole.
Indeed, self-explanations (i.e., those spontaneously generated) have been associated with
greater problem-solving success and gains in deeper understanding when students learn from
science texts (Chi, Bassok, Lewis, Reimann, & Glaser, 1989; Chi, de Leeuw, Chiu, &
Lavancher, 1994). Solomon (1986) described different modes of explanations that students
gave. These included immature precausal explanations such as reaffirmation, teleology, tautology, and simple juxtapositions. These types of explanation have little semantic movement
from the event to be explained, which is an important aspect of a successful and valuable explanation. Metz (1991) found that children progressed through three developmental phases of
explanation types: explanations that focused on the function of the object (Phase 1), those
that emphasized connections relying on spatial connections and the physical relationship between the parts of the systems (Phase 2), and mechanistic explanations which were fundamentally more adequate (Phase 3). Woodruff and Meyer (1997) also found three salient levels
in the evolution of students explanations during group discourse. First-order explanations described properties of objects, and second-order explanations described a set of relations among
variables. Third-order explanations went beyond descriptions, proposed mechanisms, and dealt
with the coherence of various conditions and effects. These studies on students explanations
showed a progressive trend from precausal explanations to causal explanations which focus on
causeeffect relationships.
Such causal explanations which relate to a logical chain of causality have been considered
to hold more explanatory power and are highly valued (Gilbert, Boulter, & Rutherford, 1998a;
Horwood, 1988). Indeed, as pointed out by Tamir and Zohar (1991), in the physical sciences,
providing an explanation always implies causeeffect relationships. Also, models play key
roles in the explanations of science (Gilbert, Boulter, & Rutherford, 1998b), particularly in theory building, and expert scientists frequently rely on models to concretize abstract ideas, explain
mechanisms and processes, and explicate real-world phenomena. However, the typical school
students naive conception of models as physical copies of reality rather than constructed representations that serve to test ideas and explanations about phenomena (Grosslight, Unger, &
Jay, 1991) could influence their appreciation of explanations.
Asking questions is a key element in the learning process. Questioning helps students direct their learning as they try to reconcile their prior knowledge and new information in their
attempts to make sense of these ideas. Student questioning, particularly at the higher cognitive
level, is also an essential aspect of problem solving (Pizzini & Shepardson, 1991; Zoller, 1987).
Students questions play a significant role in meaningful learning and motivation, and can be

112

CHIN AND BROWN

very revealing about the quality of students thinking and conceptual understanding (Watts,
Gould, & Alsop, 1997), their alternative frameworks and confusion about various concepts
(Maskill & Pedrosa, 1997), their reasoning (Donaldson, 1978), and what they want to know
(Elstgeest, 1985). Also, interesting and productive answers are dependent on being able first to
come up with good questions for eliciting them (Shodell, 1995). Low levels of questioning and
explaining on the part of the students have been found to be correlated with lower achievement
(Tisher, 1977). The number and type of questions that students ask may be influenced by their
age, experiences, prior knowledge, skills, the attitude of the teacher, teaching style, nature of
the topics, reward structure, classroom evaluative climate, and social interaction patterns (Biddulph & Osborne, 1982).
Closed questions and those requiring simple recall of information are easier to generate and
thus more commonly asked than questions requiring deep processing of ideas such as those involving application or extension of taught ideas. Few students spontaneously ask high-quality
thinking questions (White & Gunstone, 1992, p. 170). There has been relatively little research
on students questions, particularly the types of questions that students ask. One study, however, that attempted to investigate students questions of understanding in science was reported by
Watts et al. (1997). The authors discussed three categories of students questions which were
seen to illuminate distinct periods in the process of conceptual change: consolidation questions
in which students attempted to consolidate understanding of new ideas in science; exploration
questions in which they sought to expand knowledge and test constructs; and elaboration questions in which students attempted to examine claims and counterclaims, reconcile different understandings, resolve conflicts, test circumstances, and track in and around the ideas and their
consequences. A lack of domain-specific prior knowledge may influence the kinds of questions
that students ask. For example, Scardamalia and Bereiter (1992) found that students asked mainly basic information questions for a less familiar science topic but concentrated on wonderment questions which reflect curiosity, puzzlement, skepticism, or knowledge-based speculation for a more familiar topic.
When students engage in meaningful learning, they are purposeful and constantly monitor
and reflect on the process of learning to evaluate the results of their own learning efforts. Conversely, a lack of awareness of deficient processing habits in the learning process precludes
learning with understanding (Baird & White, 1982a). Learners are appropriately metacognitive
if they consciously undertake an informed and self-directed approach to recognizing, evaluating, and deciding whether to reconstruct their existing ideas and beliefs (Gunstone, 1994). This
is particularly important if any conflicting beliefs need to be resolved, as this resolution requires
elements of metalearning (conscious control over ones learning) that will facilitate the necessary conceptual change (White & Gunstone, 1989).
The studies by Hennessey and Beeth (1993) and Hennessey (1993) attempted to elucidate
the nature of metacognitive processes in facilitating conceptual knowledge development when
elementary students were engaged in thinking and reasoning about science content and the status of their conceptions. The metacognitive reflections ranged from a minimal level of awareness of ones conceptions through various levels of sophistication. These included reflections
that explicitly (a) referred to students own or a peers conceptions; (b) referred to the reasoning behind the students conceptions; (c) considered the implications or limitations inherent in
the students conceptions; (d) referred to the students thinking or learning processes; (e) referred
to the status of the students conceptions; and (f) referred to the components of the students
conceptual ecologies. Failure on the part of students to examine their conceptual understanding
and the cognitive processes that produce that understanding cannot lead to learning scientific
knowledge of a conceptual nature.

LEARNING IN SCIENCE: A COMPARISON

113

Recent research in science education has shown the relationship between students epistemologies and their approaches to learning science which, in turn, influence their choices of
learning strategies and whether they integrate what they learn. For example, Edmondson and
Novak (1993) found that students identified as logical positivists tended to be rote learners oriented to grades, whereas those identified as constructivists used meaningful learning strategies
as the primary goal of their understanding of the material. Tsai (1998) found that students having constructivist epistemological beliefs engaged in more active learning as well as used more
meaningful strategies when learning science, whereas students having epistemological beliefs
more aligned with empiricism tended to use more rotelike strategies because they believed science was like a collection of correct facts. Thus, students epistemological beliefs seem to shape
their metalearning assumptions and influence their learning orientations, and the adoption of a
constructivist epistemology is related to more meaningful learning.
Hogan (1999) examined relationships between students personal frameworks (motivational
and epistemological perspectives comprising self-referenced, learning-referenced, and disciplinereferenced perspectives) and their sociocognitive engagement with peers during collaborative
knowledge-building tasks in science. Students who held exogenous constructivist perspectives
on learning were less sociocognitively engaged in the task, whereas those with relativist perspectives were highly engaged. Students with low self-efficacy for science, higher ability than
effort attribution, and low-to-medium interest in science displayed lower levels of sociocognitive engagement.
Students epistemological beliefs about knowing and learning affect not only the learning
strategies employed, but also their stance toward activities in the classroom (Roth & Roychoudhury, 1993, 1994). For example, the objectivists who consider scientific knowledge to be
absolute would regard the textbook as a store of knowledge to be memorized and practised. Such
students do not try to find things out on their own, but always rely on peers and teachers to provide them with guidance to get things right. On the other hand, those with a more constructivist
perspective would use laboratory activities to arrive at new knowledge on their own, aided by
feedback. Also, when students view science as dynamic, they are more inclined to seek principles to explain and integrate their ideas and to build more predictive ideas about science (Songer
& Linn, 1991). In contrast, if they view science as static, they may prefer to leave the grappling
of complex ideas to scientists and then wait until facts for them to memorize become available.
Lee and Anderson (1993) and Lee and Brophy (1996) have described how cognitive, motivational, and affective factors interact to influence students approaches to learning tasks and
the quality of their task engagement. Students exhibiting different patterns of motivation engaged in science tasks with different goals and levels of task engagementthose who were motivated to learn science demonstrated high quality of cognitive engagement. Thus, if students
are to learn meaningfully and engage in deep cognitive processing and self-regulated construction of meaning that is necessary for conceptual understanding, the students must engage in
learning tasks with certain motivational orientations that predispose them to use the appropriate
meaningful cognitive and metacognitive strategies. Students who are not motivated to learn are
likely to content themselves with strategies for meeting accountability pressures with the least
possible effort.
The concept of motivation to learn is related to students goals and learning strategies during task engagement, which in turn influences the quality of their cognitive engagement in the
activity (Lee & Brophy, 1996). When students pursue learning or task goals, they are interested in learning as an end in itself and seek to understand the task. They are likely to employ deep
cognitive and self-regulated strategies such as integrating information and monitoring comprehension which result in meaningful learning or conceptual understanding (Ames & Archer,

114

CHIN AND BROWN

1988; Meece, Blumenfeld, & Hoyle, 1988; Nolen & Haladyna, 1990). In contrast, students who
pursue performance (Dweck & Elliot, 1983) or ego (Nicholls, 1984) goals seek to demonstrate
their ability or gain social approval, and tend to use more surface-level strategies such as memorization. Students who pursue work-avoidant goals seek to complete their tasks without thinking too hard, and this orientation is associated with use of rote-level rather than deep-processing strategies (Nolen, 1988).
Students different goal orientations may also lead them to adopt different inquiry strategies during science experimental tasks. When their goal is to understand relations among causes and effects, they may adopt a science model of experimentation where there is a relatively
wide search of the experimental space and exploration of all levels of manipulable variables
(Schauble, Klopfer, & Raghavan, 1991). In such an approach, the emphasis is on the means and
process of experimentation. However, when the goal is to manipulate variables to produce a particular desired outcome, students often employ an engineering model and focus on the variables
most likely to affect the outcome, overlooking those presumed noncausal and thus leading to a
narrower search.
Rath and Brown (1996) described the different ways students orient themselves toward the
world emotionally, conceptually, and socially during science inquiry. They used the construct
modes of engagement (viz., exploration, engineering, pet care, procedural, performance, and
fantasy) to describe students orientations toward phenomena in a science classroom and the
ways in which these orientations influence the learning experience. The ways in which students
explore, manipulate, fantasize about, ignore, or orient themselves toward objects of study have
an impact on their motivation and ability to engage in sense-making activities.
Background to the Problem in the Present Study
Although the construct, approaches to learning, was originally conceptualized in the context of reading from texts (Marton & Saljo, 1976), it has been extended and generalized to other tasks such as writing (e.g., Biggs, 1988) and other content areas (Entwistle and Ramsden,
1983). One early study by Laurillard (1978, cited in Entwistle & Ramsden, 1983) of students
learning approaches in science based solely on self-reports found that a deep approach was associated with a sense of purpose in carrying out a task, attempts to relate ideas from different
topics to the task in hand, and the experience of the physical world to scientific concepts. On
the other hand, a surface approach was characterized by unthinking use of procedures in performing a task, and a focus on memorizing formulas and facts.
A few more recent studies on students learning in science also suggest that a students
learning approach is a factor influencing his or her learning outcome. The study by Cavallo and
Schafer (1994) on Grade 10 students understanding of genetics topics showed that the more
meaningful the students learning orientation was, the more meaningful was the understanding
they tended to attain (as measured by the interrelatedness and complexity of their knowledge exhibited). Studies by Hegarty-Hazel and Prosser (1991a, 1991b) showed that for student learning
in both electricity and photosynthesis, deeper and more meaningful strategies were associated
with better developed propositional knowledge and more surface strategies with less developed
propositional knowledge. A study by BouJaoude (1992) found that high school students learning approaches accounted for a statistically significant proportion of the variance on their performance on a misunderstandings posttest in chemistry. The relatively meaningful learners performed significantly better than the rote learners and, for the most part, developed more coherent
understanding of the underlying concepts.
These studies focused on the relationship between students learning approaches and their
science performance. Little research, however, has been done on how students learning ap-

