2022 Conceptual Change Uin Chemistry
2022 Conceptual Change Uin Chemistry
2022 Conceptual Change Uin Chemistry
Abstract
Practice-oriented studies on conceptual change in chemistry education remain relatively scarce. In agreement with the
constructivist approach, in which learners build their own cognitive structure, many results have shown that learning through
modelling positively contributes to conceptual adaptation. From this point of view, this study presents a cross-sectional study
of secondary school learners’ schematic modelling abilities in the upper secondary school. The schematic modelling skills and
competences of 216 students (15–18 years old, 10th–12th grade), ranging from purely macroscopic conceptions to an adequate
interplay between the three levels of Johnstone’s triangle, have been assessed within selected contextualized situations. The
data were collected using a three-part instrument, namely the analysis of the information contained in a graphical scheme,
the perception of the role and relevance of schematic modelling, and the autonomous production of schematic modelling. The
collected data show that 12th-grade students have a significantly higher capability of analysing the information of a scheme
compared to 10th and 11th grades. Moreover, autonomous schematic modelling skills follow the same trend. The data also
provide some evidence that students with better model analysis competences design higher quality autonomous models at
the submicroscopic level.
Key words: conceptual change, schematic models, Johnstone’s triangle, macroscopic representations, submicroscopic repre-
sentations
Résumé
Les études axées sur la pratique qui portent sur le changement conceptuel dans l’enseignement de la chimie demeurent
relativement rares. Tel que le veut l’approche constructiviste, dans laquelle les apprenants construisent leur propre structure
cognitive, de nombreux résultats ont montré que l’apprentissage par la modélisation contribue positivement à l’adaptation
conceptuelle. De ce point de vue, nous présentons dans cet article une étude transversale des capacités de modélisation sché-
matique des apprenants de deuxième cycle du secondaire. Les aptitudes et compétences en modélisation schématique, allant
de conceptions purement macroscopiques à une interaction adéquate entre les trois niveaux du triangle de Johnstone, ont
été évaluées chez 216 élèves (de 15 à 18 ans, de la 10e à la 12e année) dans des situations contextualisées sélectionnées. Les
données ont été recueillies à l’aide d’un instrument en trois volets, à savoir l’analyse des informations contenues dans un
schéma graphique, la perception du rôle et de la pertinence de la modélisation schématique, et la production autonome de
modélisation schématique. Les données recueillies montrent que les élèves de 12e année ont une capacité significativement
plus élevée d’analyser les informations d’un schéma comparativement à ceux de 10e et de 11e année. De plus, les compétences
de modélisation schématique autonome suivent la même tendance. Les données tendent également à montrer que les élèves
ayant de meilleures compétences en matière d’analyse de modèles conçoivent de manière autonome des modèles de meilleure
qualité à l’échelle inframicroscopique. [Traduit par la Rédaction]
Mots-clés : changement conceptuel, modèles schématiques, triangle de Johnstone, représentations macroscopiques, représen-
tations submicroscopiques
Introduction
issues, subject to adaptation.1 Confronted to a particular sit-
A learning process results from a sequence of steps rang- uation, the learners mobilize representation systems, men-
ing from the simplest to the most complex level, in which the tal plans and a series of actions, questions, and meanings.2,3
learner is guided through a reasoning pathway. The concep- Read detailed the construction process of a new cognitive
tual change framework considers constructed mental mod- structure or the modification of an existing structure.4 He
els at a given educational stage as temporary, intermediate developed a conceptual change model related to the four
conditions already identified by Posner et al.: (i) the learner students. The lack of such competences may impede their un-
is not satisfied with his or her existing conceptions, (ii) the derstanding during the learning process. The interpretation
new conception is understandable, (iii) the new conception and production of such schematic models thus involve mo-
looks plausible, and (iv) the new conception is expected to be bilizing the three levels described by Johnstone’s triangle— —
fruitful for further work.5 macroscopic, submicroscopic, and symbolic.14 The students
Conceptual change provides a framework to describe and are required to also relate these different kinds of informa-
interpret an adaptation process during which the student is tion across scales. The explanatory role of the submicroscopic
faced with several situations that require to generate and test level, which is focusing on unseen particles and their inter-
alternative propositions.6 Although Posner5 mainly insisted actions, is instrumental.15
on the role of the cognitive conflict to promote conceptual Working with such models involves both the interpreta-
change, focusing on the epistemological dimension, educa- tion of existing schematic models met in textbooks, which
tion research since then has also identified other important we will call reference schematic models, and the autonomous
aspects, in particular the ontological (the nature of the in- design of models to interpret specific scientific situations
vestigated concepts) and affective ones, which contribute to (Fig. 1). Both aspects have been assessed in this work in a
progressively replace alternative conceptions by scientifically cross-sectional fashion. 216 learners at three stages of the
recognized ones.7 In the end, a progressive adjustment leads upper secondary school (15–18 years old, 10th–12th grade)
to a significant conceptual change.6,8–9 The learners must were asked to answer questions on reference schematic mod-
succeed in using appropriate conceptions in different situa- els chosen from textbooks used in the French- and German-
tions. This means that the state of a particular concept can speaking regions of Belgium (Part A of this work). In Part B,
change according to various conditions. they were invited to assess the relevance of schematic mod-
Tackling with models, and especially designing models au- els for their learning of scientific content. In Part C, they were
tonomously in a given learning situation, contributes to de- assigned the task of designing personal schematic models of
velop students’ abilities to shape their own thinking frame phenomena discussed in the upper secondary school curric-
by supporting reasoning, creativity, and decision making.1 ula. Particular attention has been devoted to the comparison
Concepts are constructed and used according to several in- between the three grades (10th to 12th).
terconnected components: representation systems (natural
language, graphical language, mathematics, etc.) and men-
Analysing the information of a schematic
tal plans (properties of the concepts, relations between them,
operations involving the concepts, and situation-adapted and model
theory-based models).10 To transmit chemical knowledge, textbooks use schematic
Analysing students’ modelling competences in specific sit- graphical representations of scientific phenomena. There-
uations allows identifying which obstacles the learners meet fore, representation skills become a required ability in learn-
in the elaboration of their knowledge as well as how possible ing chemistry.16 Schematic models in science significantly
misconceptions may arise. This forms a basis for a more fo- contribute to the construction of knowledge. They are de-
cused guidance of the students to overcome these difficulties. signed in textbooks to optimize their representational con-
The didactic, epistemological, and psychological dimensions tent and may also offer some kind of synoptic conceptual map
must be considered in this analysis. promoting cognitive integration.