LEARNING IN SCIENCE: A COMPARISON

115

proaches and learning strategies relate to the construction of their conceptual knowledge, and
how the use of these strategies figures into educational discourse, particularly in a naturalistic
setting. The focus of the above-mentioned studies was on the quantitative outcome of learning.
Although some studies (e.g., Baird & White, 1982b) provided some information about qualitative differences in learning in terms of application of cognitive strategies, few studies have had
a sufficiently descriptive approach about individual differences in science learning to offer
teachers alternatives to help improve students learning. Hence, more detailed information about
the finer differences in learning approaches that students adopt as they explore their ideas when
learning science will provide a more complete picture of what is meant by a deep or surface approach in learning science, and may shed light on the processes of knowledge acquisition that
lead to different learning outcomes. Furthermore, some of the tasks involved in learning science
(e.g., problem solving and hands-on investigations) go beyond merely reading texts and writing, and so students would use other strategies. Thus, whereas the meaning of the deep and surface constructs may be fundamentally the same across different subject areas, the application to
a particular subject such as science and its associated tasks may be manifested in slightly different ways.

Purpose of Study
The primary purpose of this study was to compare in greater depth the qualitative differences between what previous researchers have called a deep versus surface approach to learning science, and to describe in more detail the differences in the process of learning that lead to
different outcomes among individuals. Specifically, the following research question was addressed: What are the qualitative differences between a deep and surface approach to learning
science? The focus was on articulating the subtleties of the general constructs, deep and surface,
in the context of science learning.
This problem and its related findings are important because although some research has
been done on students learning approaches, there exists no principled and detailed account of
the characteristics associated with learning approaches specifically in the context of science
learning. This information would shed light on understanding individual differences, provide
further insight into what is meant by a deep and surface approach in the specific context of science learning, and extend our understanding of students learning approaches beyond earlier
conceptualizations which were more generic in nature.
Method
Selection of Students
As part of a larger study, the Learning Approach Questionnaire, which measured students
tendency to learn meaningfully using a deep approach or by rote using a surface approach, was
administered to 102 students from five Grade 8 classes in a school in a midwestern U.S. university town. This questionnaire containing 31 Likert items, was based on a slight modification
of the meaning and reproducing orientation scales of Entwistle and Ramsdens (1983) Approaches to Studying Inventory, which was originally written for university students, and which
referred only to general aspects of deep and surface approaches in noncontent domains. To focus students attention on learning science in particular and make the language more appropriate for middle school students, some of the original statements in Entwistle and Ramsdens

116

CHIN AND BROWN

instrument were revised by changing a few words. For example, the statement, I generally put
a lot of effort into trying to understand things which initially seem difficult was revised to I
generally put a lot of effort into trying to understand things in my science classes which seem
difficult at the beginning. The statement, Lecturers seem to delight in making the simple truth
unnecessarily complicated was revised to Science teachers seem to make simple ideas unnecessarily complicated, and the statement, I find that studying academic topics can often be
really exciting and gripping was revised to I find that studying science can often be really exciting and gripping.
When responding to the questionnaire, the students were told to consider each statement in
the context of their science lessons, and to respond to the statement using a 5-point Likert scale
ranging from 1 (never or rarely true) to 5 (always or almost always true). The students were
told that the purpose of the questionnaire was to gather information about how students usually learn science. No reference was made to learning approaches, to avoid cueing them that their
responses would serve to identify their typical learning approach. Item responses were coded so
that a higher score corresponded to a higher degree of meaningful or deep learning. The Cronbach alpha reliability coefficients of internal consistency for the meaning orientation and reproducing orientation scales were .86 and .65, respectively. Test-retest reliabilities over a 9-day period were .86 and .80, respectively, for these two scales.
One class was then observed during instruction of a chemistry unit which lasted 9 weeks.
In consultation with the teacher, six target students were selected for more in-depth study.
They represented learners typically using learning approaches which ranged from deep to
surface, as identified by the Learning Approach Questionnaire and their teachers evaluation
of their learning approaches in science. It was important that the teachers evaluation matched
the students score on the Learning Approach Questionnaire. Other selection criteria included: good attendance, being verbally expressive and on task, having at least average success
in science, and having the ability to work well with each other. The students were not informed about how they were representative of learners using different learning approaches
along the deep-surface continuum, to avoid any self-fulfilling prophecy or their conforming
to a stereotype.
The six target students worked in two groups of three during their class activities. The group
assignment was done by the teacher who in the past had experienced greater success with samesex groups. The boys group consisted of Rick, Quin, and Carl; the girls group was composed
of Mary, Bess, and Dale (all pseudonyms). Rick and Mary were identified independently by both
the Learning Approach Questionnaire and the teacher, as meaningful learners who used a predominantly deep learning approach, Carl and Dale as rote learners who typically used a more
surface approach while Quin and Bess used an approach that lay somewhere between a deep
and surface approach. Rick and Mary were A students, Quin and Bess were B students, and
Carl and Dale were C students. In designating these students as relatively meaningful or rote
learners, it does not imply that they were entirely consistent in their learning approaches or that
the learning approaches were stable in-built traits of the students. Often, the students used a
combination of both approaches which depended on contextual influences. Rather, the designations were intended to reflect their predominant learning mode and dispositions to the extent
that they could be determined on the basis of a questionnaire and the teachers evaluation. Furthermore, although students may adopt different learning approaches across different subjects
(e.g., science vs. social studies), all the tasks considered here are in the context of science learning. Hence, it seems reasonable to assume that there would be some consistency in a students
learning approach, and that variations from task to task would not be too extreme. Also, although
the focus of this study was on deep versus surface approaches, students who typically used a

LEARNING IN SCIENCE: A COMPARISON

117

mixed approach (viz., Bess and Quin) were also included. This would reflect real-life situations,
as some students may not use a predominantly deep or surface approach.
Procedure
The boys were audiotaped and the girls were videotaped (as only one video camera was
available) while engaged in science hands-on activities during their regular science classes, and
were encouraged to think aloud and to verbalize their thoughts. Field notes were taken based on
classroom observations which focused on classroom discourse and science activities. The topics covered in this chemistry unit included the nature of matter (elements, mixtures, compounds,
atoms and molecules), states of matter and changes of state, physical and chemical changes,
acids and bases. Instructional activities included teacher presentation of topics, hands-on activities, and whole-class discussions. Laboratory activities included the separation of a saltsand
mixture, plotting the temperature graphs of water and salt water, and paper chromatography using different colored pen inks.
The target students were also interviewed individually both before and after instruction of
the chemistry unit using a semistructured interview protocol to find out more about their ideas
and understanding of the science concepts in this unit, as well as their reasoning behind them.
They were told that the purpose of the interview was not to test them, but rather to obtain information about the diversity of students ideas regarding the chemical concepts of interest. They
were encouraged to tell what they really thought, and not to worry about whether they gave the
scientifically correct answers. The interviews were audiotaped. In addition, during the postinstructional interviews, stimulated recall was used to obtain further information about how the
students tackled the tasks and what they were thinking of while engaged in the laboratory activities. This provided information about silent thoughts which were not always verbalized and
captured on tape. For example, they would be given a description or narration of a critical
episode or quote from a segment of the tape pertaining to a specific activity, and asked to elaborate on what they were thinking of regarding what they did or said. They were also asked about
any ideas, predictions, observation, explanations, and questions that they had before, during, and
after doing the activities.
The use of verbal data from such individual or small-group think-aloud protocols has
been used widely in research on cognition and has been considered a suitable means of studying students thought processes (Ericsson & Simon, 1993). The latter authors have addressed
the issues regarding the completeness, veridicality, and epiphenomenality of verbal reports,
and using their model of human information processing and verbalization, have argued that
data from both concurrent and retrospective reports are pertinent to and informative about individuals cognitive processes and memory structures. Such verbalizations about what students were thinking and doing while completing a task in hand help to externalize cognitive
and metacognitive strategies, and are useful in giving researchers information about covert
strategic activity that is not accessible except as described by the strategy users (Garner,
1988). The use of stimulated recall in uncovering covert cognitive processes not observable
by either on-site observation or videotape viewing has also been suggested in studying students thinking (Garner, 1988; Peterson, Swing, Braverman, & Buss, 1982). The findings of
the study by De Grave, Boshuizen, and Schmidt (1996) also suggest that the information derived from stimulated recall of students thinking processes provide more and unique information about their cognitive, and metacognitive processes than that obtained by group verbal
interaction alone. This is because thinking aloud is not a natural process for many individuals, and the social interaction situation imposes constraints on the interaction in that not every