As noted by Kermen and Méheut11 based on previous works Research has shown that students have difficulties inter-
by Tiberghien et al.12 and Walliser,13 “[t]he model mediates preting and using representations of scientific concepts.17,18
between the theoretical field which it interprets and the em- Abd-El-Khalick et al. showed that textbooks have a large im-
pirical field which it formalises”. In the present communica- pact on students’ conceptions and help them to develop con-
tion, we investigate a kind of pedagogical tool very often used ceptions about the nature of science.18
in teaching and learning context and which we call schematic Analysing a schematic graphical representation does not
models, because they share common features with the sci- only require to understand the scientific concepts but also
entific models defined above. These are graphical drawings the ability to interpret the language of its content. Translat-
presented in a schematic manner, thus voluntarily simpli- ing the content of a chemical representation implies thinking
fied to eliminate any aspect considered as contingent, of phe- about a phenomenon on three different levels.14,19,20 How-
nomena (chemical in our case, e.g., the dissolution of a salt ever, according to Cheng and Gilbert, students have diffi-
in water), provided with explanatory annotations (legend of culties to make links between the chemical phenomena on
symbols and chemical equation). These representations com- the macroscopic scale, the usual chemical symbolic repre-
bine a descriptive dimension insofar as the main aspects of sentations, and the concepts relevant to the submicroscopic
the experimental situation are schematized (beaker, precipi- scale.21
tate, etc.) and an explanatory dimension, ideally at the sub-
microscopic level of the particles involved (atoms, molecules, Autonomous schematic modelling
ions, and electrons), insofar as the behaviour (motions and in- The process of modelling is widespread in teaching ap-
teractions) of these particles is suggested. Scale distortions—
— proaches and in general human interaction with the world.
atoms are not represented at the same scale as, for exam- When facing a new situation in everyday life, students’ most
ple, the beaker— —are inevitable and often implicit, requiring common reaction is to try to find a prototype from their
specific reading and interpretation skills on the side of the own conceptual field. This prototype can then be seen as a
Fig. 1. The two investigated didactic activities correlated to schematic models. [Colour online.]
first model of the considered phenomenon with, however, correlation between a high level of understanding of basic
some limitations. Models are indeed defined by their spe- chemical concepts and representational competences.17
cific content and by the way they are elaborated or struc-
tured. An interesting and fruitful distinction has been intro- Research motivation and research questions
duced by Gouvea and Passmore, who emphasize that models This study is part of a project that aims at improving the
are both “models of” (some object and some phenomenon) impact of chemistry learning through active modelling ac-
and “models [designed] for” a given aim, a perspective that tivities. This research is performed in the context of the Bel-
highlights the role of the modelling agent in defining how gian French- and German-speaking education systems. To op-
a model is built, which elements are included or excluded, timize the design of model-based learning activities, it is nec-
etc.22 In the context of science teaching, it is commonly ob- essary to first explore how students react when confronted
served that students misunderstand the exact role of models to existing representation systems and to analyse the rela-
and underestimate their limitations. For instance, they often tionship between modelling abilities and age in the upper
believe that models represent a definite truth, ignoring their secondary school. This contribution focuses on the following
adaptability.23 Moreover, the importance of the modelling aspects for students of the 10th–12th grades:
process and of an adequate analysis of the different dimen-
sions spanned by the model concept itself is not sufficiently r How do student abilities to analyse, that is, interpret
addressed in the various curricula.24,25,26 To overcome these and extract information from, reference schematic models
misconceptions, modelling approaches are developed for in- compare from the 10th to the 12th grade?
troducing various concepts associated with, e.g., chemical ki- r How do students progressively change from a macroscopic
netics, electrochemistry, and chemical equations.11,21,25,27,28 description of a phenomenon towards a model-based inter-
This active participation in modelling processes allows devel- pretation at the molecular level?
oping the ability to translate mental representation in a sci- r Is there a correlation between the ability to interpret a ref-
entific scheme and to make connections between representa- erence schematic representation and the autonomous mod-
tions and concepts. The quality of student schematic model elling skills?
productions is a measure of their understanding of chemical
phenomena.29 Cheng and Gilbert also observed that a deeper
understanding of a chemical concept following a teaching se- Methods
quence manifests itself in a significant progression of their
modelling at the submicroscopic level.21 Sample
Related studies also showed that, to explain the basic An assessment tool has been developed for Belgian French-
concepts of scientific phenomena, students need to acquire and German-speaking learners (N = 216) at three stages of the
a natural reasoning relating the submicroscopic scale to upper secondary school (15–18 years old, 10th–12th grade)
macroscopic events.28,30–32 These studies have consistently (Table 1). Requests for authorization to participate in the
demonstrated that conceptual understanding of chemical study were sent to general education secondary schools in
changes is a concern. Sim and Daniel provided evidence for a the Belgian province of Liège (in majority French-speaking
Table 1. Profile of participants in terms of upper secondary school grade and mother language.
Grade Age (years) No. of students tested Type of education
French speaking German speaking
10th 15–16 60 31 General (two or three 50 min periods of
chemistry/week)
11th 16–17 46 19 General (two or three 50 min periods of
chemistry/week)
12th 17–18 44 16 General (two or three 50 min periods of
chemistry/week)
Total 150 66
but also including the smaller size German-speaking region). perceive schematic models for their learning process. A cross-
A series of meetings were scheduled with the teachers in- sectional assessment of these skills and perceptions was per-
volved to explain the research aims and the study approach. formed from the 10th to the 12th grade. This also allowed
The questionnaires were originally written in French and us to evaluate the conceptual changes of the students and to
then translated into German by a native bilingual chemistry identify the obstacles met in the appropriation of the various
teacher. concepts. Due to organisation constraints, the test time was
All participants followed an advanced chemistry course limited to one lecture period of 50 min: this time limit forced
(two or three periods of 50 min/week) with similar assess- us to focus on a limited number of questions in Part A and on
ment practices. After a general scientific education course in one modelling task in Part C.
grades 7 and 8, these students had started their specific chem-
istry education in the 9th grade for which the curriculum in-
Part A: capability to analyse the information of
cludes the following topics:33,34
a schematic model
r Composition and classification of matter: elements, com- This part of the questionnaire consisted of six multiple
choice questions (MCQs): each one was associated with an
binations, metals and non-metals, metallic conductivity,
image (Supplementary 1.A). These different schematic rep-
atoms and molecules, atomic structure (protons, neutrons,
resentations of scientific concepts have been selected from
and electrons) and associated models, cations and anions,
textbooks used in Belgian secondary schools. The schematic
and electronegativity.
r Qualitative approach of chemical reactions: interpretation representations and the questions were selected based on
the following criteria: (i) the illustrated concepts had already
of a chemical reaction at the particle level, chemical equa-
been taught previously (see the summary of the curricula in
tion, inorganic chemical functions, oxidation numbers, Ar-
the previous subsection); (ii) the schemes refer to central con-
rhenius model, ionic dissociation of a salt, and electrolysis.
cepts in chemistry; (iii) the distractors test whether students
are able to infer a correct information that is effectively con-
These new science curricula have been introduced in 2014
tained in the schematic model; in other words, they must
(French-speaking community) and 2015 (German-speaking
be able to focus on the specific information content of the
community). The development of modelling abilities is par-
schematic model; and (iv) the distractors focus on the trans-
ticularly emphasized both in the general recommendations
lation of the sometimes implicit symbolic language of the
and in the detailed descriptions of the specific topics to be
schematic models into conceptual explanatory information
taught.
and on the awareness of the intrinsic simplified representa-
We have not attempted to systematically compare the re-
tional character of such a model.36
sults for the two populations (French- and German-speaking).