118

CHIN AND BROWN

student can reveal his or her thoughts at the same time, and some things are better not said in
public.
Data Analysis
Data from multiple sources (field notes, transcripts of classroom discourse from the audiotapes and videotapes, audiotaped interviews with the students, and students written work) were
analyzed in relation to each other; this served to triangulate the data and to help enhance the
credibility of the findings and assertions made (Lincoln & Guba, 1985; Stake, 1995). For example, segments of interview transcripts that corresponded to specific sections on the students
laboratory reports or classroom discourse were checked for congruence. Observation fieldnotes
used as secondary data sources provided a context for the interpretation of data. The target students taped interviews and discourse during class activities were transcribed verbatim and analyzed to determine the nature of their reasoning, argumentation, and sense making during their
engagement in problem-solving activites and in the development of their conceptual knowledge.
To obtain information about what the students did during the laboratory activities, transcribed
discourse from the videotapes was also supplemented with descriptive notes obtained by viewing the videotapes; this pertained to the students actions, behaviors, and expressions.
The interview and group discourse transcripts were analyzed using an iterative process.
The transcripts were first read through several times to get an impression and sense of the totality of the data. The next task was to determine an appropriate unit of analysis. Coding segments of the transcript were first identified; these were separated by transitional changes in
the situation (e.g., a change in episode such as an event or specific activity, or a shift in focus to a different discussion question). Coding categories (Bogdan & Biklen, 1992) were then
developed to organize the data. In particular, students ways of thinking about objects, events,
and phenomena, and various aspects of scientific discourse such as hypotheses, predictions,
observations, explanations, questions, evidence, arguments, models, and theories were noted
as these were considered to be relevant to their learning approaches in science. During the
coding process, annotated descriptive and interpretive comments were made in the margins
of the transcripts. The inductively derived categories that emerged became the tentative coding categories. Subsequent transcript segments were then annotated with the appropriate code.
In developing the coding categories, it was necessary to see if they could be operationalized
in the context of the data and how much coverage of the transcripts they accounted for. A constant comparative method (Glaser & Strauss, 1967) was used to cluster the codes into progressively more inclusive categories forming a hierarchical taxonomy or working typologies.
During this forward-and-backward testing process, categories were refined by analyzing
further transcript segments and by adding to, deleting from, or modifying the existing list of
categories. This resulted in a number of subcodes which categorized information at different
levels of the major codes.
Of particular interest were the differences between learners using a deep and surface approach to learning. The transcripts were studied closely to determine whether certain patterns
of behavior, thinking, or events repeated and stood out, and to search for any regularities pertaining to these differences. Assertions were made based on patterns observed which were
grounded in the data. Excerpts from the transcripts were used to exemplify these assertions.
These assertions encompassed areas for which there was most substantiation, and from which
grounded working hypotheses were developed. Working back and forth among the data from
the various sources helped detect relationships among the categories and refine the working hypotheses on the basis of confirming and disconfirming evidence (Lincoln & Guba, 1985).

LEARNING IN SCIENCE: A COMPARISON

119

In analyzing students learning approaches, a large part involved the study of students
thinking, behavior, and their internal states manifested in their observed performance. However, since it is not possible to gain direct access to the mind or internal state of the learner, the
analysis must be inferential to some extent. Thus, in developing grounded interpretive analyses
of the students learning approaches, the purpose was not to characterize what was really going
on in students, since this was not open to direct inspection. Rather, it was to articulate the subtleties of the general constructs deep and surface by looking closely at naturalistic student interactions and using interpretive analyses to assist in this articulation.
Results
From an analysis of the data, five categories emerged as meaningful perspectives that would
illuminate the differences between deep and surface approaches to learning. These categories
were five areas where such differences could be compared and contrasted, and thus used as a
framework to present the findings. They were (a) generative thinking, (b) nature of explanations,
(c) asking questions, (d) metacognitive activity, and (e) approach to tasks. Although the deep
strategies were used more consistently by the students representing the deep approach learners
than those representing the surface approach learners and were characteristic of their predominant learning mode, there were occasional instances when those representing the surface approach learners used deep strategies and when those representing the deep approach learners
used surface strategies. The Learning Approach Questionnaire and the teachers judgments were
used to select these representative students, and the students initially judged as using a predominantly deep approach were observed to use deep approaches more. However, the judgments of
whether particular types of thinking or behaviors were deep or surface were not made on the
basis of who was making them. Rather, they were made based more on theoretical considerations of what deep and surface mean, and which then constitute articulations of these more general terms. The target students were identified based on general criteria pertaining to deep and
surface learning approaches, and the examples given below serve to articulate and refine further
the idea of deep and surface learning approaches in the context of science learning. Thus, this
study should be considered to be more an empirically inspired and grounded conceptual analysis of students learning approaches than any kind of validation or correlation with the criteria
used for identifying the target students.
In the discourse segments, the following transcript conventions are used: [ . . . ] means a
pause of 3 s or less; [ . . . , . . . ] means a long pause (.3 s).
Generative Thinking
Generative thinking refers to the students ability to generate an answer when they do not
have an immediate ready-made solution to a problem, particularly when the problem is not familiar and the students cannot rely on simple recall of facts or something learned by rote. It embodies ideas of creativity, lateral thinking, and fluency in the generation of ideas. Four main levels of generative thinking emerged, corresponding to differing degrees of generativity shown by
the students. In Level 1, the students remained stuck, saying I dont know and did not seem
to proceed with thinking further. In Level 2, the students gave an evasive response that did not
directly answer the question but which was related to it. In Levels 3 and 4, the students attempted to think hard and give an answer which might be correct or incorrect. Whereas the answers at the third level were short and lacking in details, answers at the fourth level were more
elaborate and often more richly illustrated with specific examples and references to real life

120

CHIN AND BROWN

experiences. Levels 1, 2, and 3 would be characteristic of a surface approach to learning; Level 4 would be typical of a deep approach.
The following example illustrates these differences in generative thinking. During the
postinstructional interview, the students were asked questions pertaining to the class activity on
paper chromatography of pen inks. As they had performed the activity in a rather procedural
manner and seemed not to have thought about why it might be useful or important to calculate
the retention factor, Rf, they were asked what they thought chromatography was used for and
what was the purpose and significance of calculating the Rf, for each color spot. In class, the
teacher had only introduced the Rf as simply a formula, Rf 5 DC / DS, the distance the color
moved (DC) divided by the distance the solvent moved (DS).
Carl had blindly followed the instructions to calculate the Rf and said, I didnt think. Just
did it. He could not think of the purpose and significance of finding the Rf and remained stuck,
saying, I dont know, I have no idea repeatedly when coaxed for a possible answer. This would
be an example of Level 1 generative thinking. Bess had also just simply followed the procedures and was unable to come up with an idea of what the Rf might represent. When asked what
was the purpose and significance of finding the Rf, Bess gave a response, so that we could understand it a little bit more, that did not specifically answer the question (Level 2). Subsequently, upon further coaxing, she was still unable to generate an interpretation of the Rf, remaining stuck and saying, I dont know (Level 1). Finally, when asked what chromatography
was used for, she responded with to match something, without giving any specific referents
or elaborating further (Level 3).
Compare the answers given by Rick and Mary. Although their ideas of the Rf were not in
accordance with the scientific view, they were able to generate on the spot answers that seemed
plausible and quite complete in themselves compared to the other students, despite not having
extra content knowledge about the Rf. Rick thought of the Rf as being associated with the relative proportions of the component colors of the ink, and gave a specific hypothetical example
to illustrate his ideas. His example was that in comparing two shades of blue ink, the ink that
was darker in color would contain a higher percentage of dark blue dye component than the
other, and that this percentage would be reflected in a higher Rf value. Also, unlike Carl and
Bess, when asked about the applications of chromatography, he gave an example about mixing
different shades of colors in artists paints that was slightly different from the activity done in
class, but consistent with his idea of the Rf as a measure of the relative proportions of the component colors:
How to mix different colors. Like, if you want artists paints. Maybe he doesnt want
brown but he wants lightish brown. So he needs to find out what the standard brown is
made up of [ . . . ] so maybe he can add more of one substance and less of the other and
form different colors of brown.

Ricks response to the question would be illustrative of Level 4 generative thinking activity.
When Mary was asked to suggest a possible use of the Rf value, she came up with the idea
of Rf as an index of the purity of a substance, saying that an Rf of 1 must mean that its like a
pure color. . . . It doesnt have any colors that mix up to make with it. . . . So its a pure color,
so I know this isnt going to kill me if I ingest it. As for the applications of chromatography,
she also gave examples that were not simply exact reproductions of the crime lab chemistry
class activity regarding a ransom note and pen inks. Rather, her ideas were extensions of the activity and she mentioned the use of the Rf in identifying samples of food dyes, blood, and clothing dyes in forensic science. Here again, there seems to be generative thinking at Level 4. Rick

LEARNING IN SCIENCE: A COMPARISON

121

and Mary were not inclined to say I dont know when they were confronted with something
for which they did not have a ready answer. Instead, they would attempt to think and give a
plausible answer.
When students used a deep approach, they tended to venture their ideas more spontaneously. Their responses were longer, more sustained and elaborate, they dwelt more on a single idea,
and their ideas were more interconnected. The students usually resorted to daily life experiences,
past episodes, examples, and self-generated analogies as tools to keep the thinking going. This
indicates a greater degree of cognitive processing which helped the students to maintain their
thinking continuously as a chain reaction compared to the piecemeal thinking in spurts shown
by the other students when they were stuck for an answer. The language used was also more precise with specific referents. On the other hand, students using a surface approach tended to give
up thinking more easily, or gave a response that did not answer the question directly or that was
brief and unelaborated. Their thinking was more piecemeal and the students moved from one idea
to another without a sense of directional link between the isolated ideas. They seemed to be more
dependent on recall of factual knowledge. The language used was also more vague.
Nature of Explanations
There were four main types of explanations, reflecting different levels of depth and sophistication. The explanations associated with a surface learning approach tended to be reformulations of the question (Level 1); of a black box variety (Level 2) which did not refer to a
causal mechanism; or macroscopic (Level 3), which referred only to what was visible. Black
box explanations included mere observations or descriptions of what happened instead of telling
why or how an event occurred, rote explanations in which the student gave textbook-like answers that sounded scientific but did not relate meaningfully to the situation, global explanations which lacked specific referents, cyclic explanations in which the student switched back
and forth between two ideas without advancing, and anthropomorphic explanations. These explanations were also sometimes vague with nonspecific referents, and less elaborate. They were
typically given only when solicited and the student required more probing to generate a more
complete explanation. In contrast, a learner using a deep approach tended to give microscopic explanations (Level 4) which described nonobservable theoretical entities and causeeffect
relationships. This type of explanation was like a model or minitheory which served as a link
between the macro and micro levels. Microscopic explanations also included references to relevant personal experiences in trying to understand a phenomenon but might not articulate unseen mechanisms. In general, microscopic explanations were also given more spontaneously by
the student. To generate microscopic explanations, the student often used mental imagery, analogies, real-life experiences, or hypothetical examples to articulate his or her ideas. The use of
analogies, for example, may have a concretizing role in that they can help students construct
new explanatory models with unobservable (but imageable) structures or mechanisms (Brown,
1993; Brown & Clement, 1989).
The boiling point activity will be used to illustrate these different types of explanations. After plotting the temperature graph for ice water as it was heated to the boiling point in the absence and presence of salt, the students were asked why the temperature stayed constant at
1008C when plain water boiled, and about the effects of adding salt to the water. Since the
teacher had not discussed explanations of these phenomena in class, the students had to give
their own explanations instead of relying on any memorized ones.
Rick had initially thought that the temperature of plain boiling water would increase beyond 1008C. However, what he noticed during the activity seemed counterintuitive to him, and

122

CHIN AND BROWN

he apparently addressed this conceptual conflict by self-generating an explanation to account for

the anomalous data. The following excerpt from the group discourse illustrates this:
Rick: 100 [8C]. Its not going anywhere.
Carl: Okay, ready [ . . . ] now.
Rick: 100. I think its getting hotter but because the water is evaporating, the temperature
doesnt go up.

During the postinstructional interview, Rick elaborated on his new idea that heat was leaving
with the evaporated water, saying:
It could have been heating up but because the water was leaving, it wasnt trapping as
much heat. So it kind of stayed at the same level. . . . I guess it [heat] went away with the
evaporation of the water. It was like leaving with it.

Rick had formed an idea of the heat of vaporization although this concept was not formally introduced in class. What was it that made him offer an explanation so spontaneously during the
activity which he later refined and elaborated on during the interview? Perhaps it was his inclination to anticipate outcomes that made him alert to discrepant data.
Also, unlike the other four students (Carl, Quin, Bess, and Dale) who could not think of a
reason for why salt lowered the melting point of ice and raised the boiling point of water, Rick
had ready explanations for these observed phenomena. Before doing the activity, he had thought
that the melting point of salted ice would be lower since it [salted ice water] had more particles, it kind of trapped more cold than just the plain water. He also thought that the boiling
point of salted water would be higher, because there were more particles in it and it would attract more heat than just plain water. It would kind of trap the heat while the water just lets it
run through. After carrying out the experiment, Rick found that the boiling point of salt water
was 1028C. He explained:
Salt in it . . . makes the water thicker. And it kind of took more heat to melt the water that
had salt in it. . . . It [salt] kind of fills up a lot of empty spaces between the [water] molecules. And so the heat couldnt pass through it as fast as it did through the plain water.
So it had to add more heat to break through the salt particles and heat up the water.

Rick gave an explanation at the molecular level beyond what was perceptually obvious.
None of the other students gave an explanation in molecular terms. It is possible that his observation that salt makes the water thicker could have triggered him to think of the salt molecules filling up the empty spaces between the water molecules. Another noteworthy point about
Ricks explanations is that although they were not completely normative, they were coherent in
themselves and were attempts to come up with a model or minitheory that would explain the
mechanism of how things worked in the physical world beyond what was visible. He referred
to unobservable theoretical entities such as particles and molecules, and attempted to link these
at the micro level with his observation at the macro level, describing a causeeffect relationship. His microscopic explanation would be an example of Level 4.
Mary also predicted that salt would raise the boiling point of water. She thought about how
she had used salt on the driveway to melt ice in winter and reasoned that the temperature of
salt water would rise a little bit higher than regular water because it would conduct more heat
due to the salt. She also thought that salt would create more heat to speed up the process of
heating the water. Referring to the chemicals in salt and antifreeze, she thought that it had
something to do with the chloride in it that makes the thing heat hotter. . . . Like antifreeze for

LEARNING IN SCIENCE: A COMPARISON

123

your car has some kind of chloride in it. And thats what keeps the pipes in your car from freezing. Regarding the effect of salt on the boiling point of water, Mary thought of boiling baby
bottles, saying saltwater sterilizes things. . . . Itll make the water hotter so that it will kill more
bacteria and stuff that are on the baby bottles. I learned that from a friend who I babysit for.
Like Rick, it was also natural for Mary to predict and theorize about how and why things happened the way they did. The striking thing about the way she generated her explanations was
that she often related the phenomenon being considered to her experiences in daily life (e.g.,
use of salt in melting snow, as an antifreeze in cars, and in sterilizing baby bottles) and attempted
to make connections between them. She also seemed to be trying to postulate a mechanism and
causeeffect relationship; her explanation would also be an example of Level 4.
When asked why she thought the temperature remained constant when water boiled, Dale
did not give a causal explanation but described her graph instead, merely stating her observation. Upon further probing she replied, because it cant get any hotter. Her answer was more
a reformulation of the question in different words (Level 1). This type of response which is reiterative without adding any new information has little semantic movement (Solomon, 1986)
which is an important aspect of a successful and valuable explanation. Before doing the activity, Dale thought that the salt would make the water heat up faster, although she could not
give a specific reason for this. Unlike Ricks answers, which referred to a mechanism that might
account for the observation, Dales response, because . . . just adding something might help it,
was vague in that it did not specify a causeeffect relationship and there were no specific referents. Her ideas were of the black box variety and had little explanatory power (Level 2).
Besss attempts to explain the effects of salt on the boiling of water resulted in a cyclic explanation in which her responses kept returning to the former idea without advancing in any
way. When she was asked how saltwater affected the temperature curve, she said, it gets hotter faster. Then, in response to why that might be so, she replied because it has got salt in it.
When she was then again asked how salt might affect it, she returned to her former idea that it
just makes it get hotter. Thus, her responses were cyclic in that she kept going back and forth
between these two ideas. Such a cyclic explanation is also of the black box variety because it
does not offer a mechanism or explain how the observed effect might occur.
Quins explanations for the temperature remaining constant during the melting of ice was
anthropomorphic in nature, in that he attributed human qualities to the ice. Referring to the bottom plateau of the graph depicting the melting of ice, he said it was like it couldnt make up
its mind. . . . It was battling like the heat and the cold. To explain the lowered melting point of
ice in the presence of salt, Quin said that he had probably put more ice into the beaker, making
it colder. This would be an example of a macroscopic explanation (Level 3) in which the depression of the melting point was attributed to visibly more ice.
In summary, Level 4 explanations that were associated with a deep learning approach were
more coherent, focused on the mechanism of how things worked in the physical world, and tended to be more like minitheories or models which attempted to account for what was not perceptually obvious or related to personal experiences in daily life. In contrast, explanations at
Levels 13 focused on the gross and macroscopic aspects of students observations or were restatements of the question. Sometimes, they also had a black box quality to them in that they
were not really explanations, as they did not describe causal links or a mechanism for how things
happened.
Asking Questions
Questions associated with a surface approach tended to refer to more basic, factual (requiring only recall of information), or procedural information. Factual questions were often

124

CHIN AND BROWN

closed questions with a single unambiguous answer. They typically related to information in the
textbook, or some simple observation made about an event. Examples would be, What does
the dictionary say about salt? and What color is that? Blue? Procedural questions were those
which sought clarification about a given procedure or asked how a task was to be carried out,
particularly where step-by-step instructions had been given. Examples would be, Did she
[teacher] say to put it in a pan? and Could we pour this out now?
Questions associated with a deep approach to learning tended to be wonderment questions
(Scardamalia & Bereiter, 1992) which reflected the students curiosity, puzzlement, skepticism,
or speculation. They focused on explanations and causes, predictions, or on resolving discrepancies in knowledge, and were more open, imaginative, and reflective. They were pitched at a
conceptually higher level, required an application or extension of taught ideas, and apparently
sprang from a deep interest of the students or arose from an effort to make sense of the world.
They were asked when students tried to relate new and existing knowledge, integrate complex
and divergent information from multiple sources, or build internal associations among different
aspects of the new knowledge in their efforts to understand. Unlike basic factual and procedural questions, which were typically either ignored or simply responded to with a simple short answer, wonderment questions served to direct further inquiry, tended to elicit responses that were
of a more conceptual nature, and had a greater potential contribution for an advancement in conceptual understanding. When students used a deep approach to learning, they tended to ask wonderment questions more spontaneously, perhaps because asking such questions is, to a large extent, related to the inclination to self-generate explanations and construct theories.
Wonderment questions included (a) comprehension questions which typically sought an explanation of something not understood (e.g., Why do some pen inks run faster than others?),
(b) prediction questions which were of the What would happen if . . . variety involving some
speculation or hypothesis-verification (e.g., If you put more than one color, would it separate
into just more colors?), (c) anomaly detection questions in which the student expressed skepticism or detected some discrepant information and sought to address this anomalous data (e.g.,
So its staying at 1008C?), (d) application questions in which the student wondered of what
use was the information that he or she was dealing with (e.g., What is the Rf used for?), and
(e) planning or strategy questions in which the student was temporarily stuck and wondered how
best to proceed next when no prior procedure had been given (e.g., How are we going to recover the salt from the solution?). This list of categories of wonderment questions is not exhaustive, but includes most of the common questions that students usually asked.
Metacognitive Activity
This refers to the students use of comprehension-monitoring and evaluative strategies that
indicate they are reflecting on the learning process and their strategy of thinking and are aware
of the status of their comprehension. It includes two essential features: knowledge about cognitive states and processes, and regulatory (i.e., control or executive) aspects (Baird, Fensham,
Gunstone, & White, 1991; Brown, 1978). When using a deep approach, the students displayed
more cognitive self-appraisal and regulatory control of the learning process through ongoing reflective thinking. When using a surface approach, they engaged in less self-monitoring and selfassessment. When students used a deep approach, they (a) constantly self-evaluated their ideas
by making statements that expressed their understanding (e.g., Ive figured out what I wanted
to say!), that recognized a comprehension failure or impasse (e.g., No, I dont get it), and
that expressed value judgments (e.g., I know Im just rambling on, Oh no, I didnt draw that
right, Im getting confused); (b) made self-questioning statements when they encountered

LEARNING IN SCIENCE: A COMPARISON

125

something that confused them or when they realized that there was a block in their understanding (e.g., Oh no, what am I going to do?); (c) detected their errors and spontaneously selfcorrected them; (d) noted and attended to anomalous data and counterintuitive events; (e) considered a range of possible answers and attempted to understand alternative ideas; and (f)
considered limitations in their own or others ideas and critiqued them. The use of illustrative
cases focusing on students with predominantly different learning approaches best serves the purpose of comparing differences among such learners.