In a validation phase, the questionnaire has been submit-
However, to check the validity of the instrument, we per-
ted to university and secondary school colleague teachers to
formed Kruskall–Wallis tests for all items of the instru-
estimate the time required to perform the test and to opti-
ment to detect possible differential item functioning between
mize the level of difficulty of each question. The students’
French- and German-speaking participants. For the major-
responses were encoded as one point per correct answer.
ity of items, the null hypothesis was retained.35 As a mat-
Cronbach’s alpha coefficient has been computed to assess the
ter of fact, because the regional education system in Belgium
internal coherence of the six items (see the result section).
evolved from a previously unique system, the differences be-
tween the French- and German-speaking teaching environ-
ments are not significant and may be assumed not to affect Part B: student perceptions with respect to
the data analysis and conclusions. schematic models: Likert scale
Part B (see Supplementary 1.B) was designed with three
The instruments items using a Likert scale (“1” = strongly disagree, “2” = dis-
A questionnaire was established to assess both the pre- agree, “3” = agree, and “4” = strongly agree).37 It aims at iden-
viously described skills (analysing a schematic model and tifying the strengths and weaknesses of schematic modelling
designing such a model) and how relevant the students as perceived by the learners in their learning process. As
discussed, e.g., by Liu et al.38 and by Treagust et al.,7 students’ tackle a modelling task where they feel at ease (or believe so)
affective attitudes play a determinant role when learning sci- with the scientific content, even though we are aware of un-
entific concepts: one question focused therefore on the per- avoidable biases linked to the variable knowledge level of the
ceived power of iconic language and another one on the aes- students (based on both personal qualifications and quality of
thetic dimension. A third item, more at the metacognitive previous teaching). It is also clear that the degree of complex-
level and referring, among others, to the work of Fredlund et ity of the three different phenomena is not exactly the same,
al.,36 addressed the question whether the students are aware involving, e.g., more types of particles in the case of the ionic
of the need to help deciphering the scheme language by a dissociation of NaCl or the current flow in a metal compared
textual explanation. to the vaporization of water. A core content was present in
the three situations, however: the need to address, in an ideal
modelling, the spatial organization of the particles, and their
Part C: autonomous modelling task motions and interactions. The possible influence of the com-
As mentioned by Treagust et al.,7 students have seldom plexity differences will be dealt with in the result and discus-
the opportunity to engage into autonomous schematic mod- sion sections.
elling tasks despite the proven benefits for a deeper concep-
tual understanding.39 From the vantage point of the present Data collection
study, confronting learners with open modelling tasks and In all three schools, data collection was completed during
analysing their productions are expected to allow us to iden- the first semester of the 2018–2019 school year under the su-
tify their spontaneous reasoning, in particular the way they pervision of the chemistry teacher responsible for the tested
connect the three aspects of Johnstone’s triangle. group and one of the co-authors of this work, who informed
For the purpose of assessing the schematic modelling skills the students on the goals and the framework of the study.
through the last 3 years of the secondary school, learners
were invited to draw, based on their knowledge acquired in Data analysis
the previously described curriculum (see also below), a graph- The student answers to the questionnaire were processed
ical schematic model describing (“model of”)22 and explaining and analysed using the MS Excel and SPSS (version 23) soft-
(“model for”)22 a scientific phenomenon within three prese- ware. Independent sample mean comparison tests and mul-
lected choices that were: water vaporization, dissolution of tiple comparisons tests were used to check the significance
NaCl in water, or electric current flow through a metal (see of differences between groups of students according to their
Supplementary 1.C). different grades in the upper secondary school, to identify
The following competences have been mainly targeted: the possible differences across the secondary school curricu-
representing molecules, displaying explanatory elements, lum. The statistical methods that were used are mentioned
and establishing the link between the particle and the macro- in Table 2. A significance (alpha) level of 0.05 was used for all
scopic level. Three criteria led to the choice of the selected statistical tests.
phenomena. First, all three ones had already been taught ei-
ther in the 9th grade (NaCl ionic dissociation in water and
current flow in a metal conductor) or even in the 7th–8th Results
grades (vaporization of water) with further deepening in the
next years (see the section describing the sample and the Part A: capability to analyse the information of
curricula). A second selection criterion is that for the three a schematic model
phenomena, both a macroscopic description (sketch of an The reliability of Part A of the test can be considered as ac-
idealized experiment) and a rudimentary or more advanced ceptable based on the computed Cronbach’s alpha value of
representation at the particle level are possible and may be 0.72. The results of the different items could then be aggre-
spontaneously chosen by the students, even though it is clear gated to provide scores between 0 and 6 (one point per correct
that only the particle level integrates the explanatory dimen- answer). The difficulty indices of the items spanned a broad
sion. The third criterion is the modelling of a phenomenon range between 0.10 and 0.83 and the indices of discrimina-
rather than an object. This includes at least partly an explana- tion ranged from 0.33 to 1. The Kolmogorov–Smirnov test in-
tory dimension that requires that the students generate a dicated that the data of Part A are not normally distributed.
model for explaining rather than a model of a situation.22 A non-parametric test was thus used, the Kruskal–Wallis test,
The aim of the work was here to unravel the spontaneous to analyse the scores.
way students deal with models at different stages of their The box plots of Fig. 2 provide information on the average
secondary school education. This is the very reason why no scores and on the score dispersions for the learners over the
specific instruction on the expected level of description and three successive grades (10th, 11th, and 12th).
explanation (macroscopic vs submicroscopic) has been pro- Figure 2 shows that 12th-grade students achieved the best
vided to best promote spontaneous modelling by the stu- scores in Part A. The 10th- and 11th-grade students presented
dents. similar median scores. The variability of the score of the 12th-
By letting the students choose freely the phenomenon they grade students was smaller as indicated by the interquartile
had to model, we intended to have them focus on the situa- range.
tion that they feel to master best. The primary aim was not The Kruskal–Wallis test was performed and indicated sig-
to test their knowledge per se but to identify the way they nificant differences of capability to analyse the information
Fig. 2. Box plots for the schematic model analysis scores ac- Part B: student perception with respect to
cording to the student grades of the upper secondary level. schematic models: Likert scale
The bold lines indicate the median score for each population. Figure 3 summarizes the answers of students about how
The rectangles represent the interquartile ranges. they perceive the role of schematic models in their learn-
ing process. The answers of the students to these three items
have not been aggregated, contrary to Part A, because they
refer to three different aspects of model reception, for which
the rational versus emotional dimensions are not equally
weighted.