Example 1. This is an example of a negative instance showing a lack of self-questioning,

in which the student used a surface approach and simply followed procedures blindly without
thinking much about the purpose of what he was doing. In the activity on the separation of the
saltsand mixture, after the boys had already drained the salt solution from the wet sand and
were waiting for the salt to dry while the water from the solution was being evaporated over the
alcohol burner, Carl suggested that they transfer the salt onto a piece of filter paper.
Carl: Here. . . . Take one of those [filter paper] and put this on the top. And you put salt
on the top and salt stays on [. . .] the salt goes through and the water doesnt.
Quin: Why would we do that?
Carl: Thats what Mrs. Jones told me to do.
Rick: Salt goes through that [filter] paper ?
Carl: Yeah.
Rick: But why? We already have salt separated.

Earlier on in the activity, the teacher had told Carl that he could use a piece of filter paper
to filter the salt solution from the sand and then heat the salt solution to recover the dry salt. But
Carl had not understood the purpose of doing this, and thus thought of putting the dry salt on
the filter paper after the water had evaporated. When Quin asked him why they would do that,
Carl merely said, Thats what Mrs. Jones told me to do. Rick then pointed out to him that they
already had the salt separated. This episode shows the surface strategy of following procedures
blindly, and the absence of self-questioning as a comprehension-monitoring strategy. It seems
that Carl had not asked himself what would be the purpose of using the filter paper, and wanted to use it merely because the teacher had suggested doing so.

Example 2. This shows how Mary noted errors in her own thinking and spontaneously
made changes to them. Her thoughts were constantly in flux, thus leading her to reconstruct her
ideas and revise her original answers. When discussing her understanding of a chemical change
and a physical change, she initially thought that burning paper to ashes and smoke was a physical change, since she saw ashes as simply paper is in a new form and not chemically
changed. Upon further reflection, she said, But if you burn the paper . . . it could be a chemical change because the heat may have changed some of the properties of the paper. Then, she
thought about baking, saying, You can mix flour and sugar, and baking soda and eggs and combine them all together to make a chemical reaction with the heat thats baking it to make a new
thing. After comparing burning paper with baking, she reasoned that heat was a factor common to the two events, noticed the inconsistency in her thinking, and added, The heat makes
like a char on the paper. And thats a chemical reaction. This finally led her to conclude that
burning paper was a chemical change.

126

CHIN AND BROWN

The difference between learners using a deep approach and those using a surface approach
was not so much due to the amount of content knowledge they possessed as to the strategies
used when they were presented with a problem or new information. For example, when Mary
used a deep approach, she displayed more planning and both reflection-in-action as well as reflection-on-action. When faced with a problem, she tried to recall her past experiences and think
of things she knew that were associated with the problem in hand. For example, she would think
of similar tasks which matched the new task requirements, and ask herself questions of the type,
Have I come across this before? and What do I know about this? While executing a task,
she would constantly reflect to assess the extent to which her implemented ideas were working,
revise or modify them when necessary, make constant on-line adjustments, and use alternative
strategies if needed. She stopped periodically to assess the progress she was making to check
whether what she was thinking or saying made sense, and was alert to discrepancies in her thinking. She also used self-questioning as a source of feedback to correct her misunderstandings.
By being strategic, self-regulated, and reflective, she was more aware of when she made errors
and was better able to identify the loci of her misunderstandings spontaneously.
In contrast, when the students used a surface approach, they rarely reflected on their own
performance and seldom critically evaluated the new information, compared it with their prior
knowledge, or reflected on the efficacy of the procedures used to process the information. They
hardly made self-evaluation statements or asked themselves questions such as, Why am I doing what am I doing? (e.g. Carl, when he wanted to transfer dry salt onto a piece of filter paper after the water had already evaporated from the salt solution), What does this mean? (e.g.,
when the students did not question the meaning of the Rf ), Why does this happen? and Does
this seem correct? In this mode of learning, the students did not use an active, interrogative approach. Instead, they followed given procedures blindly and relied more on memorized facts.
Thus, the difference between a learner using a deep approach and a surface approach was more
of a qualitative difference in the strategies that they used.
Approach to Tasks
One characteristic of a deep learning approach was that the student was more persistent in
following up on an idea with some sustained interest before moving to another one, whereas
one using a surface approach might give up on an idea as soon as it did not work. When using
a deep approach, the student would also attempt to generate ideas on his or her own, whereas
one using a surface approach would be more reliant on external resources such as other peers
or the teacher for ideas. The student using a deep approach to learning engaged in on-line theorizing in which he or she spontaneously generated explanations for causeeffect relationships
and minitheories to account for observations of phenomena while carrying out the tasks and also
tended to think ahead and predict outcomes when performing an activity. He or she was also
more capable of attending to multiple foci, and could focus on several aspects of a phenomenon at the same time; a surface approach learner had a more limited or single focus. When a
student used a deep approach, he or she did not ignore information that was incomprehensible
but rather ruminated over this puzzlement instead of letting it slip by. Such a student also showed
a more sophisticated level of observation and was better able to discriminate finely between differences. in not only what was visible and obvious, but also inferred patterns and trends. In contrast, a learner using a surface approach noticed mainly gross macroscopic features. Finally, a
deep approach learner was also more likely to engage in talk at the conceptual, analytical, and
metaconceptual levels, beyond the procedural and observational levels that the surface approach
learner typically engaged in.

LEARNING IN SCIENCE: A COMPARISON

127

To illustrate the above-mentioned differences, consider the girls group. Mary, who used a
predominantly deep learning approach had an inquiring mind, was curious about things, and displayed a predominantly exploration mode of engagement toward phenomena (Rath & Brown,
1996) in which her primary interest was to find out about the phenomena and study their basic
properties. Bess was very dependent on external resources and often relied on Mary, other classmates, and the teacher for instructions and sources of ideas. She displayed more of a busy bee
orientation toward the activities, often walking back and forth between her group and other
groups to see what others were doing and how they were making use of the materials and apparatus provided. Her mode of engagement would typically be procedural, where she conscientiously followed or imitated others actions and carried out the given instructions as best as
she could. Dale, in contrast, was relatively passive in task engagement, often assuming the role
of an onlooker, and spoke little. She would typically sit back and watch Mary and Bess perform
the activity first before trying them out herself later, and let her partners perform the more challenging tasks which required more thinking and active engagement. Her mode of engagement
typically fluctuated between procedural and that of an onlooker. Selected episodes from the
hands-on activities will be presented to illustrate the behaviors, verbal comments, and thinking
processes associated with these approaches to tasks.

Separation of SaltSand Mixture. This was a problem-solving activity in which the students had to devise a method of separating the salt from the sand. Mary first suggested using a
sifter and then looking up a dictionary to find out the chemical composition of sand and salt.
Unlike Mary, who attempted to generate her own ideas and dwell more on them, Bess kept visiting other groups to see what they were doing and tried to get ideas from them, giving up one
idea as soon as it did not work. Together with Mary and Dale, she tried using a magnet, creating static electricity by using a plastic filter funnel and a plastic spoon and rubbing them with
her jacket, heating the saltsand mixture dry, and then finally adding water to the mixture followed by draining off the sand residue and evaporating the water from the salt solution over an
alcohol flame.
Mary displayed not only a hands-on orientation to the activity, but also a minds-on approach
to the tasks, paying attention to several details. For instance, when the salt solution was boiling,
Bess seemed to have only a single focus and was fixated on the presence of salt in the solution,
making comments such as, I still see a lot of salt, and Its, like, crystallizing. Mary, on the
other hand, was better able to notice nuances in her observations as well as focus on multiple
aspects of the phenomena. For instance, she noticed the sizzling effect of adding cold water to
the hot dry mixture, the salt clumping together when water was first added to it, the salt dissolving in the hot water, the salt floating around and mixing with the bubbles when the water
was boiling, the salt settling down to the bottom of the pan when more water evaporated, the
salt grains spluttering and sticking to the edges of the pan, and the smell of salt burning when
it was dry.
One major difference between Mary and her partners was her engagement in on-line theorizing where she came up with tentative hypotheses, generated self-explanations for causeeffect relationships, and theories to account for her observations of phenomena. In this capacity,
she was able to go beyond the kind of thinking displayed by Bess and Dale. For example, when
she noticed the water sizzling and the salt gathering around the edges of the pan, she did not
simply let these observations slip by without trying to self-explain what caused this to happen.
Instead, she theorized about how heating the salt solution could have helped to separate the salt
from the water, and how adding cold water catalyzed this separation process when it sizzled.

128

CHIN AND BROWN

[Drawing on paper] These are water [molecules]. When you pour the salt into them, the
molecules combine, they stick together. . . . When you heat the water, the molecules separate. The heat acts as a reactant to separate the molecules. . . . So that the salt, the sodium chloride molecules gather round at the edges to separate from the water molecules.
And the water molecules are left in the middle. . . . Ill just label this salt water theory. . . .
When we were boiling the water and the salt was gathering around the edges, the water
in the pan was hot. And the water that I added was cold. So it made kind of like a spark
explosion reaction. Like the molecules didnt agree. . . . And it made the little sizzling
thing. And I think that it increased the process of the salt separating because it made the
molecules spin faster. . . . I was thinking that maybe if the cold water is not agreeing with
the hot water, that would speed up the process of the salt separating faster.

Although her minitheories may seem naive, they reveal how a deep approach learner might
react to ordinary, commonplace situations and perhaps see the extraordinary in the ordinary. Furthermore, Mary did not ignore statements which puzzled her. Rather, she ruminated about them
and referred to them again later on. This can be seen for example, when she suddenly commented on her understanding of how dry salt could be recovered when salt water evaporated,
saying, Now I know what she meant by evaporating the water. . . . It will evaporate and the
salt will be back several minutes after Bess had told her that the teacher had suggested this
idea to her. She had made a connection between the current situation and her previous experience at her grandparents beach house by the sea where she thought about the sun, warm ocean
waters, sandy beaches, and the thin film of salt left on the rocks.
Boiling Point Activity. The students had to plot the temperature graphs of plain ice water
and salted ice water as they were heated in a beaker over a hot plate until the water boiled. Bess
wanted to record the temperature and observations, Dale volunteered to keep time, and Mary
agreed to read the temperature. It is interesting that Bess and Dale chose to perform the less
challenging tasks when Mary gave them the option first to select what they wanted to do.
Bess: 1, 4, 5, 7 [8C] (reading the temperatures recorded for the first 2 min from her data
sheet)
Dale: 30 s.
Mary: 9 [. . . , . . .] Im hypothesizing that the first initial leap in temperature was like the
first burst, and then it would gradually go up by 2 [8C] every 30 s.
Dale: 30 s.
Mary: 11. See, its pretty evident that it is going to keep going up by 2. Although as the
hot plate gets hotter, it might go faster [. . . , . . .] Feel the heat radiating off this
[hot plate].
Bess: The ice is about half gone.
Dale: 30 s.
Mary: 12.
Bess: Okay, and ice is about half gone [. . .] more than half.
Dale: 30 s.
Mary: 16.
Bess: And the ice is almost gone now. Almost gone [writing].
Mary: Its going up a lot faster now. Its already 208C and we are not even at the 30 s
mark. 21.
Dale: 30 s.
Mary: 22.
Bess: Only like one or two ice.

LEARNING IN SCIENCE: A COMPARISON

129

Mary:
Dale:
Mary:
Bess:
Mary:

Say ice melted. Thats not even ice.

30 s.
29.
Ice gone [writing].
1008C is 212 [8F]. And 212 is boiling point. So once it reaches 100, it should be
boiling.
Dale: 30 s.
Mary: 36. Look on the bottom and see if its, like, going to be small bubbles. Are the small
bubbles forming yet?
Bess: Not yet. No.

As the temperature increased, Bess noted bubbles forming, getting bigger, and starting to
rise to the surface of the water. At 878C, Mary remarked that This is the third time in a row it
had 78C as a consecutive number [the temperature difference was 78C for the last three times].
When the temperature reached 1008C, Mary predicted that Its boiling now, so it will slow
down. When the temperature did not rise above 1008C upon further heating, both Bess and
Mary were puzzled by this counterintuitive observation. This prompted Mary to generate an explanation for her observation, saying, Because if its boiling, its already at that temperature.
Mary told Bess that she thought salt water would get hotter than plain water because salt is a
heat conductor and that it wouldnt go above 1058C. . . . Because by then the water would
have all evaporated. This was another example of on-line theorizing where she went beyond
thinking at the observational level to a more conceptual level.
Again, in this activity, there were obvious differences in the approach to tasks displayed by
the three girls, particularly regarding the nature of the comments made and the different levels
of noticing. Dale was extremely passive, did not participate hands-on, and hardly said anything
other than 30 s when keeping time. In this extreme case, her orientation to the activity was
purely routine, mechanistic, and procedural, and she displayed a very low level of cognitive engagement. She only had a single focusto keep time. Bess was more observant. Her primary
concerns were keeping track of when and how much ice had melted, and noting when the bubbles appeared and subsequent changes in bubble formation during the heating process. Most of
Besss comments could at best be described as observational, which referred to physical phenomena that she noticed.
Mary showed a deeper level of observation, used trends in temperature changes to predict
the subsequent temperatures and noted the appearance of bubbles, the formation of steam, and
increased evaporation of water at the boiling point. Her observations included not only gross
macroscopic features (such as bubbles and steam), but also inferred patterns and trends (e.g.,
temperature differences). She also made the most use of her senses. This was evident when she
placed her hand above the hot plate to feel the heat radiated, tasted her finger which she had
dipped into the salted ice, and commented on the burning smell. In addition, she often thought
ahead and anticipated possible outcomes such as the temperature at which water would boil and
the effect of salt on the heating of water. She also attempted to explain why the temperature of
water remained constant at the boiling point and why salt water would boil at a higher temperature than plain water, although her reasons were not completely normative. Nevertheless, it was
her ability to concentrate on multiple aspects of the phenomenon and her spontaneous on-line
theorizing that made her different from the other two girls. Her comments were not only pitched
at the observational level, but went beyond this to include those targeted at the conceptual and
analytical levels. These conceptual/analytical comments were attempts to integrate current information with prior knowledge, or speculations and explanations as to why an event might
occur.

130

CHIN AND BROWN

Discussion, Implications, and Conclusions

In this study, meaningful perspectives (with respect to generative thinking, nature of explanations, asking questions, metacognitive activity, and approach to tasks) were developed as
part of a framework to describe the characteristics associated with a deep and a surface approach
to learning science and compare the differences between them. In summary, when students used
a deep approach, they generated their ideas more spontaneously and their responses were more
precise and elaborate. They also gave microscopic theory-like explanations which described
nonobservable entities and causeeffect relationships or referred to relevant daily life experiences in trying to understand a phenomenon. Explanations associated with a surface learning
approach tended to be reformulations of the question, a black box variety which did not refer
to a mechanism, or macroscopic descriptions which referred only to what was visible. Questions associated with a deep approach focused on explanations and causes, predictions, or resolving discrepancies in knowledge, and had a greater potential to lead to an advancement in
conceptual understanding. Questions associated with a surface approach referred to more basic
factual or procedural information. When students used a deep approach, they also displayed
more cognitive appraisal and regulatory control of the learning process through ongoing reflective thinking. In their approach to tasks, they tended to think ahead and predict outcomes,
showed a more sophisticated level of observation, and were more likely to engage in talk at a
higher cognitive level beyond the procedural and observational levels that learners using a surface approach typically engaged in. To facilitate the generation of ideas and expand on their
thinking, the students using a deep approach used strategies such as generating mental images,
creating analogies, hypothesizing, constructing thought experiments and predicting possible outcomes, giving self-explanations and constructing theories, invoking personal experiences and
prior knowledge and applying them to new situations, thinking of specific examples, and asking questions.
In this study, the nature of the students learning approach was studied at a more fine-grain
level involving analysis of their thinking processes, and cognitive and metacognitive strategies.
While Martons (1983) conceptualization of the deep and surface approaches (in noncontentspecific domains) was used as an overarching framework, the identification and articulation of
finer differences between these approaches has provided a more elaborate description and complete picture of what is meant by deep and surface learning approaches in the context of science
learning. This goes beyond earlier descriptions which were more general, broadly defined, less
domain specific, and more pertinent to reading of texts. Moreover, it takes into account those
learning strategies that are used when students are engaged in hands-on investigations which are
an integral part of learning in science. In this capacity, it has expanded on the previous framework for reading text to include hands-on experiments in inquiry settings, problem solving, and
reasoning. The importance of this study also lies in the creation of a terminology that makes it
possible to describe and discuss in an explicit and specific way, the differences between a deep
and surface learning approach in science. A situated account of how these differences could be
manifested and several concrete examples were given.
This study has expanded the knowledge base about students learning approaches by including more ecologically valid data. The data were derived not only from interviews with students and written tasks, but also from the students engagement with actual hands-on activities
which were carried out during regular science classes. As such, they were situated in the authentic context of a naturalistic setting (Brown, Collins, & Duguid, 1989; Hennessey, 1993), as
opposed to a setting outside the classroom context or involving ad hoc tasks. That is, the students were actually observed performing the science tasks during normal instruction, and the

LEARNING IN SCIENCE: A COMPARISON

131

analysis of their learning approaches and use of strategies was based on this on-site performance,
together with information from follow-up interviews. This differs from earlier research which
typically used only students self-reports from interviews and questionnaires to obtain information about their approaches to learning. This procedure has yielded insights which might not
have been obtained from the use of self-report data alone. Furthermore, unlike most previous
research on students learning approaches where the student typically interacted only with the
researcher in a one-to-one individual interview (e.g., DallAlba, 1986), data from this study also
pertained to science activities in which the students interacted with their peers in a group setting. This allowed direct comparisons to be made of the students and their learning approaches. A combination of detailed observation, video recording of hands-on activities, interviews,
and written tasks also complemented each other in providing a more holistic picture of the students learning approaches.
Using a deep approach can sometimes lead a learner to form idiosyncratic ideas which are
scientifically incorrect. For example, during the separation of saltsand activity, Mary constructed her saltwater theory to explain how she thought heat would facilitate the separation
of salt from the water molecules. Rick used noncanonical ideas to explain the melting point depression and boiling point elevation of salted water. Despite this, such arguments were considered to be deep and assigned to a high level. Unlike the other students, Mary and Rick had the
special quality of being able to theorize and generate their own ideas spontaneously and could
elaborate on them in more depth. When students use a deep approach, they do not necessarily
have to come up with correct ideas all the time. One reason could be that because of their deep
thinking, they do not always simply accept standard scientific answers unthinkingly and can thus
generate alternative ideas. With regard to this, in the history of science, great scientists of the
past (e.g., Aristotle, Lamarck) postulated theories which were subsequently found to be inadequate. Newton also held inappropriate conceptions before he wrote the Principia (Steinberg,
Brown, & Clement, 1990). This does not make them any less of a deep thinker. Thus, it may be
that although active on-line theorizing could potentially lead students to generate nonscientific
ideas, other deep processing strategies could also help them detect conflicts between the evolving ideas and what is correct. On the whole, this may have a compensatory effect and lead to
improved learning in the long run.
Although the students who used more of a deep approach shared some similarities with the
other students in that they also had alternative conceptions about phenomena and limited content knowledge in some areas, what was striking about them was the difference in the way they
approached a task or problem. This was evident, for example, in differences in their generative
capacity, the nature of the explanations they gave, and their use of more sophisticated metacognitive strategies. Some of these differences are reminiscent of the findings obtained by other
researchers. For example, Chi, Bassok, Lewis, Reimann, and Glaser (1989) found that good
students (i.e., successful problem solvers) generated many self-explanations while studying
worked-out examples of mechanics problems from text and during problem solving. However,
poor students did not generate sufficient self-explanations and monitored their learning inaccurately. Also, Ertmer and Newby (1996) pointed out that expert learners are more strategic,
self-regulated, and reflective than novice learners.
Implications
One finding that emerged from this study was the recognition of a variety of types of explanation (reformulations of the question, black box, macroscopic, and microscopic) that students typically give. This typology which provides a framework within which different levels