Figure 3 (item 1) shows that the majority of tested stu-
dents of all grades consider that the contents of an image
must be supplemented by other information (other models,
explanations, etc.) to reach a better understanding of reality.
The Kruskal–Wallis test did not show any significant differ-
ence between the different grades (χ 2 (2, N = 216) = 5.095,
p = 0.078).
Student answers to item 2 (Fig. 3) suggest that a majority
of them feel more attracted by graphics than by verbal lan-
guage. Again, no statistically significant differences between
the three groups of students were found for this item (χ 2 (2,
N = 216) = 2.009, p = 0.366). The inter-quartile interval of
10th-grade students is broader than the inter-quartile inter-
val of 11th- and 12th-grade students.
The answers to item 3 (Fig. 3) do not show any clear-cut
tendency on the role of aesthetics in a scheme. According to
the median, 10th- and 11th-grade students seem to be more
of a schematic model between learners of the three groups sensitive to aesthetical qualities of an image than 12th-grade
corresponding to the three different grades (χ 2 (2, N = 216) students. However, the differences among the three groups
= 17.134, p < 0.001). of students are not statistically significant (χ 2 (2, N = 216) =
Dunn’s post hoc test for multiple comparisons was sub- 2.561, p = 0.278).
sequently performed to determine which pairs of samples
provided significant differences. The test results (Table 3) re-
vealed significant differences between two pairs: 10th versus Part C: autonomous modelling task
12th scores and 11th versus 12th level, with p < 0.05. In Part C, students are confronted with an autonomous
Finally, for the purpose of comparison with the data of Part schematic modelling task. We expect this part of our study
C, the 216 students have been categorized, following Ferk et to allow us to answer two of our research questions: (i) Are
al.,40 into three levels according to their scores in Part A: Poor higher grades students more confident with the submicro-
(score < average (M) − one standard deviation (σ ); n = 20 stu- scopic view of chemical phenomena than lower grade ones?
dents), Average (M − σ ≤ x ≤ M + σ ; n = 179 students), and (ii) Are the autonomous modelling competences correlated to
Superior (score > M + σ ; n = 17 students). As mentioned be- the level of ability to interpret an existing reference model?
fore, the data of this Part A are not normally distributed. We The percentage of students having chosen each phe-
believe, however, that the procedure introduced by Ferk,40 nomenon is reported in Table 4.
who gave no information on the normality of the data, is ro- This table shows that 40%–45% of the students chose the
bust enough and will be useful to compare the scores in parts water vaporization phenomenon, whatever the grade. The
A and C, in particular the correlation between the analysis reason for this choice has not been investigated. A hypothesis
abilities and the qualitative categorization of the au- would be that, as mentioned before, this process seems less
tonomous modelling as macroscopic or submicroscopic. complex because it involves only one type of particles and
Table 3. Dunn–Kruskal–Wallis multiple comparison tests for the capability to analyse the information of a graphical schematic
model.
Levels of Part A (pairs of different stages) Mean difference Standard error Chi-square (χ 2 ) p value
10th versus 11th 0.1582 0.1288 0.467 0.494
10th versus 12th 0.6379 0.1319 12.188 p < 0.001
11th versus 12th 0.7961 0.1420 14.384 p < 0.001
Fig. 3. Box plot of the answers on items 1 (top left), 2 (top right), and 3 (bottom left) of Part B according to the student grades in
the upper secondary level. “4” corresponds to “strongly agree”, “3” to “agree”, “2” to “disagree”, and “1” to “strongly disagree”.
that students might be more familiar with it since it is taught - In a second approach, the student schematic models
already in the 7th and 8th grades. were quantitatively assessed for the accuracy and pre-
To infer a maximum of information from the collected cision of their scientific information content, accord-
data, the following three levels of analysis were imple- ing to six criteria listed in the Supplementary material
mented: (Supplementary 2).
- The third level of analysis looks for a possible correlation
- In a first step, each student schematic production
in the ability levels for schematic level interpretation, de-
was assigned a category according to the balance be-
fined in Part A, and the scores reached in the quantitative
tween the macroscopic and submicroscopic aspects rep-
assessment of the autonomous schematic modelling skills.
resented in the drawing. Three categories were defined:
macroscopic, elementary submicroscopic, and advanced
submicroscopic, as described in Table 5. It must be Type of model representation
emphasized here that this analysis assessed only the Depending on the pedagogical goals aimed at, and on the
submicroscopic–microscopic balance and not the scien- emphasis put on the macroscopic and submicroscopic lev-
tific accuracy, which is the focus of the next step. els of Johnstone’s triangle, schematic models often appear to
belong to one of the three categories defined in Table 5. Based To look for a possible link with the student abilities to in-
on research on the modelling capacities of students,17,41 we fer information from a reference schematic model, Fig. 6 dis-
posit that this categorization represents a key to reading the plays the percentage of each model category for the students
students’ productions. recognized as having poor, average, or superior model inter-
The schematic models drawn by the tested students are rec- pretation skills (see Results: Part A).
ognized to be very diverse and sometimes intermediate be- Figure 6 suggests a correlation between the model inter-
tween two categories, so that choices had to be made. It is nev- pretation skills and the capability of designing autonomously
ertheless observed that in most cases the categorization pro- an advanced submicroscopic model. The percentage of such
cess was rather obvious and that the collected data shed light models increases as the interpretation skills improve.
on the students’ spontaneous modelling approach across up-
per secondary school. For this part of analysis, the three co-
authors therefore assessed collectively the students’ models Quantitative assessment of the content of the
and easily built a consensus view. To illustrate this categoriza- student schematic models
tion procedure, Fig. 4 shows examples of students’ represen- The learner productions have been quantitatively analysed
tations for the three phenomena. Consider, e.g., the dissolu- using a list of six criteria that had to be met in an ade-
tion of NaCl in water. Figure A focuses on a sketch of the ex- quate representation (see Supplementary 2). These criteria
perimental apparatus (also introducing a non-necessary heat- have been specifically designed in the frame of this study to
ing source). Figure B introduces clearly submicroscopic parti- span three levels of increasing modelling competences: level
cles, distinguishing between cations and anions and suggest- 1 (criteria 1 and 2), level 2 (criteria 3 and 4), and level 3 (crite-
ing the transition from the solid state (left) to the dissoci- ria 5 and 6). Adapted criteria had to be selected for the three
ated solution state (right). Interactions with the solvent (sol- phenomena to be modelled, considering their specificities, as
vatation) are, however, omitted in this elementary submicro- already discussed before. Each fulfilled criterion was counted
scopic approach. This makes the difference with Fig. C that for one point, leading to a score in the 0–6 range. When am-
clearly introduces the role of the solvent in the ionic dissocia- biguities existed whether a criterion was fulfilled or not, half
tion process and can be considered as reaching the advanced points could be assigned.
submicroscopic stage. This assessment was more difficult than assigning one of
The distribution of student schematic models according to the categories discussed in the previous subsection (Table 5
the three categories of Table 5 is displayed in Fig. 5 for the and Figs. 5 and 6). To limit the influence of subjective judg-
three selected situations. ments on the criteria fulfilment, all student productions were
Despite some fluctuations, Fig. 5 shows a clear trend to- assessed independently by the three co-authors. Cohen’s κ pa-
wards a more advanced submicroscopic representation as the rameter was computed to check the level of agreement.42 All
level of instruction increases. For example, for the salt dis- κ values lie between 0.7 and 0.9. The average κ values for mod-
solution phenomenon, the percentage of students drawing elling the vaporization of water, dissolution of NaCl in water,
an advanced submicroscopic model increases from 0% (10th and electric current flow through a metal are, respectively,
grade) to 28% (11th grade) and then to 58% (12th grade). 0.80, 0.83, and 0.90. These κ values indicate a high agreement
Similar tendencies are observed for the other topics. It between the three appraisers for the assessment of the three
must be noticed, however, that students were least suc- phenomena.43
cessful to represent the particle level for the electric cur- To illustrate the assessment based on the criteria of Sup-
rent flow process, whatever their age. For example, only plementary 2, let’s consider again the dissolution of NaCl in
31% of the 12th-grade students proposed an advanced sub- water (Fig. 4). The model in Part A barely meets criterion 1,
microscopic model for this process, compared to 58% for although it mentions the salt (and suggests possibly macro-
the salt dissolution process. This might be linked to the al- scopic tiny NaCl crystals). Criterion 2 is fulfilled. The four
ready mentioned complexity of this task that required to other criteria are not met. This model reaches then a score
consider the motion of conduction electrons in the field of of 0.5. Model B fulfils criteria 1, 2, and 3 but not 4, 5, and 6
ionic cores. No significant differences were, however, ob- and reaches then a score of 3. Model C reaches a score of 6,
served when comparing water vaporization and NaCl disso- all criteria being fulfilled.
ciation, although the latter process involves ions and solvent First, a classification of students’ schematic models as
molecules. a function of the number of met criteria has been
Fig. 4. Examples of students’ representations of the three selected phenomena: (A) macroscopic model, (B) submicroscopic
model: elementary level, and (C) submicroscopic model: advanced level. [Colour online.]
performed. A representation was globally considered as “ad- level criteria also reached the lower level ones but some
vanced” if it met 5 or 6 criteria, as “intermediate” for 3 exceptions were unavoidable. The results of this assess-
or 4 met criteria, and as “basic” when only 1 or 2 crite- ment are shown in charts A in Fig. 7 for the three mod-
ria were reached. In most cases, students reaching higher elled situations, where the percentage of students in each
Fig. 5. Distribution of student schematic models over the three categories depending on their grade in the upper secondary
school. [Colour online.]
Fig. 6. Distribution over the categories of autonomous schematic models as a function of the capability level of inferring
information from a reference model. [Colour online.]
of the three groups (basic, intermediate, and advanced) which also confirms the relevance of the hierarchy of chosen
is displayed. criteria.
To check the consistency of the data, they have been rep- Similar conclusions emerge for the dissolution of NaCl, but
resented in a different but complementary way in charts B the students’ scores are worse than for the vaporization of wa-
of Fig. 7. In these charts, the average result reached for each ter. For example, still 42% of the 12th-grade students remain
of the three criterion levels (see Supplementary 2) is plotted: at basic level. This might be due to the fact that this phe-
this corresponds to a value between 0 and 2 because there are nomenon is in fact a more complex process involving cleav-
two criteria per level. age of ionic bonds and ion hydration. For the modelling of
The data reported in charts A of Fig. 7 highlight a positive the third phenomenon, electric current flowing through a
trend from the 10th to 12th grade. For example, for the water metal, 10th- and 11th-grade students were not very success-
vaporization, the percentage of students at only basic level ful in building models, reaching mainly a basic level (92% of
dropped from 76% (10th grade) to 62% (11th grade) and then the 10th-grade students and 72% of the 11th-grade students).
to 22% (12th grade). These results are confirmed in Fig. 7B, 12th-grade students had less difficulties in designing models
Fig. 7. Assessment of students’ autonomous schematic modelling tasks for the three phenomena according to different grades.
(A) Assessment according to the number of criteria met in each representation, and (B) average number of criteria fulfilled (0,
1, or 2) for each level: level 1 (criteria 1 and 2), level 2 (criteria 3 and 4), and level 3 (criteria 5 and 6). [Colour online.]
of this phenomenon (42% basic level, 40% intermediate, and three levels of criteria, which confirms a tendency towards an
18% advanced level). intermediate or even advanced modelling ability.
An important point has to be made when comparing charts The differences between the different grades are statisti-
A (representing percentages) and B (representing scores). cally significant as shown by the Dunn–Kruskal–Wallis mul-
Consider, e.g., the dissolution of NaCl in water. In chart A, tiple comparisons of the scores in Part C (Table 6).
a decrease of the percentage of basic models from the 10th
to the 12th grade appears in parallel with a percentage in-
Comparison between student competence
crease of the intermediate and advanced models. In chart B,
the scores of all three levels increase, which is consistent with levels in inferring information from a reference
the data of chart A. Whereas on average 10th-grade students schematic model and autonomous modelling
do not even reach completely level 1 (score < 0.5), 12th-grade competences
students reach better scores (even if non-ideal) for all three The aim here is to investigate whether the compe-
levels. In all three B charts, a positive trend is observed for the tences to analyse a reference schematic model have some
Table 6. Dunn–Kruskal–Wallis multiple comparisons of modelling task of the three modelled phenomena.
Levels of Part C (pairs Modelled
of different grades) phenomenon Mean difference Standard error Chi-square (χ 2 ) Level of significance (p)
10th versus 11th grade Water vaporization 0.47 0.13 4.640 0.03
Dissolution of NaCl in 1.23 0.19 28.184 < 0.001
water
Electric current flow 0.79 0.20 7.928 0.005
through a metal
10th versus 11th grade Water vaporization 1.39 0.14 11.162 < 0.001
Dissolution of NaCl in 1.95 0.18 48.867 < 0.001
water
Electric current flow 1.64 0.23 22.197 < 0.001
through a metal
11th versus 12th grade Water vaporization 0.92 0.15 33.567 0.001
Dissolution of NaCl in 0.72 0.20 8.137 0.004
water
Electric current flow 0.85 0.24 4.702 0.030
through a metal
Fig. 8. Box plot of competences to design autonomously The Kruskal–Wallis test (χ 2 (2, N = 216) = 6.609, p = 0.037)
a schematic model as a function of the competence level showed a significant difference among students of the three
reached in interpreting a reference model. different levels. However, as shown by Dunn’s post hoc mul-
tiple comparisons (Table 7), significant differences arise only
between the “poor” and “superior” groups, as might also have
been expected from Fig. 8.