132

CHIN AND BROWN

of student generated explanations can be accommodated could help teachers to recognize the
type and quality of explanations that their students give in relation to their learning approaches, as well as the attributes of effective and not so effective explanations. A taxonomy of question types which classifies students questions according to different conceptual levels was also
developed and shown how they could be related to learning approaches. Such a classification
could help teachers to recognize the learning approaches adopted by their students and to encourage their students to ask questions at a deeper level.
In looking closely at individual students, it was evident that even among the students who
used a deep approach, they showed a tendency to be deep in different ways. For example, Rick
tended to think of specific hypothetical examples and seemed to have a theoretical approach,
whereas Mary was more inclined to draw on personal experiences in daily life. This can be seen,
for example, in their explanations pertaining to the chromatography of pen ink and the effect of
salt on the boiling point of water. The students were deep in different ways reflecting individual differences even among those considered as deep approach learners. What this suggests is
that students may be more prone to deeper thinking in some dimensions and contexts than in
others. By being aware of the multidimensionality of this, teachers could help students deepen
their thinking starting from contexts and dimensions where they already show some depth. By
the same token, when students adopt a surface approach, this also seems to be manifested in different ways. For example, in adopting a busy bee approach to tasks, Bess was diligent in following procedures but seemed to be externally regulated in that she was dependent on others
for sources for ideas. Dale, who often assumed the role of onlooker, was typically apathetic in
task engagement, preferring to watch others perform an activity first before trying it out herself.
These approaches to learning seem related to Vermunts (1998) description of four learning
styles (viz., meaning directed, application directed, reproduction directed, and undirected) and suggest that students learning approaches are more differentiated than can be denoted with a bipolar
deep-surface distinction. For instance, the approaches adopted by Rick, Mary, Bess, and Dale seem
in some ways to correspond respectively to the meaning directed, application directed, reproduction directed, and undirected styles. As Vermunt pointed out, these bipolar [deep vs. surface] descriptions insufficiently cover the empirical variation found among students in these learning components, and underestimate the complexity of study behavior in a real study context (p. 167).
During the group work activities, the roles assumed by the students were found to some extent to follow their stereotypes even though the students were not assigned by their teacher to
any specific roles in their groups. Rather, the roles adopted were self-selected. The students had
also not been informed about how they were representative of learners using different learning
approaches. In the girls group, for example, the three girls seemed to work reasonably well together, each comfortable with the role she assumed which was in accordance with her natural
disposition, personality, and learning approach. Dale seemed content playing a passive role as
mainly an onlooker and letting the other two girls lead the activity. Mary, being a more active
and self-initiated person, also seemed quite satisfied in that she was able to initiate the tasks,
lead the others in discussion, and dominate most of the thinking. This natural state of affairs
may reinforce and perpetuate individual differences in students learning approaches and allow
the gap between them to widen even further. Teachers need to think about ways to support students learning activities (e.g., regarding group composition and roles, instructional design, task
structure) without accentuating differences that may disadvantage some students.
Another noteworthy point about the students in this study who used more of a deep approach was that although they were naturally more inclined than the other students to use deep
processing strategies (e.g., predicting, self-explaining), this did not always seem spontaneous
but was manifested only upon further probing or prompting during the interviews and when the

LEARNING IN SCIENCE: A COMPARISON

133

students were specifically asked to explain a phenomenon. For example, during the activity on
chromatography of pen inks, both Rick and Mary did not think seem to much about the meaning of the Rf, as their teacher had not emphasized this. Instead, they were more concerned with
the procedure involved in determining the Rf value and getting this task done. However, when
they were probed during the interview, they were perhaps stimulated to use deeper strategies to
extend their thinking about the Rf and were able to generate plausible ideas about the significance of this value. This suggests that these deep thinking processes are sometimes latent in students and are manifested only under optimal conditions such as through another persons scaffolding, prompting, or probing as a result of the interaction between the students dispositions
and situational circumstances. Also, in students using more of a surface approach, the deep
strategies seem more deeply embedded, whereas in those using more of a deep approach, the
deep processing strategies seem to surface more readily. Thus, if left to their own devices, students may sometimes find it difficult to develop deep processing strategies on their own and
may not use deep processing strategies as often as teachers would like them to. If deep strategies are considered desirable in facilitating learning, then this suggests that teachers could scaffold students thinking to encourage their students to adopt deep processing strategies.
Future Research
How might teachers scaffold students thinking to encourage deep processing? Future research could be directed at devising instructional strategies that attempt to induce a deep approach to learning and then studying the effects of such strategies. For example, one might study
how the teacher can scaffold student discourse and provide a task structure to engage the students in deep processing strategies that would lead to higher-level cognitive and metacognitive
talk. Activities which involve questioning, explaining, and predicting could be used to bolster a
discussion and the students could be explicitly required to do these rather than hope that they
would just occur spontaneously. The teacher could ask students to write their predictions and
formulate questions before performing the activity to help them direct their own inquiry and use
these questions as a springboard for investigation and discussion. The students could also think
about particular questions as they work on their tasks. Then, at the end of the activity, the students can write questions reflecting what they wondered about, what had puzzled them, or what
they needed to know to obtain a fuller understanding of the material they have just studied.
To encourage higher levels of constructive activity and foster deep processing strategies in
students, the teacher could also ask students to explain how and why certain phenomena occur
and to provide explanations to their own questions. When students attempt to account for observed phenomena, they might generate minitheories. Over time and with continual practice,
this on-line theorizing may become more of a habit and thus occur more spontaneously in students. Students could also be told about what counts as an explanation and the attributes of a
good explanation, and be encouraged to move to a deeper level of explanation. To encourage
students to generate their own questions and explanations and guide stimulation of deeper processing strategies which students may not activate on their own, the teacher could provide selfmonitoring and reflection prompts. Such prompts could provide conceptual guidance and the
impetus for explanation, and encourage reflection at a level that students do not generally consider. They could also elicit elaboration of students ideas and encourage integration (Davis,
1996). While some students may not naturally engage in deep processing strategies on their own,
the instructional scaffolding strategies and task prompts might serve a cueing purpose to enhance the students cognitive and metacognitive capabilities and lead to the behaviors associated with a deep learning approach.

134

CHIN AND BROWN

Although the above suggestions may appear promising, there is a caveat. It is a common
assumption that if one makes students perform the same kinds of activities, this would mean
that they, too, would use a deep approach. However, this kind of logical reasoning may not always lead to the expected results when applied to human behavior. For example, Marton and
Saljo (1976) found that their attempts to induce a deep approach through forcing students to answer certain types of questions while reading text resulted in a technification of the learning
process. This demand structure of the learning situation led to an extreme form of surface learning instead, in which students simply complied with the demands without engaging in deep
thinking. Also, a study by Arzi and White (1986) found that after students were trained in asking good reflective questions in an attempt to actively engage them in thinking, some of the
questions that students asked were simply a glib modification of versions of standard question
stems which had been introduced to them. These questions, however, were generated in the context of reviewing for a test. On the other hand, King (1994) found that using prompts to teach
students how to question and explain during science lessons led to more complex knowledge
construction and enhanced learning in the students.
It is also possible that providing students with a task structure may benefit students using
different approaches to different extents. For example, learners who already know how to use
deep processing strategies may find them irrelevant or stifling because they may interrupt independent reasoning and distract them. How different scaffolding structures could specifically be
embedded within an activity and how different students react to and use them could also be the
focus of further research.
In conclusion, this study has provided further information on the finer differences associated with deep versus surface learning approaches to learning science. This information could help
teachers to become more perceptive towards their students learning approaches by becoming
more aware of the subtleties in students thinking.
The authors thank the students and teacher who participated in this study, and the two anonymous reviewers for their helpful comments on an earlier draft of the manuscript.

References
Ames, C., & Archer, J. (1988). Achievement goals in the classroom: Students learning
strategies and motivation processes. Journal of Educational Psychology, 80, 260267.
Arzi, H.J., & White, R.T. (1986). Questions on students questions. Research in Science Education, 16, 8291.
Ausubel, D.P. (1968). Educational psychology: A cognitive view. New York: Holt, Rinehart
& Winston.
Baird, J.R., Fensham, P.J., Gunstone, R.F., & White, R.T. (1991). The importance of reflection in improving science teaching and learning. Journal of Research in Science Teaching,
28, 163182.
Baird, J.R., & White, R.T. (1982a). A case study of learning styles in biology. European
Journal of Science Education, 4, 325337.
Baird, J.R., & White, R.T. (1982b). Promoting self-control of learning. Instructional Science, 11, 227247.
Biddulph, F., & Osborne, R. (1982). Some issues relating to childrens questions and explanations. LISP(P) Working paper no. 106, University of Waikato, New Zealand.
Biggs, J. (1987). Student approaches to learning and studying. Melbourne: Australian Council for Educational Research.

LEARNING IN SCIENCE: A COMPARISON

135

Biggs, J. (1988). Approaches to learning and to essay writing. In R.R. Schmeck (Ed.),
Learning strategies and learning styles (pp. 185228). New York: Plenum.
Biggs, J. (1994). Approaches to learning: Nature and measurement of. In T. Husen and T.N.
Postlethwaite (Eds.), The international encyclopedia of education (2nd ed., Vol. 1) (pp. 319
322). Oxford: Pergamon.
Bogdan, R.C., & Biklen, S.K. (1992). Qualitative research for education. Boston: Allyn &
Bacon.
BouJaoude, S.B. (1992). The relationship between students learning strategies and the
change in their misunderstandings during a high school chemistry course. Journal of Research
in Science Teaching, 29, 689 699.
Brown, A.L. (1978). Knowing when, where, and how to remember: A problem of metacognition. In R. Glaser (Ed.), Advances in instructional psychology (Vol. 1) (pp. 77157). Hillsdale, NJ: Erlbaum.
Brown, D.E. (1993). Refocusing core intuitions: A concretizing role for analogy in conceptual change. Journal of Research in Science Teaching, 30, 12731290.
Brown, D.E., & Clement, J. (1989). Overcoming misconceptions via analogical reasoning:
Abstract transfer versus explanatory model construction. Instructional Science, 18, 237261.
Brown, J.S., Collins, A., & Duguid, P. (1989). Situated cognition and the culture of learning. Educational Researcher, 18, 3242.
Cavallo, A.M.N., & Schafer, L.E. (1994). Relationships between students meaningful
learning orientation and their understanding of genetics topics. Journal of Research in Science
Teaching, 31, 393418.
Chi, M.T.H., Bassok, M., Lewis, M.W., Reimann, P., & Glaser, R. (1989). Self-explanations: How students study and use examples in learning to solve problems. Cognitive Science,
13, 145182.
Chi, M.T.H., de Leeuw, N., Chiu, M.H., & Lavancher, C. (1994). Eliciting self-explanations
improves understanding. Cognitive Science, 18, 439477.
Clement, J. (1988). Observed methods for generating analogies in scientific problem solving. Cognitive Science, 12, 563586.
DallAlba, G. (1986). Learning strategies and the learners approach to a problem solving
task. Research in Science Education, 16, 1120.
Davis, E. (1996, April). Metacognitive scaffolding to foster scientific explanations. Paper
presented at the meeting of the American Educational Research Association, New York.
De Grave, W.S., Boshuizen, H.P., & Schmidt, H.G. (1996). Problem based learning: Cognitive and metacognitive processes during problem analysis. Instructional Science, 24, 321341.
Donaldson, M. (1978). Childrens minds. London: Falmer.
Driver, R., Asoko, H., Leach, J., Mortimer, E., & Scott, P. (1994). Constructing scientific
knowledge in the classroom. Educational Researcher, 23, 512.
Dweck, C., & Elliot, E.S. (1983). Achievement motivation. In E.M. Hetherington (Ed.),
Handbook of child psychology: Vol. 4. Socialization, personality, and social development
(pp. 643691). New York: Wiley.
Edmondson, K.M., & Novak, J.D. (1993). The interplay of scientific epistemological views,
learning strategies, and attitudes of college students. Journal of Research in Science Teaching,
30, 547559.
Elstgeest, J. (1985). The right question at the right time. In W. Harlen (Ed.), Primary science: Taking the plunge. London: Heinemann.
Entwistle, N.J. (1981). Styles of learning and teaching. Chichester: Wiley.
Entwistle, N.J., & Ramsden, P. (1983). Understanding student learning. London: Croom Helm.

136

CHIN AND BROWN

Ericsson, K.A., & Simon, H.A. (1993). Protocol analysis (rev. ed.). Cambridge, MA: MIT
Press.
Ertmer, P.A., & Newby, T.J. (1996). The expert learner: Strategic, self-regulated, and reflective. Instructional Science, 24, 124.
Garner, R. (1988). Verbal-report data on cognitive and metacognitive strategies. In C.E.
Weinstein, E.T. Goetz, & P.A. Alexander (Eds.), Learning and study strategies (pp. 6376). New
York: Academic.
Gilbert, J.K., Boulter, C., & Rutherford, M. (1998a). Models in explanations, part 2: Whose
voice? Whose ears? International Journal of Science Education, 20, 187203.
Gilbert, J.K., Boulter, C., & Rutherford, M. (1998b). Models in explanations, part 1: Horses for courses? International Journal of Science Education, 20, 8397.
Glaser, B.G., & Strauss, A.L. (1967). The discovery of grounded theory: Strategies for qualitative research. New York: Aldine de Gruyter.
Grosslight, L., Unger, C., & Jay, E. (1991). Understanding models and their use in science:
Conceptions of middle and high school students and experts. Journal of Research in Science
Teaching, 28, 799822.
Gunstone, R.F. (1994). The importance of specific science content in the enhancement of
metacognition. In P. Fensham, R. Gunstone, & R. White (Eds.), The content of science: A constructivist approach to its teaching and learning (pp. 131146). London: Falmer.
Hegarty-Hazel, E., & Prosser, M. (1991a). Relationship between students conceptual
knowledge and study strategiespart 1: Student learning in physics. International Journal of Science Education, 13, 303312.
Hegarty-Hazel, E., & Prosser, M. (1991b). Relationship between students conceptual
knowledge and study strategiespart 2: Student learning in biology. International Journal of
Science Education, 13, 421429.
Hennessey, M.G. (1993, April). Students ideas about their conceptualization: Their elicitation through instruction. Paper presented at the annual meeting of the National Association for
Research in Science Teaching, Atlanta, GA.
Hennessey, M.G. & Beeth, M.E. (1993, April). Students reflective thoughts about science
content: A relationship to conceptual change learning. Paper presented at the American Educational Research Association annual meeting, Atlanta, GA.
Hennessey, S. (1993). Situated cognition and cognitive apprenticeship: Implications for
classroom learning. Studies in Science Education, 22, 141.
Hogan, K. (1999). Relating students personal frameworks for science learing to their cognition in collaborative contexts. Science Education, 83, 132.
Horwood, R.H. (1988). Explanation and description in science teaching. Science Education,
72, 4149.
King, A. (1994). Guiding knowledge construction in the classroom: Effects of teaching children how to question and how to explain. American Educational Research Journal, 31, 338
368.
Laurillard, D.M. (1978). A study of the relationship between some of the cognitive and contextual factors involved in student learning. Unpublished doctoral dissertaion, University of Surrey, UK.
Lee, O., & Anderson, C.W. (1993). Task engagement and conceptual change in middle
school science classrooms. American Educational Research Journal, 30, 586610.
Lee, O., & Brophy, J. (1996). Motivational patterns observed in sixth-grade science classrooms. Journal of Research in Science Teaching, 33, 303318.
Lincoln, Y.S., & Guba, E.G. (1985). Naturalistic inquiry. Newbury Park, CA: Sage.

LEARNING IN SCIENCE: A COMPARISON

137

Marton, F. (1983). Beyond individual differences. Educational Psychology, 3, 289303.

Marton, F., & Saljo, R. (1976). On qualitative differences in learning. I: Outcome and
process. British Journal of Educational Psychology, 46, 411.
Maskill, R., & Pedrosa de Jesus, H. (1997). Pupils questions, alternative frameworks and
the design of science teaching. International Journal of Science Education, 19, 781799.
Meece, J., Blumenfeld, P.C., & Hoyle, R. (1988). Students goal orientations and cognitive
engagement in classroom activities. Journal of Educational Psychology, 80, 514523.
Metz, K. (1991). Development of explanation: Incremental and fundamental change in childrens physics knowledge. Journal of Research in Science Teaching, 28, 785798.
Nicholls, J.G. (1984). Achievement motivation: Conceptions of ability, subjective experience, task choice, and performance. Psychological Review, 91, 328346.
Nolen, S.B. (1988). Reasons for studying: Motivational orientations and study strategies.
Cognition and Instruction, 5, 269287.
Nolen, S.B., & Haladyna, T.M. (1990). Motivation and studying in high school science.
Journal of Research in Science Teaching, 27, 115126.
Novak, J.D. (1988). Learning science and the science of learning. Studies in Science Education, 15, 77101.
Osborne, R.J., & Wittrock, M.C. (1983). Learning science: A generative process. Science
Education, 67, 489508.
Osborne, R., & Wittrock, M. (1985). The generative learning model and its implications for
science education. Studies in Science Education, 12, 5987.
Peterson, P.L., Swing, S.R., Braverman, M.T., & Buss, R. (1982). Students aptitudes and
their reports of cognitive processes during direct instruction. Journal of Educational Psychology, 74, 535547.
Pizzini, E.L., & Shepardson, D.P. (1991). Student questioning in the presence of the teacher
during problem solving in science. School Science and Mathematics, 91, 348352.
Ramsden, P. (1988). Context and strategy: Situational influences on learning. In R.R.
Schmeck (Ed.), Learning strategies and learning styles (pp. 159184). New York: Plenum.
Rath, A., & Brown, D.E. (1996). Modes of engagement in science inquiry: A microanalysis of elementary students orientations toward phenomena at a science summer camp. Journal
of Research in Science Teaching, 33, 10831097.
Roth, K.J. (1990). Developing meaningful conceptual understanding in science. In B.F.
Jones & L. Idol (Eds.), Dimensions of thinking and cognitive instruction. Hillsdale, NJ: Erlbaum.
Roth, W.M., & Roychoudhury, A. (1993). The nature of scientific knowledge, knowing and
learning: The perspectives of four physics students. International Journal of Science Education,
15, 2744.
Roth, W.M., & Roychoudhury, A. (1994). Physics students epistemologies and views about
knowing and learning. Journal of Research in Science Teaching, 31, 530.
Scardamalia, M., & Bereiter, C. (1992). Text-based and knowledge-based questioning by
children. Cognition and Instruction, 9, 177199.
Schauble, L., Klopfer, L.E., & Raghavan, K. (1991). Students transition from an engineering model to a science model of experimentation. Journal of Research in Science Teaching,
28, 859882.
Schmeck, R.R. (1988). An introduction to strategies and styles of learning. In R.R. Schmeck
(Ed.), Learning strategies and learning styles (pp. 319). New York: Plenum.
Shodell, M. (1995). The question-driven classroom. The American Biology Teacher, 57,
278281.

138

CHIN AND BROWN

Solomon, J. (1986). Childrens explanations. Oxford Review of Education, 12, 4151.

Songer, N.B., & Linn, M.C. (1991). How do students views of science influence knowledge integration? Journal of Research in Science Teaching, 28, 761784.
Stake, R. (1995). The art of case study research. Thousand Oaks, CA: Sage.
Steinberg, M.S., Brown, D.E., & Clement, J. (1990). Genius is not immune to persistent
misconceptions: Conceptual difficulties impeding Isaac Newton and contemporary physics students. International Journal of Science Education, 12, 265273.
Tamir, P., & Zohar, A. (1991). Anthropomorphism and teleology in reasoning about biological phenomena. Science Education, 75, 5767.
Tisher, R.P. (1977). Practical insights gained from Australian research on teaching. Australian Science Teachers Journal, 23, 99 104.
Tsai, C.C. (1998). An analysis of scientific epistemological beliefs and learning orientations
of Taiwanese eighth graders. Science Education, 82, 473489.
Vermunt, J.D. (1998). The regulation of constructive learning processes. British Journal of
Educational Psychology, 68, 149171.
Watts, M., Gould, G., & Alsop, S. (1997). Questions of understanding: Categorising pupils
question in science. School Science Review, 79, 5763.
White, R.T., & Gunstone, R.F. (1989). Metalearning and conceptual change. International
Journal of Science Education, 11, 577586.
White, R.T., & Gunstone, R.F. (1992). Probing understanding. London: Falmer.
Wittrock, M.C. (1994). Generative science teaching. In P. Fensham, R. Gunstone, & R.
White (Eds.), The content of science: A constructivist approach to its teaching and learning
(pp. 2938). London: Falmer.
Wong, E.D. (1993a). Self-generated analogies as a tool for constructing and evaluating explanations of scientific phenomena. Journal of Research in Science Teaching, 30, 367380.
Wong, E.D. (1993b). Understanding the generative capacity of analogies as a tool for explanation. Journal of Research in Science Teaching, 30, 12591272.
Woodruff, E., & Meyer, K. (1997). Explanations from intra- and inter-group discourse: Students building knowledge in the science classroom. Research in Science Education, 27, 25 39.
Zoller, U. (1987). The fostering of question-asking capability: A meaningful aspect of problem-solving in chemistry. Journal of Chemical Education, 64, 510512.