Discussion
In this section, we discuss globally the results of the three
parts (A, B, and C) in relation with the three research ques-
tions. For the ease of the discussion, we recall here these ques-
tions:
Table 7. Comparison of students with different scheme analysis competence levels confronted to autonomous model design.
Levels of Part C (pairs of
educational level) Mean difference Standard error Chi-square (χ 2 ) Level of significance
Superior versus Poor 0.987 0.234 6.521 0011
Superior versus Average 0.839 0.193 3.480 0.062
Average versus Poor 0.148 0.157 2.509 0.113
reactions) are taken up in detail in the 12th grade. It might information from existing submicroscopic graphical repre-
then be assumed that during their 10th and 11th grade years, sentations (Fig. 6).
the students get progressively more and more familiar with Models at the submicroscopic, particle level have a bet-
reasoning at the three Johnstone’s levels so that they can ter unifying power because they are not limited to a spe-
reach what we called an “advanced” level only after complet- cific macroscopic situation but encompass a variety of con-
ing grade 11th. texts. Because they make decontextualization possible, they
These results support other studies reported in the litera- also enable a more thorough understanding of a class of
ture, showing that a majority of learners need to revisit a spe- chemical phenomena. This is, however, barely recognized by
cific situation several times during their chemistry education young learners. According to Greca and Moriea,45 and Fred-
to develop progressively their cognitive abilities to interpret lund et al.,36 the relationship that is established between re-
chemical representations of this situation.17,18 ality and a physical model is complex, and the textbook im-
Figure 3 summarizes the results collected on the percep- ages that illustrate physical concepts are not spontaneously
tion that students have about schematic models. Kruskal– understood because the correspondence between their con-
Wallis tests did not detect any particular trend through tents and the macroscopic reality is not straightforward. Ferk
the upper secondary school curriculum, so that students’ et al. found a significant correlation between spatial visual-
global perceptions appear to remain stable. Students have ization abilities of students and their educational level from
a positive perception of models and they appreciate the elementary (aged 13–14 years) to secondary (aged 17–18 years)
compact information they provide but are aware of the school, and then to university (aged 21–25 years).40 They ob-
need of additional verbal-type information to complement served that the youngest students have difficulties to under-
the graphical schemes to make an optimal use of it. stand the unifying nature of a model. Grosslight et al. found
They are rather neutral as far as aesthetical aspects are that “these students have a poor understanding of models;
concerned. they probably believe that models are incomplete copies of
Let us focus now on the competencies for autonomous real objects, which is the lowest level of modelling”.46 Hurst
model design, illustrated by a few examples in Fig. 4. Such has claimed that “this oversimplified presentation misleads
student graphical representations of a chemical process can- chemistry students and may actually cause learning impedi-
not be considered as consistent with a recognized scientific ments”.47
view if they remain restricted to a simple rearrangement The quantitative analysis based on the six assessment cri-
of particles, which can be referred to as a “basic” particle teria of Supplementary 2 and summarized in Fig. 7 reveals
model, and if they do not include information on interac- a significant positive trend through the three grades of the
tions between particles and dynamical aspects that are cen- upper secondary school. A correlation is observed between
tral for an understanding of the considered phenomenon.44 the scores obtained in the autonomous modelling task and
Johnstone showed that students need time to become able to in the reference model analysis task but statistically signif-
create a link between the three levels of thinking in chem- icant differences appear only between the two extreme cat-
istry (macroscopic, submicroscopic, and symbolic), and to egories (“poor” and “superior”). We note, however, that the
make connections between schematic representations and dispersion for the “superior” category is larger.
the taught information.14,19,20 The data displayed in Figs. 5 A finer analysis of the data requires to confront them to
and 6 are consistent with these considerations. Lower grade the objectives of the official chemistry curricula for French-
students are more prone to spontaneously propose a macro- and German-speaking schools in Belgium, as described in the
scopic descriptive model, similar to the idealized operating Methods section. Most concepts involved in the proposed sit-
mode of an experiment, whereas a tendency to switch to uations to be modelled are taught in the 7th to 9th grades, in
a submicroscopic symbolic representation, even up to an the chemistry but also in the physics course (followed in par-
advanced level (including interactions and (or) particle mo- allel by the tested students). These concepts and competences
tion), is observed for upper grade students (Fig. 5). Figure 5 are revisited at a deeper level in the next years.
shows that, respectively, 70%–90% of the 10th-grade students, A first observation concerns the symbolic level of a
30%–70% of the 11th-grade students, and 10%–30% of the schematic modelling. A majority of students omitted to anno-
12th-grade students limited their models to the macroscopic tate the scheme and to mention the meanings of the symbols
level. Interestingly, the tendency to favour a submicroscopic used: the percentage of those learners who did represent an-
representation in a modelling task seems to correlate posi- notations of the symbols increases, however, from 7% (10th
tively with the ability, assessed in Part A, to infer relevant grade) to 22% (11th grade), and then to 40% (12th grade). The
second important observation concerns concepts that con- comparison of results of parts A and C also indicates that the
tribute to the transition between basic to advanced submi- autonomous model design competence level is related to the
croscopic modelling (Table 5), that is, intermolecular interac- level of capability to analyse the information of a scheme. The
tions and thermal motion. Interactions between particles are perception of the learners with respect to schematic models
not considered by the students of the 10th grade and only is globally positive and remains stable along the three grades.
3% of the 11th-grade students and 9% of the 12th-grade stu- Let us mention that our study is complementary to the con-
dents mentioned them explicitly. The results for 10th-grade cept cartoons that have been developed by Keogh and Nay-
students are probably not surprising based on the school cur- lor.50 It is a simple alternative to develop learners’ thinking
ricula but the very low improvement in the next 2 years is by drawing.
deceiving as far as molecular interactions (hydrogen bonds Our approach allows assessing the conceptual change of
in particular) are taught in the 11th grade. Particle thermal the students and identifying the obstacle met in the appro-
motion, even though it is taught in the physics course in priation of the concepts.21,25,28,29 The majority of the 10th-
the 10th grade and revisited in the chemistry course, was and 11th-grade students proposed a modelling production at
also strongly ignored by students: only 2%, 10%, and 29% of macroscopic level. However, the 12th-grade students showed
them mentioned it in the 10th, 11th, and 12th grades, re- better abilities to reason at the submicroscopic level. The pro-
spectively. These results are, however, better than those for portion of students taking into account intermolecular in-
the intermolecular interactions. The concept of ion, which teractions and particle thermal motion in their models re-
the learners are confronted very early to, already in grade 9, mains deceiving even though some progression takes place.
is also very poorly implemented in the students’ drawings: Although they are part of the official curriculum, these con-
0% (10th grade), 10% (11th grade), and 31% (12th grade) intro- cepts appear to remain too abstract for the students who
duced them in an appropriate way in their schematic model do not mobilize them to explain a specific chemical phe-
productions. Globally, except for the intermolecular interac- nomenon.
tions where the results remain very unsatisfactory, the per-
formances can be considered to increase steadily with a more
significant jump between the 11th and 12th grades. These de- Limitations and perspectives of this study
tailed results are consistent with the overall data shown in A first limitation is that the study performed is cross-
Fig. 7, where the advanced level is seen to be poorly achieved sectional rather than longitudinal. It would be interesting to
even by 12th-grade students. As some researchers already no- follow during a 3-year period a given test group from the 10th
ticed, these concepts present strong conceptual difficulties grade to the 12th grade along with appropriate modelling
that contribute to hamper their active understanding and training activities and to compare the results with a control
mobilization by students.45,48,49 group.
A second limitation is that we did not complete the study
by using more qualitative tools. It has not been possible to
Conclusions and perspectives interview students after they had realized the three parts of
The present work investigated the interaction between up- the questionnaire, to describe more precisely their opinions
per secondary school students and schematic models along on the schematized concepts and their model choices to ex-
four directions: (i) their ability to analyse a reference graph- plain the selected scientific phenomena.
ical scheme met in a textbook, that is, to extract the rele- This study did not specifically follow the goal of analysing
vant scientific information; (ii) their psychologic-emotional student preconceptions even though some weaknesses we
reception of schematic models in a scientific context; (iii) identified could be interpreted as such (e.g., the neglect of
their competences for the autonomous design of a schematic intermolecular interactions). We believe, however, that it is
model for a scientific phenomenon which they had already possible to design specific modelling tasks for well-chosen
been confronted to in their science education; and (iv) the phenomena to identify student preconceptions.
possible correlation between the ability to interpret a refer- We are aware of the fact that this study only examined 216
ence graphical scheme and to design such a scheme. students with unequal distribution among the three groups
The sample consisted of 216 students of 10th, 11th, and of students of different grades. A complementary study could
12th grades, from Belgian French- and German-speaking be envisaged with a larger number of students. The correla-
schools with similar science curricula. Because the data tion between the quality of different schematic models de-
were not normally distributed (according to the Kolmogorov– signed by each student could also be investigated.
Smirnov test), the Kruskal–Wallis test was used to identify This research focused on the analysis of individual student
significant differences between groups, with additional post productions and neglected the social dimension of a teach-
hoc Dunn’s tests when needed. ing sequence where designed models can be the basis for
A positive trend in the investigated competences, both interactions between peers. This paper reports on tools and
for interpreting a reference schematic model and for au- methods that could be applied to assess how learners’ con-
tonomously designing an original one, is in most cases ob- ceptions and modelling skills evolve either during a learning
served through the three grades but the most significant im- sequence within a given year or over consecutive years. Model
provements appear at the transition between the 11th and design can be integrated in a teaching sequence where the
12th grades. A tentative explanation based on the structure particle model plays a central explanatory role (e.g., chem-
of the relevant chemistry curricula has been put forward. The ical kinetics) and contributes to a thorough understanding.
Peer reviewing of the designed individual models, or social (8) Pluta, W.; Chinn, C.; Duncan, R. J. Res. Sci. Teach. 2011, 48, 486.
interactions within peer groups to develop a common insti- doi:10.1002/tea.20415.
(9) Chen, J.; Wang, M.; Grotzer, T. A. J. Res. Sci. Teach. 2018, 55, 1239.
tutionalized model, under the guidance of the teacher may
doi:10.1002/tea.21450.
be suggested.51 This approach can also integrate visualization (10) Lehrer, R.; Schauble, L.In Converging Perspectives on Conceptual
tools along with active learning strategies using, e.g., inter- Change. Mapping an Emerging Paradigm in the Learning Sci-
active websites to support schematic drawing.52–54 The use of ences;Amin,Levrini, Eds.; Routledge: London and New York, 2018;
pp 163–169.
activities on an interactive website can improve students’ un-
(11) Kermen, I.; Méheut, M. Chem. Educ. Res. Pract. 2009, 10, 24. doi:10.
derstanding of chemical phenomena by visualizing the links 1039/B901457H.
between the macroscopic and submicroscopic levels. (12) Tiberghien, A.; Psillos, D.; Koumaras, P. Instr. Sci. 1995, 22, 423.
(13) Walliser, B. Systèmes et modèles; Seuil: Paris, 1977.
(14) Johnstone, A. H. J. Comput. Assist. Learn. 1991, 7, 75. doi:10.1111/j.
Acknowledgements 1365-2729.1991.tb00230.x.
(15) Kapon, S.In Converging Perspectives on Conceptual Change. Map-
The research presented in this paper was the result of a col-
ping an Emerging Paradigm in the Learning Sciences;Amin,Levrini,
laboration between several schools, teachers, and Belgian stu- Eds.; Routledge: London and New York, 2018; pp 198–205.
dents. We acknowledge the help of Mrs Kathy Mawet and (16) Treagust, D.; Chittleborough, G.; Mamiala, T. Int. J. Sci. Educ. 2002,
Véronique Lonnay for the critical review of the questionnaire. 24, 357. doi:10.1080/09500690110066485.
(17) Sim, J.-H.; Daniel, E.-G.-S. Cogent Educ. 2014, 1, 991180. doi:10.1080/
We are grateful to Mrs Betty Guilliams, Naima Saidi, and
2331186X.2014.991180.
Sylvie Larock for the opportunity to collaborate in their sec- (18) Abd-El-Khalick, F.; Myers, J.-Y.; Summers, R.; Brunner, J.; Waight, N.;
ondary school classes. Wahbeh, N.; Zeineddin, A.-A.; Belarmino, J. J. Res. Sci. Teach. 2017,
54, 82. doi:10.1002/tea.21339.
(19) Johnstone, A. H. J. Chem. Educ. 1997, 74, 262. doi:10.1021/
Article information ed074p262.
(20) Johnstone, A. H. Chem. Educ. Res. Pract. 2006, 7, 49. doi:10.1039/
B5RP90021B.
History dates (21) Cheng, M. M.; Gilbert, J. K. Int. J. Sci. Educ. 2017, 39, 1173. doi:10.
Received: 12 June 2021 1080/09500693.2017.1319989.
Accepted: 29 March 2022 (22) Gouvea, J.; Passmore, C. Sci. Educ. 2017, 26, 49. doi:10.1007/
Version of record online: 27 July 2022 s11191-017-9884-4.
(23) Lopes, J. B.; Costa, N. Int. J. Sci. Educ. 2007, 29, 811. doi:10.1080/
09500690600855385.
Copyright (24) Gilbert, J. K. Int. J. Sci. Math. Educ. 2004, 2, 115.
© 2022 The Author(s). Permission for reuse (free in most (25) Prins, G. T.; Bulte, A. M. W.; Van Driel, J. H.; Pilot, A. Int. J. Sci. Educ.
2008, 30, 1867. doi:10.1080/09500690701581823.
cases) can be obtained from copyright.com.
(26) Valk, T.; Driel, J. H.; Vos, W. Res. Sci. Educ. 2007, 37, 469. doi:10.1007/
s11165-006-9036-3.
(27) Ghirardi, M.; Marchetti, F.; Pettinari, C.; Regis, A.; Roletto, E. J. Chem.
Author information Educ. 2014, 91, 59. doi:10.1021/ed3002336.
(28) Maia, P. F.; Justi, R. Int. J. Sci. Educ. 2009, 31, 603. doi:10.1080/
Author ORCIDs 09500690802538045.
(29) Chang, H. Y.; Quintana, C.; Krajcik, J. J. Sci. Educ. Technol. 2014,
Hamad Karous https://orcid.org/0000-0003-4060-9673
23(3), 355. doi:10.1007/s10956-013-9468-2.
Bernard Leyh https://orcid.org/0000-0001-8447-5836 (30) Vosniadou, S. Learn. Instr. 1994, 4, 45. doi:10.1016/0959-4752(94)
90018-3.
(31) Coll, R. K.; Taylor, N. Chem. Educ. Res. Pract. 2002, 3, 175. doi:10.
Supplementary material 1039/B2RP90014A.
(32) Copper, C. L.; Koubek, E. J. Chem. Educ. 1999, 76, 1714. doi:10.1021/
Supplementary data are available with the article at https: ed076p1714.
//doi.org/10.1139/cjc-2021-0162. (33) Communauté Française Wallonie–Bruxelles. Compétences et Savoir
Requis en Sciences GéNéRales, 2014, http://www.gallilex.cfwb.be/d
ocument/pdf/40140_001.pdf .
References (34) Ministerium der Deutschsprachigen Gemeinschaft. Rahmenplan
Naturwissenschaften, 2015, https://ostbelgienbildung.be/PortalDat
(1) Carey, S. J. Appl. Dev. Psychol. 2000, 21, 13. doi:10.1016/ a/21/Resources/downloads/schule_ausbildung/schulische_ausbildun
S0193-3973(99)00046-5. g/rahmenplaene_neu/Rahmanplan_NAWI.pdf.
(2) Vergnaud, G. Revue française de pédagogie . 1991, 10, 133. (35) Arjoon, J. A.; Xu, X.; Lewis, J. E. J. Chem. Educ. 2013, 90, 536. doi:10.
(3) Vergnaud, G. Int. J. Educ. Dev. 2013, 36, 131. 1021/ed3002013.
(4) Read, J. R. Children’s Misconceptions and Conceptual Change (36) Fredlund, T.; Linder, C.; Airey, J.; Linder, A. Phys. Rev. ST Phys. Educ.
in Science Education, 2004, Retrieved 11 September 2021from Res. 2014, 10, 020129. doi:10.1103/PhysRevSTPER.10.020129.
https://asell.org/wp-content/uploads/2018/05/Childrens-Misconcepti (37) William, M. K. T. Research Methods Knowledge Base, 2006, [On-
ons-and-Conceptual-Change-in-Science-Education-1.pdf. line document]. URL. http://www.socialresearchmethods.net/kb/sca
(5) Posner, G. J.; Strike, K. A.; Hewson, P. W.; Gertzog, W. A. Sci. Educ. lgen.php.
1982, 66, 211. doi:10.1002/sce.3730660207. (38) Liu, C.-J.; Hou, I.-L.; Chiu, H.-L.; Treagust, D. F. Res. Sci. Educ. 2014,
(6) Amin, T. G.; Levrini, O.In Converging Perspectives on Concep- 44, 133. doi:10.1007/s11165-013-9373-y.
tual Change. Mapping an Emerging Paradigm in the Learning Sci- (39) Waldtrip, B.; Prain, V., Carolan, J. Res. Sci. Educ. 2010, 40, 65. doi:10.
ences;Amin,Levrini, Eds.; Routledge: London and New York, 2018. 1007/s11165-009-9157-6.
(7) Treagust, D. F.; Won, M.; McLure, F.In Converging Perspectives on (40) Ferk, V.; Vrtacnik, M.; Blejec, A.; Gril, A. Int. J. Sci. Educ. 2003, 25,
Conceptual Change. Mapping an Emerging Paradigm in the Learn- 1227. doi:10.1080/0950069022000038231.
ing Sciences;Amin,Levrini, Eds.; Routledge: London and New York, (41) Chittleborough, G.; Treagust, D. F. Chem. Educ. Res. Pract. 2007, 8,
2018; pp 121–128. 274. doi:10.1039/B6RP90035F.
(42) Cohen, J. Educ. Psychol. Meas. 1960, 20, 1960. doi:10.1177/ (49) Özman, H. J. Sci. Educ. Technol. 2004, 13, 147. doi:10.1023/B:JOST.
001316446002000104. 0000031255.92943.6d.
(43) Landis, J. R.; Koch, G. G. Biometrics, 1977, 33, 159. doi:10.2307/ (50) Keogh, B.; Naylor, S. Int. J. Sci. Educ. 1999, 21, 431. doi:10.1080/
2529310. PMID: 843571. 095006999290642.
(44) Johnson, P. Int. J. Sci. Educ. 1998, 20, 393. doi:10.1080/ (51) Zembylas, M. Stud. Sci. Educ. 2005, 41, 91.
0950069980200402. (52) Ardac, D.; Akaygun, S. J. Res. Sci. Teach. 2004, 41, 317. doi:10.1002/
(45) Greca, I. M.; Moreira, M. A. Sci. Educ. 2002, 86, 106. doi:10.1002/sce. tea.20005.
10013. (53) Zhang, Z. H.; Linn, M. C. J. Res. Sci. Teach. 2011, 48, 1177. doi:10.
(46) Grosslight, L.; Unger, C.; Jay, E.; Smith, C. L. J. Res. Sci. Teach. 1991, 1002/tea.20443.
28, 799. doi:10.1002/tea.3660280907. (54) Hensiek, S.; De Korver, B. K.; Harwood, C. J.; Fish, J.; O’Shea, K.;
(47) Hurst, O. J. Chem. Educ. 2002, 79, 763. doi:10.1021/ed079p763. Towns, M. J. Chem. Educ. 2016, 93, 1847. doi:10.1021/acs.jchemed.
(48) Wu, H. K.; Shah, P. Sci. Educ. 2004, 88, 465. doi:10.1002/sce.10126. 6b00234.
Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.
Alternative Proxies: