Available via license: CC BY 4.0
Content may be subject to copyright.
Metacognition and epistemic cognition in physics are related to physics identity
through the mediation of physics self-efficacy
Yaren Ulu 1and Sevda Yerdelen-Damar 2
1Department of Physics and Astronomy, College of Arts & Sciences,
Texas Tech University, Lubbock, 79409, Texas, USA
2Faculty of Education, Department of Mathematics and Science Education,
Bogazici University, Bebek, 34342, Istanbul, Turkey
(Received 2 January 2023; accepted 25 March 2024; published 26 April 2024)
This study aimed (i) to investigate how epistemic cognition in physics and metacognition, together with
three dimensions of physics identity framework—recognition, physics self-efficacy, and interest—
predicted the overall physics identity of Turkish high school students and also (ii) to investigate gender
differences in study constructs. A sample of 1197 high school students participated in the study. The
collected data were analyzed using structural equation modeling. The analysis results indicated that
the model fitted the data well, further motivating intervention studies to test the causal relations proposed in
the model. The results showed that recognition and interest directly predicted physics identity and
mediated the relation of physics self-efficacy to it. Metacognition and epistemic cognition predicted
physics identity through physics self-efficacy. The study also observed significant direct and indirect
relations among metacognition, epistemic cognition, self-efficacy, recognition, and interest. Furthermore,
gender differences were found in the current study. While no gender difference was observed in
metacognition and epistemic cognition in physics, male students scored higher than female students in
physics identity, self-efficacy, recognition, and interest. However, the mediation analysis further indicated
that gender differences in physics self-efficacy might explain gender differences in physics identity,
recognition, and interest. The results of this study could motivate future interventions testing the effect of
metacognitive and epistemic activities on both physics self-efficacy and identity, and also, the interventions
testing whether practices that reduce the gender gap in physics self-efficacy will help eliminate the gender
gap in physics identity, recognition, and interest.
DOI: 10.1103/PhysRevPhysEducRes.20.010130
I. INTRODUCTION
Physics identity refers to the degree to which a person
considers herself or himself a “physics person”[1]. Various
research studies have shown that students’physics identity
predicts their participation in physics classes and their
choice of careers related to physics [1,2]. The sophisti-
cation in identity enables learners to become active agents
in science by combining their knowledge with scientific
thinking methods to be purposeful and strategic learners
[3]. Based on research findings demonstrating the crucial
role of identity in students’learning, engagement, and
career paths, Organization for Economic Cooperation and
Development (OECD) [4] added scientific identity to the
Program of International Student Assessment (PISA, 2024)
assessment framework as a new dimension. It is claimed
that identity can be a tool to create a learning ecology;
therefore, the assessment framework should involve prob-
ing students’identities.
On the other hand, it is noteworthy that interest in
physics departments has gradually decreased worldwide.
For instance, fewer bachelor’s degrees in physics are given
out each year in the United States compared to other
scientific, technology, engineering, and mathematics sub-
jects [5]. Furthermore, although there was a slight increase
in the number of physics undergraduates in the United
Kingdom, there needed to be more growth since 2010 [6].
The inadequate increase in the rate of science, technology,
engineering, and mathematics (STEM) graduates was also
the case in Turkey [7]. While women receive just around
one-fifth of these degrees in the United States [8], the low
rate of women in STEM careers persisted between 2013
and 2019 in Turkey [9]. Similarly, according to the 2018
National Centre for Universities and Business report, only
22.2% of A-level physics students in the United Kingdom
were women [10]. Therefore, determining the factors
leading to this low choice rate and gender difference is
vital for physics education. As students’physics identity is
Published by the American Physical Society under the terms of
the Creative Commons Attribution 4.0 International license.
Further distribution of this work must maintain attribution to
the author(s) and the published article’s title, journal citation,
and DOI.
PHYSICAL REVIEW PHYSICS EDUCATION RESEARCH 20, 010130 (2024)
2469-9896=24=20(1)=010130(23) 010130-1 Published by the American Physical Society
a predictive variable in their career choices, especially
during secondary education [1], this study examined
Turkish high school students’physics identity. Also, it
investigated how two related constructs, metacognition and
epistemic cognition in physics, predicted physics identity.
II. THEORETICAL AND EMPIRICAL
BACKGROUND
A. Identity
Gee defines identity as “being recognized as a certain
‘kind of person,’in a given context”[11] (p. 99). Identity
does not only depend on the individual but also on the
social aspects, such that it is an outcome of an individual’s
actions and perceptions of significant others on that person.
For example, a person reaches their scientific identity as an
outcome of their competence and performance in science.
Also, they can reach their recognition as a science person in
their community [12].
According to Carlone and Johnson [12], identity com-
prises three components: competence, performance, and
recognition. While competence is related to one’s knowl-
edge and understanding of science and does not have to be
visible to the public, performance refers to revealing
scientific practices using tools or even talking. On the
other hand, recognition is a social dimension which means
that one’s recognition of oneself and others as a science
person affects identity significantly. One may not have each
dimension adequately. One may have exceeding skills
meeting the performance criterion, but others may not
recognize that one can perform it. One may have the
relevant knowledge but may not be able to perform it or
vice versa. In addition, since experiences gained in schools
affect skills and knowledge, they are related to performance
and competence dimensions. Although performance, rec-
ognition, and competence were the critical components of
science identity, interest was also considered a component;
however, because the researchers were already working
with practicing scientists, interest was attributed to the
participants and was not included in the model [12].
Hazari et al. [1] developed a physics identity framework
utilizing Carlone and Johnson’s science identity study
[12]. Physics identity refers to the degree to which a
person considers himself or herself a “physics person.”
Hazari et al. [1] proposed a framework that includes
performance, competency, recognition, and interest,
which are the fundamental interrelated constructs affect-
ing the formation of physics identity. For the physics
identity framework, competence is believing in the ability
to understand physics, and performance is believing in the
ability to carry out requisite physics assignments. In
addition, recognition is being recognized by others as a
physics person when interest is defined as the eagerness to
get more knowledge about physics and do more activities
related to physics.
While Hazari et al. [1] mentioned performance and
competence as separate constructs, in a later study, Lock
et al. [13] examined the effect of physics and math
identities on students’choice of physics careers. They
found that performance and competence were not inde-
pendent constructs. However, they comprise one construct,
“performance or competence,”which is defined as stu-
dents’beliefs in their abilities to carry out necessary
physics tasks like problems and experiments and under-
stand physics content. In the literature, a term similar to
performance or competence is conceptualized. Self-efficacy
is an individual’s confidence in her or his ability to carry out
behaviors required to achieve particular performance goals
[14]. This study uses these terms interchangeably and
prefers self-efficacy over performance or competence.
The later quantitative studies examining disciplinary
identity in physics, mathematics, and engineering observed
significant relations among these components. They
revealed that students’interest and external recognition
significantly related to an overall identity construct and
self-efficacy. Self-efficacy was not directly related to
physics identity but was positively related to identity
through the mediating relations of interest and external
recognition [2,8,15–18]. For example, Dou and Cian [17]
worked on expanding the STEM identity framework by
looking at the relationships between STEM recognition,
self-efficacy, interest, and identity, and relevant demo-
graphic and social factors, such as gender, ethnicity, home
support of science, parental education, and science talk.
They found that students’interests and external recognition
were strongly connected to their self-efficacy beliefs, which
were indirectly related to their overall identity through
interest and external recognition. The same mediational
relationship between the identity constructs was also
observed in the study of Verdín [18], in which the
interrelations among engineering identity, interest, recog-
nition, self-efficacy beliefs, sense of belonging, and per-
sistence in the engineering career were examined. Thus, in
the model of this study, the paths among the identity
components and overall identity were proposed based on
these research findings. Specifically, we hypothesized that
interest and recognition are directly related to physics
identity, and interest and recognition mediate the relation
of self-efficacy to physics identity.
B. Metacognition
Metacognition refers to knowledge and regulation of an
individual’s own cognition [19,20]. Brown categorized
metacognition into two components: knowledge of cogni-
tion and regulation of cognition. Knowledge of cognition
refers to declarative information regarding one’s cognition,
whereas regulation of cognition refers to the ability to plan,
monitor, control, and evaluate one’s own cognition.
Following Brown’s framework, Schraw and Dennison
[21] proposed an eight-dimensional framework to
YAREN ULU and SEVDA YERDELEN-DAMAR PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-2
operationalize metacognitive awareness. According to this
framework, knowledge of cognition has three levels:
•declarative knowledge (knowledge about facts and
strategies).
•procedural knowledge (knowledge about how to apply
strategies).
•conditional knowledge (knowledge about when to
apply strategies and why).
For the regulation of cognition, five skills are necessary:
planning, information management strategies, comprehen-
sion monitoring, debugging strategies, and evaluation [21].
The present study employed this framework to determine
students’metacognition.
C. Epistemic cognition
Epistemic cognition refers to one’s view of the nature of
knowledge, knowing, and learning [22,23]. Different
frameworks are used to probe students’epistemic cognition
in physics [24–26]. For example, Hammer [25] (p. 155)
conceptualizes students’epistemic cognition in physics as
follows:
1. Beliefs about the structure of physics knowledge
which can be a group made of individual parts or a
sole organized system.
2. Beliefs about the content of physics knowledge that
can consist of formulas or concepts.
3. Beliefs about learning physics in such a way that by
getting the information passively or being actively
involved in managing one’s learning and shaping
understanding.
The latest framework developed by Ozmen and Ozdemir
[27] (p. 1215) is built on the literature on epistemic
cognition in science and physics. The framework included
six dimensions. Table Iindicates the dimensions and
descriptions of the dimensions.
The current study used this framework to measure
students’epistemic cognition in physics. In the following
sections, the interrelations among the study variables are
discussed.
D. The relation of metacognition to identity, epistemic
cognition, and self-efficacy
The sophistication in metacognition is considered a
prerequisite for identity formation [28]. According to
Marcia [29], individuals exhibiting a developed identity
have more awareness of their own strengths and weak-
nesses when they make their path in life. Those features
are compatible with the characteristics of individuals with
high metacognitive knowledge, identified as an awareness
of one’sownstrengthsandweaknesses[21,30,31].
Likewise, Irving and Sayre [32] found that students in
different identity development stages indicated different
levels of metacognition. Students classified into the lowest
stage of identity development indicated a lack of self-
awareness of different approaches to learning, while
students classified into the highest stage of identity
development in the group demonstrated a completed
self-awareness of different approaches to learning phys-
ics. Moreover, research revealed a positive relationship
between decision making, which is necessary for
TABLE I. The dimensions of epistemic cognition in physics and explanations of them. Note that the table was adapted from
K. Özmen, and Ö. F. Özdemir, Conceptualisation and development of the physics related personal epistemology questionnaire (PPEQ),
Int. J. Sci. Educ., 41, 1207 (2019).
Dimensions Description of what dimensions probe
Structure of knowledge coherence (SKC) The degree to which the student views physics knowledge as a coherent vs
incoherent structure.
Structure of knowledge hierarchical (SKH) The degree to which the student views physics knowledge is formed by
establishing a link between previous and new physics knowledge with a
hierarchical vs fragmented structure.
Justification of knowledge and knowing (JK) The degree to which the student views physics knowledge can be justified
using mental processes (i.e., logical reasoning), evidence from
experimentation, and inquiry emanating from conflicts between previous
experiences and novel situations.
Changeability of knowledge (CK) The degree to which the student views physics knowledge is subject to
change or fixed (unchangeable).
Quick learning (QL) The degree to which the student views constructing physics knowledge takes
time (a gradual process of meaning-making), or learning happens very
quickly.
Source of knowledge (source) The degree to which the student views physics knowledge as constructed or
is accepted directly from authority (i.e., textbooks, teachers, and
scientists).
METACOGNITION AND EPISTEMIC COGNITION …PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-3
successful identity formation, and metacognition [33,34],
and metacognitive interventions fostered individuals’
decision-making performance [34,35]. Studies [36,37]
showed that individuals with a developed identity dem-
onstrated a high level of self-reflection, which is a
metacognitive process ([38–40]. The studies employing
self-reflection activities in training preservice teachers and
learning assistants fostered identity construction [39,41–
44]. According to Beauchamp and Thomas [45], reflec-
tion shapes teacher identity because by self-reflecting,
they can better understand their sense of self and how that
self is positioned in a larger community, including others.
Furthermore, metacognition can indirectly influence iden-
tity formation through its impact on learning. The inter-
twining nature of learning and identity construction has
been acknowledged in the literature [46–49]. According to
Var e le s [49], the learning process includes both content
learning and identity formation. Many studies depicted
that metacognitive interventions promoted student learn-
ing [50–53]. Thus, metacognitive activities could directly
and indirectly improve students’identity formation.
Researchers have also pointed out metacognition as a
necessary construct for epistemological development [54–
58]. For instance, Bendixen and Rule [54] have proposed a
model for explaining epistemic belief change and develop-
ment. Metacognition is a critical factor of this model. The
researchers assert that metacognition is vital for effective
and durable epistemological development. In another
integrated model of epistemic cognition and self-regula-
tion, Muis [57] argues that metacognitive strategy training
is crucial for epistemic development. Similarly, Elby and
Hammer [56] claim that metacognitive monitoring facili-
tates epistemological change and development, including
co-activation and stabilization of epistemological resources
that individuals already have. Research studies investigat-
ing the effect of metacognitive strategies on students’
epistemic views support the antecedent role of metacog-
nition for epistemic development [59–61].
Finally, the link between self-efficacy and metacognition
has also been highlighted in the literature [62].Mooreset al.
[62] discuss the relationship between metacognition and
self-efficacy as predictors of performance. According to the
researchers, self-efficacy determines behavior and indi-
rectly influences performance, while metacognition ini-
tiates behavior, monitors the level of performance, and
controls subsequent behavior, which informs the benefit of
metacognitive training for reaching a desired level of
performance. Thus, instruction should promote this meta-
cognitive feedback loop in which metacognition, perfor-
mance, and self-efficacy can interact with each other. In this
loop, metacognitive monitoring of performance can regu-
late subsequent behavior, which in turn influences one’s
sense of self-efficacy and stimulates the next cycle of
behavior that can be re-evaluated by metacognitive proc-
esses [62]. The role of metacognition in self-efficacy
development can be attributed to the relationship between
past achievement and self-efficacy. According to Bandura
[14], past achievement is the most influential source of self-
efficacy. Based on this fact, a variable contributing to
students’achievement could also contribute to their sense
of self-efficacy. In this sense, metacognition-enhancing
activities can potentially improve self-efficacy beliefs as
well. Experimental studies indicated that metacognition-
enhancing activities in instruction led to improved
self-efficacy beliefs [63–67]. The relationship between
self-efficacy and metacognition has also been found in
correlational studies [68–70].
E. The relation of epistemic cognition to identity, self-
efficacy, and interest
Similar to metacognition, epistemic cognition is key to
identity development. Several researchers observed a pos-
itive relationship between epistemic development and
identity formation [71–74]. For example, Faber et al.
[75] found that students’perceptions of themselves as
researchers were affected by their initial epistemic thinking
about researchers and research, and reflection on research
experiences promoted both their research identity and
epistemic thinking.
Furthermore, research studies showed that students with
sophisticated epistemic cognition possess higher self-
efficacy beliefs [76–78]. In a conceptual model, Muis
[57] hypothesizes that epistemic cognition is a precursor
to motivational beliefs, including self-efficacy, achieve-
ment goal orientations, interest, task value, and anxiety.
The role of epistemic cognition in self-efficacy develop-
ment can be further justified, considering the relation of
epistemic cognition to learning approaches. Studies
revealed that individuals with sophisticated epistemic
cognition are more likely to employ deep learning
approaches [79–82], positively influencing their learning
outcomes [83–86], which in turn might increase their self-
efficacy beliefs. Later experimental studies investigating
the effectiveness of epistemic interventions on students’
self-efficacy beliefs [65,87] and correlational studies using
structural equation models [78,88,89] provided supporting
evidence for the antecedent role of epistemic cognition.
Finally, consistent with Muis’theoretical model, a positive
correlation was revealed between epistemic cognition and
interest [76,77,90].
F. Gender differences in study variables
Many studies revealed gender differences in science and
physics identity and science and physics-related career
choices [1,91–95]. Male students showed higher levels of
physics identity than female students [1,93,95]. Similarly,
males chose physics as a career more than females [94,95].
Gender differences in favor of male students were also
found in recognition [13,96], interest [13,97,98], and
physics self-efficacy [13,99,100]. Thus, examining the
indirect effects of gender on physics identity through
YAREN ULU and SEVDA YERDELEN-DAMAR PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-4
competence, recognition, and interest can provide infor-
mation about underlying reasons for gender differences in
physics identity.
On the other hand, a consistent gender difference has not
been observed in metacognition and epistemic cognition.
For instance, Yerdelen-Damar and Peşman [70] did not
observe gender differences in metacognition, while Topçu
and Yılmaz-Tüzün [101] showed that female students had
higher metacognition than males. Similarly, while some
studies showed that female students had more sophisti-
cated epistemic beliefs than males [101,102],another
study revealed that male students had more developed
epistemic beliefs [103]. On the contrary, girls and boys
tended to hold similar beliefs regarding the source or
certainty and development dimensions, despite girls hav-
ing more complex beliefs regarding the justification of
knowledge than boys [104]. Due to contradictions in the
results, it is also necessary to inspect gender differences in
these two variables.
In conclusion, prior studies observed significant rela-
tions among identity, physics self-efficacy, interest, and
recognition constructs [2,8,13] in other contexts. On the
other hand, further research is needed to examine these
relationships in other cultures and to inspect the relations
of these constructs to other constructs. Metacognition and
epistemic cognition are two essential variables that the
research suggests to be used in instruction to improve
identity. However, few studies examined the interrelations
among metacognition, epistemic cognition, and identity
through structural equation modeling (SEM) [76,105].
Furthermore, to the best of researchers’knowledge, there
is no large-scale research examining the relation of
physics identity to either metacognition or epistemic
cognition in physics, although it is well-known that both
epistemic cognition and identity are domain-specific
constructs [106–108].
Therefore, the present study proposed a structural model
based on the above studies (see Fig. 1) to investigate the
interrelations among physics identity, recognition, physics
self-efficacy, interest, epistemic cognition, metacognition,
and gender. The current study extended the body of
knowledge on physics identity by inspecting the direct
and indirect relations of epistemic cognition in physics and
metacognition to identity-related constructs, which could
further motivate intervention studies to get a more com-
prehensive description of identity development. In addition,
the mediating role of physics self-efficacy in the relation of
physics identity to epistemic cognition, metacognition, and
gender was pointed out in this study. This study answered
the following research questions:
1. How are Turkish high school students’perceptions
of physics self-efficacy, recognition, and interest
related to their physics identity?
2. How is epistemic cognition related to physics
identity?
3. How is metacognition related to physics identity?
FIG. 1. The hypothesized theoretical model.
METACOGNITION AND EPISTEMIC COGNITION …PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-5
4. How is gender related to physics identity?
5. How are the interrelations among gender, meta-
cognition, epistemic cognition, and other identity
constructs?
Based on the direct relation of physics self-efficacy to
epistemic cognition and metacognition and gender revealed
in previous studies,
6. Does physics self-efficacy mediate the relation of
epistemic cognition to physics identity?
7. Does physics self-efficacy mediate the relation of
metacognition to physics identity?
8. Does physics self-efficacy mediate the relation of
gender to physics identity?
III. METHOD
A. Participants
The data were collected from 1197 high school students
from six high schools in Istanbul, Turkey. While 58.2% of
these students were female, 41.8% were male. The
students’ages ranged from 14 to 18 or above, and
24.9% were in 9th grade, 40.8% were in 10th grade,
33.2% were in 11th grade, and 1% were in 12th grade. Of
the participants, only 1% were 12th-grade students
because senior students prepare for the university entrance
exam, and they come to school less often since they have
extra courses out of school.
The success level of the students ranged from low
to high levels according to their high school entrance exam
scores. All the students were taking physics lectures based
on the Turkish curriculum. According to the students’
reports on the socioeconomic status questionnaire, 99.5%
of the students indicated they had Internet and technologi-
cal tools, such as telephone, tablet, and computer, and
94.2% had a suitable environment to study. The education
levels of their parents ranged from elementary to graduate
level. When the educational status of their mothers was
examined, it was seen that 17.1% of them were in primary
school, 12.7% were in secondary school, 35.8% were in
high school, 30.5% were undergraduate, and 3.8% were
graduate or doctoral graduates. When the educational status
of their fathers was examined, it was seen that 13% were in
primary school, 13% were in secondary school, 36.4%
were in high school, 31.2% were undergraduate, and 6.3%
were graduate or doctoral graduates.
The surveys of the study were administered in physics
classes that the students were taking. Before data collec-
tion, the students were informed by the first author
regarding what the surveys measure, the importance of
obtaining students’responses to the surveys for physics
education research, voluntary participation in the
research, participant confidentiality, and the right to
withdraw their data at any time. Students were not
rewarded with extra credit or points. However, no student
in attendance refused to answer the surveys during the
data collection. The students completed the surveys in one
class hour. The entire data collection process was com-
pleted in 1 month.
B. Instruments
1. Physics identity survey
The persistence research in science and engineering
study, as developed and validated by Hazari et al. [1],
provided the specific items used to measure physics
identity, including the dimensions of interest, self-efficacy
(performance/competence), and recognition beliefs.
Furthermore, it added an item measuring the overall
physics identity [15]. The physics identity survey was
further developed by Cheng et al. [2], including one
identity item, four for recognition, six for self-efficacy,
and four for interest. The dimensions of the Physics
Identity Survey and one example item for each dimension
are presented in Table II. The present study employed the
Turkish version of the identity scale validated by Ulu and
Yerdelen-Damar [109] to determine high school students’
physics identity and their conceptions of identity-formation
constructs. The confirmatory factor analysis (CFA) results
of the Turkish scale revealed the same factor structure as
that in the original scale. The multiple fit indices used to
evaluate the results were within the acceptable range
(χ2ð187;N ¼361Þ¼510.12;χ2=d:o:f:¼2.72; root mean
square error of approximation ðRMSEAÞ¼0.07 [90% con-
fidence interval ðCIÞ¼0.06, 0.0), standardized root mean
square residual ðSRMRÞ¼0.05; comparative fit index
ðCFIÞ¼0.99; normed fit index ðNFIÞ¼0.98]. Items had
significant factor loadings (p<0.05). The magnitude of
the factor loadings varied between 0.70 and 0.98. These
values are greater than the minimum required value of 0.40.
Cronbach’s alpha ranged from 0.90 to 0.97 for the dimen-
sions [109]. Similar to the original survey, the Turkish
identity scale is an 11-point Likert scale ranging from 0 (Not
at all) to 10 (Very much so).
2. Physics-related personal epistemology questionnaire
The physics related personal epistemology question-
naire (PPEQ) developed by Özmen and Özdemir [27]
wasusedtoprobestudents’epistemic cognition in
physics. The questionnaire was a five-point scale ranging
from 1 (strongly disagree) to 5 (strongly agree) and
included six dimensions: structure of knowledge
TABLE II. The dimensions of the Physics Identity Survey and
one example item for each dimension.
Dimensions Example item
Identity I see myself as a physics person.
Recognition My physics teacher sees me
as a physics person.
Interest Physics is fun for me.
Self-efficacy I can overcome setbacks in physics.
YAREN ULU and SEVDA YERDELEN-DAMAR PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-6
coherence (SKC), structure of knowledge hierarchical
(SKH), justification of knowledge (JK), changeability
of knowledge (CK), quick learning (QL), and source of
knowledge (SOURCE). PPEQ included 27 items.
Cronbach’s alpha was estimated as 0.92 for the scale.
Table III shows the dimensions of PPEQ and one example
item for each dimension.
3. Metacognitive awareness inventory (MAI)
The MAI was initially developed by Schraw and
Dennison [21], and it was adapted to Turkish by Akınet al.
[110]. It includes 52 items on a 5-point Likert scale ranging
from 1 (never true) to 5 (always true). The MAI consists of
two dimensions: knowledge of cognition and regulation of
cognition. Knowledge of cognition dimension includes
knowledge about facts and strategies (declarative), how to
apply strategies (procedural), and when and why to apply
them (conditional). The regulation of cognition dimension
includes planning, information management strategies,
comprehension monitoring, debugging strategies, and
evaluation. Table IV indicates the dimensions and sub-
dimensions of MAI and one example item for each sub-
dimension. The Cronbach’s alpha for the entire scale was
0.95, while it ranged from 0.93 to 0.98 for the subscales.
C. Procedure
This study applied a correlational research design to
investigate interrelationships among study variables.
Confirmatory factor analyses (CFAs) were conducted
to examine the construct validity of students’responses
to the scales in the current study. In other words, with
CFAs, we evaluated the extent to which the theoretical
factor structure, measurement model of metacognition,
epistemic cognition, and identity constructs fit the data
collected in the present study. After CFA analyses, we
performed a structural equation modeling (SEM) to
investigate the relations among the latent constructs in
the hypothesized model developed based on theoretical
and empirical studies. SEM enables us to decompose the
total relation of a predictor variable to a dependent
variable into direct and indirect relations [111,112].
The direct relation indicates a relation between the
predictor and dependent variable after controlling for
all other predictors of the dependent variable [111,113].It
can also be defined as a relation unmediated by other
variables in the model [112,113]. The path coefficient in
the path diagram estimates the direct relation [111].
However, the indirect relation refers to the relation of
the predictor variable to the dependent variable through
the intervening variable(s) after controlling for the cor-
responding direct relation [111,112]. It is estimated as the
products of path coefficients for each direct relation
composing the indirect pathway. If there is more than
one indirect pathway between the predictor and the
dependent variable, the total indirect relation is estimated
by the sum of each specific indirect relation. Finally, the
total relation refers to the sum of the direct and total
TABLE III. The dimensions of PPEQ and one example item for each dimension.
Dimensions Example item
SKC To understand a subject in physics, I need to understand the basic concepts of the subject.
SKH I understand a subject in the physics lesson through the knowledge I have already learned.
JK If the information given in the physics course contradicts what I know as correct, I question the rationale of this
information.
CK The knowledge I learned in the physics course is never-changing facts; so my knowledge will not change either.
SOURCE I accept what my physics teacher says in class without question.
QL If I spare enough time to study, I can understand the rationale of the knowledge given in physics class.
TABLE IV. The dimensions and sub-dimensions of MAI and one example item for each subdimension.
Dimensions
Knowledge of cognition Example item
Declarative knowledge I understand my intellectual strengths and weaknesses.
Procedural knowledge I find myself using helpful learning strategies automatically.
Conditional knowledge I can motivate myself to learn when I need to.
Regulation of cognition Example item
Planning I read instructions carefully before I begin a task.
Information management strategies I try to break studying down into smaller steps.
Comprehension monitoring I ask myself periodically if I am meeting my goals.
Debugging strategies I stop and go back over new information that is not clear.
Evaluation I know how well I did once I finish a test.
METACOGNITION AND EPISTEMIC COGNITION …PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-7
indirect relations of the predictor variable to the depen-
dent variable [111].
Furthermore, the mediation analysis with the bias-
corrected bootstrap method was carried out to examine
the mediating relation of epistemic cognition and physics
self-efficacy specified in the proposed SEM (see Fig. 1).
The mediating relations with 95% confidence intervals, not
including the value of zero, were considered significant.
The practical significance, the effect size of the relation-
ships observed among the constructs in the hypothesized
SEM, was evaluated based on the magnitude of the
standardized path coefficient (β), the predicted change in
the standard deviation unit of the dependent variable for
every standard deviation change in the predictor variable
when the other predictors are held constant [114]. We used
the standards recommended by Kline [111].Aβvalue of
0.10 indicates a small effect size, a βvalue of 0.30 indicates
a medium effect, and aβvalue of 0.50 or larger indicates a
large effect. Another practical significance measure, the
amount of explained variance (R2) on dependent variables
accounted for by the hypothesized model, was evaluated
using the threshold values proposed by Cohen and Cohen
[115].AnR2≤0.01 indicates a small effect size, an R2
around 0.09 suggests a medium effect size, and an R2≥
0.25 is taken as a large effect size.
By convention, in path diagrams, observed variables are
represented by rectangles, and latent variables or con-
structs estimated by observed indicators are represented
by ovals. Our hypothesized model includes a structural
model indicating hypothesized relationships among var-
iables and measurement models indicating the relation-
ships between the latent constructs and their measured
indicators. The structural model in the figure included two
observed variables, which are physics identity and gender,
and five latent constructs, which are metacognition,
epistemic cognition in physics, interest, recognition,
and physics self-efficacy. For simplicity, measurement
models presenting the indicators of the latent constructs
are not shown in Fig. 1. The entire hypothesized model is
given in Appendix A.
The data for the indicators of the latent constructs in the
model of the present study were ordered categorical data,
with the number of categories being more than ten. Finney
and DiStefano [116] recommended that when the number
of ordered categories is six or more, the data can be treated
as continuous, and the Satorra-Bentler (S-B) scaling
method can be employed as an estimation method. Thus,
the analyses were carried out with maximum likelihood
parameter estimates with standard errors and a mean-
adjusted chi-square test statistic that are robust to non-
normality (MLM) [117]. Multiple fit indices, which are
comparative fit index (CFI), Tucker–Lewis index (TLI),
root mean square error of approximation (RMSEA), and
standardized root mean square residual (SRMR), were
employed for evaluating the degree to which measurement
models and the hypothesized model fit the observed data.
The rule of thumb for model fit is that RMSEA ≤0.05,
CFI ≥0.95, TLI ≥0.95, and SRMR ≤0.05 suggest a good
fit, while RMSEA ≤0.08, CFI ≥0.90, TLI ≥0.90, and
SRMR ≤0.10 suggests an acceptable fit [118–120].
Descriptive statistics, correlation, and reliability analyses
were carried out in spss 27. Cronbach’s alpha and mean
interitem correlation (MIIC), measures of internal consis-
tency across items, were used as reliability measures. A
value of Cronbach’s alpha bigger than 0.70 is required for
reliable results [121]. As the size of Cronbach’s alpha is
also influenced by the number of items for short scales,
Briggs and Cheek [122] suggest MIIC, which is indepen-
dent of the length of scales. Therefore, MIIC was employed
to estimate the reliability of the subdimensions of the
scales. Briggs and Cheek recommended the minimum
magnitude of MIIC as 0.20 for reliable results.
All students completing the scales of the study were
included in the study. However, there were missing data per
item. The portions of missing data ranged from 0.8% to
5.1% per item across the scales. We used multiple impu-
tations in spss 27 to fill in missing values in the data before
the CFA and SEM analyses [123].
Students reported their gender as male or female by
selecting a binary gender option. In the data, female
students were represented with 1, while male students
were coded with 2. Thus, gender entered the analysis as a
dichotomous variable, the baseline category of which is
female students. As the baseline category represents female
students, a positive sign in a correlation or a path coefficient
indicates that male students had higher scores than female
students.
IV. RESULTS
In this section, the results regarding the measurement
model were given; descriptive statistics and correlations
among study variables were presented, and the results
related to the hypothesized model were discussed.
A. Measurement models
Considering the recommendation of Anderson and
Gerbing [124], first, the measurement part of the proposed
SEM was examined by CFA to determine the degree to
which the hypothesized factor structure of metacognition,
epistemic cognition, and identity constructs fit the observed
data. As seen in the model given in Appendix A, the
indicators of metacognition and epistemic cognition are
subdimensions discussed in the previous sections. The
subdimensions were entered into the model as observed
variables whose scores were estimated by the sum of scores
across items significantly loading on their hypothesized
subdimensions. Before estimating the total scores, we
conducted CFA for the metacognition and epistemic
cognition scales to check whether all items were
YAREN ULU and SEVDA YERDELEN-DAMAR PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-8
significantly loaded on the respective subdimensions. In
addition, MIIC for each subconstruct was estimated to
evaluate the measurement error of the indicators. The CFA
results indicated that all items significantly loaded on the
hypothesized subconstructs. According to MIIC values,
the reliability level of each indicator was also satisfactory.
Appendix Bpresents the CFA and MIIC results for the
subdimensions. Thus, after ensuring a good level of
internal consistency among the items in each subscale,
the total score of each subdimension was calculated,
summing scores of all items intending to measure the
respective subdimension. Then, the second CFAs were
employed to test the eight-indictor measurement model of
metacognition, the five-indicator measurement model of
epistemic cognition, and the factor structure of the physics
identity framework (see Appendix A). The CFI, TLI,
SRMR, and RMSEA values for all scales are presented in
Tab le V. All fit indices suggest that the measurement
models fit the data well based on the aforementioned
cutoff values [118–120].
B. Descriptive statistics and correlations
The student’s total score for each construct was obtained
by summing the scores of all items measuring the construct
and dividing the total by the number of items. For example,
the total score on epistemic cognition was obtained by
adding the scores of 27 items and dividing the sum by 27.
The possible minimum and maximum scores that students
can have on physics identity, recognition, self-efficacy, and
interest are 0.00 and 10.0, respectively. The possible
minimum and maximum scores that students can obtain
on metacognition and epistemic cognition are 1.00 and
5.00, respectively. Table VI shows the mean, standard
TABLE V. Fit indices for the measurement model of metacognition, epistemic cognition, and physics identity.
Scale CFI TLI RMSEA (90% CI) SRMR
Metacognition 0.99 0.99 0.06 (0.046, 0.071) 0.01
Epistemic cognition 0.99 0.98 0.05 (0.028, 0.073) 0.02
Identity 0.97 0.97 0.06 (0.051, 0.063) 0.03
TABLE VI. Descriptive statistics, reliabilities, and correlations among all study variables. Correlations in bold are significant at the
0.001 level.
Minimum Maximum Mean SD
Reliability
(Cronbach’s alpha) 1 2 3 4 5 6 7
1. Gender 1
2. Identity 0.00 10.0 5.56 2.73 0.26 1
3. Recognition 0.00 10.0 4.86 2.57 0.88 0.19 0.87 1
4. Physics Self-Efficacy 0.00 10.0 5.82 2.40 0.92 0.28 0.82 0.88 1
5. Interest 0.00 10.0 5.38 2.92 0.95 0.22 0.75 0.72 0.82 1
6. Metacognition 1.00 5.00 3.38 0.63 0.96 0.02 0.38 0.40 0.45 0.42 1
7. Epistemic cognition 2.48 5.00 3.89 0.45 0.88 0.01 0.46 0.48 0.53 0.54 0.62 1
TABLE VII. Descriptive statistics according to gender.
Gender Variable Mean Standard deviation
Females Identity 4.96 2.65
Males 6.40 2.61
Females Interest 4.82 2.90
Males 6.17 2.78
Females Recognition 4.49 2.56
Males 5.39 2.50
Females Self-efficacy 5.29 2.39
Males 6.56 2.22
Females Metacognition 3.38 0.63
Males 3.38 0.63
METACOGNITION AND EPISTEMIC COGNITION …PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-9
deviation (SD), minimum, and maximum observed score
for each construct, reliabilities, and correlations among all
study variables. All mean scores were above the midpoint
of the scale except for recognition. The mean score of
recognition was slightly below the midpoint of the 11-point
Likert scale. Table VII presents the mean and standard
deviation of male and female students’scores on each
construct. The mean scores for male students were higher
than the midpoint for all constructs, whereas only the mean
of female students’self-efficacy scores was above the
midpoint; however, it was still smaller than that of male
students’self-efficacy.
The reliability analysis of the scales measuring the
related constructs revealed that Cronbach’s alpha was
bigger than 0.70 for each scale; thus, the internal consis-
tency reliability for each scale was satisfactory [121].
Table VI shows that all correlations are statistically
significant except the correlations of gender with meta-
cognition and epistemic cognition. As the reference group
of the dichotomous-gender variable was female students, a
significant positive correlation of gender with a variable
indicates that male students exhibited significantly higher
scores than female students on the variable. Thus, male
students had significantly higher scores than female
students on all constructs except metacognition and
epistemic cognition. Gender differences in the study
variables are discussed in more detail in Sec. IV.C.4.
Based on the cutoff values recommended by Cohen and
Cohen [115], there was a very high positive correlation
between identity, recognition, self-efficacy, and interest
constructs. In contrast, identity, recognition, self-efficacy,
and interest were moderately correlated with epistemic
cognition and metacognition. Finally, there was a high
positive correlation between metacognition and epistemic
cognition.
C. The analysis of the hypothesized
structural model
After establishing the construct validity of the question-
naire results with CFAs, the measurement models of the
constructs were combined in a single model, and the
hypothesized paths were added among the latent constructs
of the study. Then, SEM was performed to analyze the
resulting model (see Fig. 1). Considering the cutoff values,
all goodness of fit measures were within the acceptable
range, which suggested the proposed model adequately
fitted the data (CFI ¼0.95, TLI ¼0.94, RMSEA ¼0.05
(90% CI ¼0.050, 0.055) SRMR ¼0.04). Figure 2indi-
cates the model with significant and insignificant path
estimates in dashed lines and explained variances (R2)of
dependent variables. The model accounted for 80%, 78%,
69%, 38%, and 39% of the variance in physics identity,
recognition, interest, physics self-efficacy, and epistemic
FIG. 2. The model with the solid and dashed lines representing significant and nonsignificant relations, path estimates on lines, and the
explained variances.
YAREN ULU and SEVDA YERDELEN-DAMAR PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-10
TABLE VIII. Path coefficients and standard error of measurement (SE) for direct, indirect, and total relations. All path coefficients in bold are significant at 0.001, except the
coefficient of the path from gender to recognition is significant at 0.05.
Metacognition Epistemic Cognition Self-efficacy Recognition Interest Identity
Variables Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total Direct Indirect Total
Gender β0.02 0.02 0.00 0.01 0.01 0.27 0.01 0.28 −0.06 0.25 0.19 0.02 0.20 0.22 0.08 0.18 0.26
SE 0.03 0.03 0.03 0.02 0.03 0.02 0.02 0.03 0.02 0.03 0.03 0.02 0.02 0.03 0.02 0.03 0.03
Metacognition β0.62 0.62 0.18 0.26 0.44 0.40 0.40 0.42 0.42 0.00 0.38 0.38
SE 0.03 0.03 0.04 0.03 0.03 0.03 0.03 0.02 0.02 0.02 0.03 0.03
Epistemic Cognition β0.41 0.41 0.37 0.37 0.15 0.31 0.46 0.01 0.36 0.37
SE 0.04 0.04 0.04 0.04 0.03 0.03 0.04 0.03 0.04 0.04
Self-efficacy β0.90 0.90 0.74 0.74 0.78 0.78
SE 0.01 0.01 0.03 0.03 0.02 0.02
Recognition β0.68 0.68
SE 0.03 0.03
Interest β0.24 0.24
SE 0.03 0.03
METACOGNITION AND EPISTEMIC COGNITION …PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-11
cognition, respectively, corresponding to a large effect size
[115]. Table VIII indicates direct, indirect, and total
relations to the dependent variables in the structural model.
Among hypothesized direct relations to identity, only
recognition, interest, and gender had significant direct
relations to physics identity. In contrast, metacognition
and epistemic cognition did not have significant direct
associations with physics identity. On the other hand, as
seen from Table VIII, all indirect and total relations to
physics identity reached statistical significance, suggesting
the existence of mediating relations we discussed in the
following sections.
1. Relationships among identity constructs
This study observed path coefficients that are in good
agreement with those found by Duo and Cain [17]. Physics
identity was significantly predicted by interest (β¼0.24)
and recognition (β¼0.68). Based on recommended effect
size measures for path coefficients, the relations of interest
and recognition had a small to medium and large effect size,
respectively. Physics self-efficacy had a significant indirect
relation to physics identity through interest and recognition
(β¼0.78, Bootstrap % 95 CI ¼0.74, 0.82). This indirect
relation had a very large effect size. Physics self-efficacy
significantly predicted recognition (β¼0.90) and interest
(β¼0.73). Both relations had a very large effect size.
2. The relationship between metacognition
and physics identity
Metacognition did not directly predict physics identity
(β¼−0.001) in the model. In contrast, the indirect relation
through epistemic cognition, physics self-efficacy, recog-
nition, and interest was significant (β¼0.38, Bootstrap %
95 CI ¼0.32, 0.42), which made the total contribution of
metacognition to physics identity significant (β¼0.37).
This indirect relation had a medium to large effect size. As
the direct relation of epistemic cognition on physics
identity was insignificant, the mediating role of epistemic
cognition was not significant in this total indirect relation.
The specific indirect relation of metacognition to physics
identity through epistemic cognition was insignificant
(β¼0.007). When other specific indirect relations of
metacognition were examined, it was seen that indirect
relation through recognition, physics self-efficacy, and
epistemic cognition contributed mainly to total indirect
relation (β¼0.16, Bootstrap % 95 CI ¼0.13, 0.20). These
results suggest the importance of a mediating role of
physics self-efficacy in the association between metacog-
nition and physics identity.
3. The relationship between epistemic cognition and
physics identity
The direct contribution of epistemic cognition to physics
identity was insignificant (β¼0.01). On the other hand, its
indirect relation through physics self-efficacy, interest, and
recognition was significant (β¼0.36, Bootstrap % 95
CI ¼0.30, 0.43). The effect size of this relation was
medium to large. The specific indirect relation through
interest was significant (β¼0.04, Bootstrap % 95
CI ¼0.02, 0.05), but its size was very small. On the other
hand, the specific indirect relation through self-efficacy and
recognition was significant (β¼0.25, Bootstrap % 95
CI ¼0.20, 0.31) and had a small to medium effect size.
Similarly, these results supported the mediating effect of
physics self-efficacy on the relation of epistemic cognition
to physics identity.
4. Gender differences in physics identity
and other study variables
A significant total relation of gender to a variable
indicates that one group reported significantly higher scores
than the other group in the variable. In contrast, a
significant direct relation of gender to a variable suggests
there would be a significant gender difference in the
variable after controlling for other predictors of that
variable, which is similar to a significant difference
obtained with the analysis of covariance. Thus, splitting
the total relation into direct and indirect relations helps us to
see this distinction.
The gender analysis on physics self-efficacy revealed
that gender had a significant direct (β¼0.27) and total
relation (β¼0.28) to physics self-efficacy with a medium
effect size. In contrast, its indirect relation to self-efficacy
through metacognition and epistemic cognition was
almost nonexistent (β¼0.01). That is, in terms of the
total relation, male students had significantly higher
scores in physics self-efficacy than female students,
and after controlling for metacognition and epistemic
cognition, male students would still have higher self-
efficacy scores.
Gender also had a significant direct relation to physics
identity (β¼0.08), with a small effect size after con-
trolling for other predictors. After adjusting for other
predictors, male students would have slightly higher
physics identity than female students. However, as seen
from Table VIII, according to the total relation, male
students possessed significantly higher physics identity
scores with a medium effect size (β¼0.26, Bootstrap %
95 CI ¼0.21, 0.32). This total relation mainly arose from
the significant indirect relation of gender to physics
identity through other study variables (β¼0.18,boot-
strap % 95 CI ¼0.13,0.23).Asaresultoftheindirect
relations, we observed significant and notable gender
differences in the identity scores of male and female
students. When the direct relations of gender were
inspected (see Fig. 2), it was seen that the significant
indirect relation mainly occurred because of the direct
relation of gender to physics self-efficacy. In other words,
gender-related differences in physics identity might
YAREN ULU and SEVDA YERDELEN-DAMAR PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-12
mainly be explained by gender differences in physics
self-efficacy.
The mediating effect of physics self-efficacy was also
observed in the relationship between gender and recog-
nition, gender and interest. The total relation of gender to
recognition was significant and positive with a small to
medium effect size (β¼0.19). That is, female students
had significantly less recognition. However, according
to the direct relation (β¼−0.06), female students would
have slightly higher recognition after controlling for
self-efficacy. (β¼0.25, bootstrap % 95 CI ¼0.19,
0.30). In other words, if female students got higher
physics self-efficacy, their perceptions of recognition
would be better.
Likewise, a significant indirect effect of gender through
physics self-efficacy was observed on interest in favor of
male students (β¼0.20, bootstrap % 95 CI ¼0.15, 0.25),
which in turn led the total relation to be positive and small
to medium size (β¼0.22). Thus, based on the total
relation, male students indicated a significantly higher
interest in physics. However, gender did not directly relate
significantly to interest (β¼0.02). If male and female
students did not differ in physics self-efficacy, they would
indicate similar interest in physics.
Finally, we did not observe a significant direct and total
relation of gender to metacognition and epistemic cognition
(see Table VIII). Thus, male and female students did not
differ in metacognition and epistemic cognition.
Gender differences observed in the present study may
raise the question of whether there is an equivalence of the
structural regression paths for females and males. Thus, a
moderation analysis (a multigroup analysis of structural
invariance) for gender was also conducted to answer this
question. The analysis results (see Appendix C) supported
the invariance of the structural model for males and
females.
5. The relation of metacognition to other
study variables
Metacognition significantly predicted both physics
self-efficacy (β¼0.18) with a small to medium effect
size and epistemic cognition (β¼0.62) with a large effect
size. Although the direct relation of metacognition on
physics self-efficacy was small, its indirect relation
through epistemic cognition was small to medium
(β¼0.26, bootstrap % 95 CI ¼0.20,0.33),whichmade
the total relation of being a medium to large effect
size (β¼0.44).
Metacognition was also indirectly related to recognition
(β¼0.40, bootstrap % 95 CI ¼0.35, 0.45) and interest
(β¼0.42, bootstrap % 95 CI ¼0.37, 0.46) due to its
direct relation to physics self-efficacy and epistemic
cognition. Both relations had a medium to large
effect size.
6. The relation of epistemic cognition to physics self-
efficacy, interest, and recognition
Epistemic cognition had a significant direct relation to
physics self-efficacy (β¼0.41, a medium to large effect
size) and to interest (β¼0.15, a small effect size). Its
indirect relation to interest through physics self-efficacy
was also significant (β¼0.30, bootstrap % 95 CI ¼0.25,
0.37) with a medium effect size, increasing the total relation
of epistemic cognition to interest (β¼0.45). Thus, physics
self-efficacy also mediated the relationship between epi-
stemic cognition and interest.
Moreover, epistemic cognition indirectly contributed to
recognition through physics self-efficacy (β¼0.37, boot-
strap % 95 CI ¼0.30, 0.45) with a medium to large
effect size.
V. DISCUSSION
The analysis of Turkish high school students’data indicated
that path coefficients among physics self-efficacy, recogni-
tion, interest, and identity were similar to those observed for
different groups of students [2,17,18]. In addition, the
significant direct and indirect relations of metacognition,
epistemic cognition, and gender to physics identity constructs
were found in this study, which motivates future experimental
studies to test the causal relations in the model. Thus,
assuming the proposed model of the current study is correct,
the following implications would be suggested:
Previous research studies pointed out the vital role of
metacognition in developing epistemic cognition, self-
efficacy, and identity [32,42,43,60,61,64,67]. In line with
previous research findings, the results of the current
study showed that metacognition predicted epistemic
cognition and physics self-efficacy directly and physics
identity indirectly through physics self-efficacy. Students
with higher metacognition also exhibited sophisticated
epistemic cognition, physics self-efficacy, and identity.
Therefore, instruction incorporating metacognition-
enhancing activities may support fostering epistemic
cognition and physics self-efficacy, which in turn could
promote physics identity. For example, in an inquiry-
based physics curriculum, Yerdelen-Damar and Eryılmaz
[52] employed several metacognitive strategies, such as
the metacognitively prompted small and whole group
discussions, predict-observe-explain strategy, error
analysis, and journal writing, to trigger students to
engage in metacognitive thinking regarding their con-
ceptual understandings and epistemic cognition. The
researchers found a significant improvement in students’
conceptual and epistemic understandings [52,61].Inthe
same curriculum, the researchers also observed that
the experimental group students exhibited higher self-
efficacy than the control group students [65].
Likewise, epistemic cognition significantly predicted
identity formation variables (physics self-efficacy,
METACOGNITION AND EPISTEMIC COGNITION …PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-13
recognition, and interest), similar to the results of previous
studies [77,78,87,89]. Epistemic cognition was not
directly related to physics identity; it indirectly predicted
it via the mediation of physics self-efficacy. Similarly,
Guo et al. found a significant indirect relationship
between chemistry identity and epistemic cognition
through self-efficacy [76]. In addition, it was also
observed that epistemic cognition in physics was medi-
ating the relation of metacognition to physics self-effi-
cacy. Based on these findings, epistemic cognition-
enhancing strategies could promote identity formation
variables, which in turn might foster physics identity.
They could also increase the effect of metacognitive
strategies on physics self-efficacy. However, research
revealed that implicitly addressing students’epistemic
cognition does not facilitate students’views about the
nature of physics knowledge, knowing, and learning
[125–127]. Therefore, it is recommended that the instruc-
tion should explicitly consider students’epistemic cog-
nition. Several studies provided evidence indicating the
effectiveness of direct interventions [61,128–131].For
instance, Redish and Hammer [131] designed an intro-
ductory algebra-based physics course explicitly address-
ing students’epistemic cognition to help students view
physics learning as the reconciliation of everyday intuitive
thinking and physics knowledge as a coherent system of
ideas. Epistemic cognition-enhancing activities used by
the researchers were explicit epistemic discussions,
epistemically modified peer instruction, and interactive
lecture demonstrations and homework assignments,
prompting students to reflect on the nature of physics
learning and knowledge. The result of the study indicated
that students demonstrated significant gains on an epi-
stemic cognition survey.
The fact that epistemic cognition and metacognition
were only indirectly related to physics identity through
self-efficacy could suggest that the effectiveness of meta-
cognition and epistemic cognition-enhancing activities in
developing physics identity may rely on students’level
of self-efficacy. In those interventions, students should
also feel confident in their abilities to achieve physics-
related tasks to build their physics identity. Thus, for
effective identity construction, metacognition or epistemic
cognition-enhancing interventions can also be enriched
with self-efficacy-supporting strategies aligning with
Bandura’s[14] four sources of self-efficacy: enactive
mastery experience, vicarious experience, social persua-
sion, and physiological and affective states. Vicarious
experiences through modeling [87,132–135], anxiety
coping strategies [133,136], providing positive feedback
about performance [132,133], providing a learner-friendly
environment in which every student can ask questions and
express their ideas [132], adjusting assignments according
to student’s level of understanding [132], providing
scaffolding [137–139] are some of the research-proven
self-efficacy-supporting strategies.
The gender differences observed in study variables also
align with previous studies. Both male and female students
demonstrated similar metacognition and epistemic cogni-
tion [70,104]. When considering total relation (direct
plus indirect relation), we observed a similar gender effect.
The male students had higher scores in physics identity,
self-efficacy, interest, and recognition [1,13,93,96–100].
However, when the mediation analysis results were further
inspected, the mediating role of physics self-efficacy in
gender differences was observed in other identity con-
structs. The superiority of male students in physics identity,
interest, and recognition might stem from their superiority
in physics self-efficacy. Therefore, physics self-efficacy
could be a key to overcoming gender differences in physics
identity constructs. However, a literature review study
conducted by Henderson et al. [140] indicated that, in
general, traditional physics courses negatively influenced
students’physics self-efficacy. In addition, male students
reported higher self-efficacy in physics courses, and gender
differences tended to increase after physics instruction
[140]. Thus, we need special teaching strategies and
classroom activities to address students’self-efficacy and
decrease gender differences. Sawtelle et al. [135] found that
modeling instruction, including collaborating learning
environments, positively influenced female students’self-
efficacy. Similarly, Espinosa et al. [141] indicated that a
project-based introductory physics class, including inquiry-
driven projects blended with peer instruction, tutorials,
estimation, and experimental design activities and problem
sets reduced the gender gap in physics self-efficacy.
Furthermore, Hazari et al. [1] recommended emphasizing
conceptual understanding and real-world or contextual
relevance in physics instruction, and the discussion of
women’s underrepresentation in science can promote
females’physics self-efficacy.
Finally, the current study had some limitations. First,
the study was correlational. Thus, in contrast to an
experimental study, the present study did not establish
causal connections among study variables. In addition,
there are also other prior studies using different SEM
models for physics identity [142–144]. Thus, future
experimental studies can be conducted to test the relation-
ships casually suggested in SEM studies on physics
identity [144,145]. Second, the model in the current study
did not include teacher or learning environment-related
variables. Future research can also include such variables
in the proposed model of the present study. Third,
instrumentation decay [146] could be a threat to the
results of the study. Survey fatigue might occur during
the study as multiple surveys were simultaneously admin-
istered to the students. However, this threat was controlled
by selecting a schedule when students felt more energized
during the earlier hours of school. The researchers gave
YAREN ULU and SEVDA YERDELEN-DAMAR PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-14
students one class hour to ensure they took time while
answering the items. In addition, the epistemic cognition
survey, placed at the end of the survey list, included
reverse items to check if students read all items. Fourth,
the researchers did not gather data regarding the nonat-
tendance rate during the data collection period. However,
the data were collected after the COVID period, on regular
class days when no exams were coming up, and the
lectures continued. During the application of the surveys,
the first author received no comments from teachers about
a high nonattendance rate in the classes where students
completed the surveys. Furthermore, students in partici-
pating schools were not informed beforehand whether
there would be data collection in the upcoming days.
Therefore, we assumed that the nonattendance rate, which
might threaten the results observed in this study, was
negligible and random. Finally, gender data were collected
with only binary options (male versus female). That
binary model of gender may limit the generalizability
of gender-related conclusions in this study and prevent
drawing more profound conclusions about gender
differences discussed elsewhere [147].
VI. CONCLUSION
This study, analyzing the data from Turkish high school
students, observed relations among physics self-efficacy,
recognition, interest, and physics identity consistent with
those found in other contexts. Recognition and interest
directly predicted physics identity and mediated the relation
of physics self-efficacy to it. The study extended the current
literature by investigating the relation of epistemic cognition
and metacognition to physics identity and identity-formation
constructs, which could motivate future experimental inter-
ventions to promote the understanding of physics identity
development. The results revealed that metacognition and
epistemic cognition in physics indirectly predict physics
identity through the mediation of physics self-efficacy, which
suggests that metacognition and epistemic cognition-enhanc-
ing strategies could be used to foster physics identity;
however, the success of those strategies may depend on
the sophistication of physics self-efficacy, which further
motivates the design of experimental studies to test these
relations. The study also observed significant direct and
indirect relations among metacognition, epistemic cognition,
self-efficacy, recognition, and interest. Metacognition directly
predicted epistemic cognition and physics self-efficacy. It also
indirectly contributed to recognition and interest through
physics self-efficacy and epistemic cognition. Epistemic
cognition was directly associated with physics self-efficacy
and interest and partially mediated the relationship between
physics self-efficacy and metacognition. It also indirectly
contributed to interest and recognition via the mediating
relation of physics self-efficacy. These significant interrela-
tions could further highlight the indirect contribution of
metacognition and epistemic cognition to the formation of
physics identity. Gender differences were also observed in the
current study. Male students scored higher than female
students in physics identity, self-efficacy, recognition, and
interest. However, the mediation analysis further indicated
that gender differences in physics self-efficacy might explain
gender differences in physics identity, recognition, and
interest, which advocates experimental interventions to test
whether practices that reduce the gender gap in physics self-
efficacy will also eliminate the gender gap in physics identity,
recognition, and interest.
ACKNOWLEDGMENTS
This study is produced from the first author’s Master’s
thesis.
APPENDIX A
The entire hypothesized model of this study.
METACOGNITION AND EPISTEMIC COGNITION …PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-15
YAREN ULU and SEVDA YERDELEN-DAMAR PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-16
APPENDIX B
As the data on epistemic cognition and metacognition
surveys were collected with a five-point Likert scale, weighted
least squares mean- and variance-adjusted (WLSMV) esti-
mator in Mplus 8.10 was used for the CFA analyses [116].The
CFA results supported the factor structures of the surveys. The
goodness of fit indices; CFI ¼0.95,TLI¼0.93,RMSEA¼
0.05 (90%CI ¼0.048, 0.055) SRMR ¼0.04 for epistemic
cognition survey; CFI ¼0.93,TLI¼0.93,RMSEA¼0.04
(90%CI ¼0.041 0.044)SRMR¼0.04.
CFA results of the Epistemic Cognition Scale
Factor loading SE pvalue MIIC
Structure of knowledge coherence
Item3 0.74 0.02 0.000 0.28
Item4 0.34 0.03 0.000
Item6 0.81 0.02 0.000
Item7 0.44 0.03 0.000
Item9 0.64 0.02 0.000
Structure of knowledge hierarchical
Item1 0.39 0.02 0.000 0.30
Item2 0.77 0.02 0.000
Item5 0.55 0.02 0.000
Item8 0.81 0.02 0.000
Justification of knowledge
Item10 0.73 0.02 0.000 0.29
Item11 0.56 0.03 0.000
Item12 0.62 0.02 0.000
Item13 0.67 0.02 0.000
Item14 0.52 0.03 0.000
Changeability of knowledge
Item15 0.20 0.03 0.000 0.26
Item16 0.68 0.02 0.000
Item17 0.74 0.02 0.000
Item18 0.32 0.03 0.000
Item19 0.79 0.02 0.000
Source of knowledge
Item20 0.33 0.03 0.000 0.35
Item21 0.43 0.03 0.000
Item22 0.46 0.03 0.000
Item23 0.69 0.03 0.000
Quick learning
Item24 0.81 0.02 0.000 0.29
Item25 0.32 0.03 0.000
Item26 0.66 0.02 0.000
Item27 0.70 0.02 0.000
CFA results of the Metacognition Scale
Factor loading SE pvalue MIIC
Declarative knowledge
Item5 0.49 0.02 0.000 0.34
Item10 0.68 0.02 0.000
Item12 0.69 0.02 0.000
Item16 0.59 0.02 0.000
Item17 0.57 0.02 0.000
Item20 0.70 0.02 0.000
Item32 0.68 0.02 0.000
Item46 0.56 0.02 0.000
Procedural knowledge
Item3 0.61 0.02 0.000 0.39
Item14 0.67 0.02 0.000
Item27 0.76 0.01 0.000
Item33 0.66 0.02 0.000
Conditional knowledge
Item15 0.53 0.02 0.000 0.33
Item18 0.71 0.02 0.000
Item26 0.60 0.02 0.000
Item29 0.64 0.02 0.000
Item35 0.67 0.02 0.000
Planning
Item4 0.52 0.02 0.000 0.30
Item6 0.62 0.02 0.000
Item8 0.57 0.02 0.000
Item22 0.61 0.02 0.000
Item23 0.67 0.02 0.000
Item42 0.54 0.02 0.000
Item45 0.60 0.02 0.000
Information management strategies
Item9 0.51 0.02 0.000 0.28
Item13 0.70 0.02 0.000
Item30 0.74 0.01 0.000
Item31 0.67 0.02 0.000
Item37 0.37 0.03
Item39 0.67 0.02
Item41 0.59 0.02
Item43 0.65 0.02
Item47 0.52 0.02
Item48 0.36 0.03
Comprehension monitoring
Item1 0.59 0.02 0.000 0.34
Item2 0.63 0.02 0.000
Item11 0.64 0.02 0.000
Item21 0.66 0.02 0.000
Item28 0.65 0.02 0.000
Item34 0.49 0.02 0.000
Item49 0.65 0.02 0.000
(Table continued)
METACOGNITION AND EPISTEMIC COGNITION …PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-17
(Continued)
Factor loading SE pvalue MIIC
Debugging strategies
Item25 0.47 0.03 0.000 0.29
Item40 0.71 0.02 0.000
Item44 0.69 0.02 0.000
Item51 0.38 0.03 0.000
Item52 0.47 0.02 0.000
Evaluation
Item7 0.40 0.02 0.000 0.28
Item19 0.57 0.02 0.000
Item24 0.55 0.02 0.000
Item36 0.66 0.02 0.000
Item38 0.61 0.02 0.000
Item50 0.61 0.02 0.000
APPENDIX C
The multigroup invariance analysis of the proposed model
for males and females was conducted following the guidelines
of Byrne [148]. First, the baseline model was tested for female
students. The baseline model fits the data well (CFI ¼0.947,
TLI ¼0.941, and RMSEA ¼0.055). Second, the configural
model without parameter constraints was tested (CFI ¼
0.942,TLI¼0.939, and RMSEA ¼0.055). The inspection
of significant and insignificant regression coefficients across
the two groups indicated that the pattern was the same for
females and males. Third, the equivalence of the model for
females and males was tested (CFI ¼0.942,TLI¼0.940,
and RMSEA ¼0.054). That the value of CFI remained
unchanged from that observed in the configural model and
the change in the value of RMSEAwas not greater than 0.015
[149,150] across two models supported the equivalence of
path coefficients for males and females in this study.
[1] Z. Hazari, G. Sonnert, P. M. Sadler, and M.-C.
Shanahan, Connecting high school physics experiences,
outcome expectations, physics identity, and physics
career choice: A gender study, J. Res. Sci. Teach. 47,
978 (2010).
[2] H. Cheng, G. Potvin, R. Khatri, L. H. Kramer, R. M.
Lock, and Z. Hazari, Physics Education Research
Conference on Examining physics identity development
through two high school interventions, in Proceedings of
the 2018 Physics Education Research Conference, Wash-
ington, DC (AIP, New York, 2018).
[3] A. C. Barton and E. Tan, We be burnin’! Agency, identity,
and science learning, J. Learn. Sci. 19, 187 (2010).
[4] PISA 2024 Strategic Direction, and Vision for Science
(OECD Publishing, Paris, 2020).
[5] P. J. Mulvey and S. Nicholson, Physics Bachelor’s
Degrees: 2018. Results from the 2018 Survey of Enroll-
ments and Degrees (AIP Statistical Research Center,
College Park, MD, 2020).
[6] Physics Students in UK Universities (n.d.). Retrieved
January 1, 2023, from https://www.iop.org/sites/default/
files/2021-12/Physics-Students-in-UK-Universities-
HESA-Data-Brief.pdf.
[7] The STEM need in Turkey for 2023, Turkish Industry &
Business Association, Retrieved January 1, 2023, from
https://tusiad.org/en/reports/item/9754-the-stem-need-in-
turkey-for-2023.
[8] R. M. Lock, Z. Hazari, and G. Potvin, Impact of out-of-
class science and engineering activities on physics iden-
tity and career intentions, Phys. Rev. Phys. Educ. Res. 15,
020137 (2019).
[9] Ö. Özkurt and I. Yakın, 2013–2019 yıllarıarasında
Türkiye’deki üniversitelerin STEM alanlarında kayıtlı
öğrenci sayılarının cinsiyet bağlamında karşılaştırılması,
J. Soc. Sci. Hum 7, 68 (2020).
[10] Talent 2030 Dashboard 2018, National Centre for Uni-
versities and Business, 2018, http://www.ncub.co.uk/
reports/talent-2030-dashboard-2018.
[11] J. P. Gee, Identity as an analytic lens for research in
education, Rev. Res. Educ. 25, 99 (2000).
[12] H. B. Carlone and A. Johnson, Understanding the science
experiences of successful women of color: Science
identity as an analytic lens, J. Res. Sci. Teach. 44,
1187 (2007).
[13] R. M. Lock, Z. Hazari, and G. Potvin, Physics career
intentions: The effect of physics identity, math identity,
and gender, AIP Conf. Proc. 1513, 262 (2013).
[14] A. Bandura, Self-Efficacy: The Exercise of Control (Free-
man, New York, 1997).
[15] J. D. Cribbs, Z. Hazari, G. Sonnert, and P. M. Sadler,
Establishing an explanatory model for mathematics iden-
tity, Child Dev. 86, 1048 (2015).
[16] A. Godwin, G. Potvin, Z. Hazari, and R. Lock, Identity,
critical agency, and engineering: An affective model for
predicting engineering as a career choice, J. Eng. Educ.
105, 312 (2016).
[17] R. Dou and H. Cian, Constructing STEM identity: An
expanded structural model for STEM identity research, J.
Res. Sci. Teach. 59, 458 (2022).
[18] D. Verdín, The power of interest: Minoritized women’s
interest in engineering fosters persistence beliefs beyond
belongingness and engineering identity, Int. J. STEM
Educ. 8, 33 (2021).
[19] A. L. Brown, J. D. Bransford, R. Ferrara, and J.
Campione, Learning, remembering and understanding,
Handbook of Child Psychology: Vol. 3. Cognitive Devel-
YAREN ULU and SEVDA YERDELEN-DAMAR PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-18
opment, 4th ed., edited by J. H. Flavell and E. M.
Markman (Wiley, New York, 1983), pp. 77–166.
[20] A. L. Brown, Metacognition, executive control, self-
regulation, and other more mysterious mechanisms, in
Metacognition, Motivation, and Understanding, edited by
F.E. Weinert and R. Kluwe (Lawrence Erlbaum Asso-
ciates, Hillsdale, NJ, 1987).
[21] G. Schraw and R. S. Dennison, Assessing metacognitive
awareness, Contemp. Educ. Psychol. 19, 460 (1994).
[22] B. K. Hofer and P. R. Pintrich, The development of
epistemological theories: Beliefs about knowledge and
knowing and their relation to learning, Rev. Educ. Res.
67, 88 (1997).
[23] A. Elby, Defining personal epistemology: A response to
Hofer & Pintrich (1997) and Sandoval (2005), J. Learn.
Sci. 18, 138 (2009).
[24] I. A. Halloun and D. Hestenes, Views about Sciences
Survey: VASS, in Proceedings of the Sociology Paper
presented at NARST, St. Louis, Missouri (1996), pp. 266–
267.
[25] D. Hammer, Epistemological beliefs in introductory
physics, Cognit. Instr. 12, 151 (1994).
[26] W. K. Adams, K. K. Perkins, N. S. Podolefsky, M.
Dubson, N. D. Finkelstein, and C. E. Wieman, New
instrument for measuring student beliefs about physics,
and learning physics: The Colorado Learning Attitudes
about Science Survey, Phys. Rev. ST Phys. Educ. Res. 2,
010101 (2006).
[27] K. Özmen and Ö. F. Özdemir, Conceptualisation, and
development of the physics related personal epistemology
questionnaire (PPEQ), Int. J. Sci. Educ. 41, 1207 (2019).
[28] M. Welsh and S. Schmitt-Wilson, Executive function,
identity, and career decision-making in college students,
SAGE Open 3, 2158244013505755 (2013).
[29] J. E. Marcia, Identity in adolescence, in Handbook of
Adolescent Psychology (Wiley and Sons, New York,
1980), Vol. 9, p. 159.
[30] J. H. Flavell, Metacognition and cognitive monitoring: A
new area of cognitive developmental inquiry, Am.
Psychol. 34, 906 (1979).
[31] P. R. Pintrich, The role of metacognitive knowledge in
learning, teaching, and assessing, Theory Pract. 41, 219
(2002).
[32] P. Irving and E. Sayre, Physics identity development: A
snapshot of the stages of development of upper-level
physics students, J. Scholarship Teach. Learn. 13,68
(2013).
[33] S. Basu and S. Dixit, Role of metacognition in explaining
decision-making styles: A study of knowledge about
cognition and regulation of cognition, Pers. Individ. Diff.
185, 111318 (2022).
[34] K. Batha and M. Carroll, Metacognitive training aids
decision making, Aust. J. Psychol. 59, 64 (2007).
[35] R. Griffith, M. Bauml, and S. Quebec-Fuentes, Promoting
metacognitive decision-making in teacher education,
Theory Pract. 55, 242 (2016).
[36] M. D. Berzonsky and K. Luyckx, Identity styles, self-
reflective cognition, and identity processes: A study of
adaptive and maladaptive dimensions of self-analysis,
Identity 8, 205 (2008).
[37] K. Luyckx, B. Soenens, M.D. Berzonsky, I. Smits, L.
Goossens, and M. Vansteenkiste, Information-oriented
identity processing, identity consolidation, and well-being:
The moderating role of autonomy, self-reflection, and self-
rumination. Pers. Individ. Diff. 43, 1099 (2007).
[38] A. M. Grant, Rethinking psychological mindedness:
Metacognition, self-reflection, and insight. Behav.
Change 18, 8 (2001).
[39] A. Graham and R. Phelps, Being a teacher: Developing
teacher identity and enhancing practice through meta-
cognitive and reflective learning processes. Aust. J.
Teach. Educ. 27, 11 (2003).
[40] L. McAlpine, C. Weston, J. Beauchamp, C. Wiseman, and
C. Beauchamp, Building a metacognitive model of
reflection. High. Educ. 37, 105 (1999).
[41] R. Yuan and P. Mak, Reflective learning and identity
construction in practice, discourse and activity: Experi-
ences of preservice language teachers in Hong Kong,
Teach. Teach. Educ. 74, 205 (2018).
[42] E. W. Close, J. Conn, and H. G. Close, Becoming physics
people: Development of integrated physics identity
through the Learning Assistant experience, Phys. Rev.
Phys. Educ. Res. 12, 010109 (2016).
[43] H. Huvard, R. M. Talbot, H. Mason, A. N. Thompson, M.
Ferrara, and B. Wee, Science identity and metacognitive
development in undergraduate mentor-teachers. Int. J.
STEM Educ. 7, 31 (2020).
[44] D. Cañabate, T. Serra, R. Bubnys, and J. Colomer,
Preservice teachers’reflections on cooperative learning:
Instructional approaches and identity construction. Sus-
tainability 11, 5970 (2019).
[45] C. Beauchamp and L. Thomas, Understanding teacher
identity: An overview of issues in the literature and
implications for teacher education, Cambridge J. Educ.
39, 175 (2009).
[46] E. Wenger, Communities of Practice: Learning, Meaning,
and Identity (Cambridge University Press, Cambridge,
England, 1999).
[47] O. Levrini, M. Levin, and P. Fantini, Fostering appropria-
tion through designing for multiple access points to a
multidimensional understanding of physics, Phys. Rev.
Phys. Educ. Res. 16, 020154 (2020).
[48] K. Silseth and H. C. Arnseth, Learning and identity
construction across sites: A dialogical approach to ana-
lysing the construction of learning selves, Cult. Psychol.
17, 65 (2011).
[49] M. Varelas, D. B. Martin, and J. M. Kane, Content
learning and identity construction: A framework to
strengthen African American students’mathematics
and science learning in urban elementary school, Hum.
Dev. 55, 319 (2013).
[50] I. Adler, M. Zion, and Z. R. Mevarech, The effect of
explicit environmentally oriented metacognitive guid-
ance and peer collaboration on students’expressions of
environmental literacy, J. Res. Sci. Teach. 53,620
(2016).
[51] P. Georghiades, Making pupils’conceptions of electricity
more durable by means of situated metacognition, Int. J.
Sci. Educ. 26, 85 (2004).
[52] S. Yerdelen-Damar and A. Eryılmaz, Promoting concep-
tual understanding with explicit epistemic intervention in
METACOGNITION AND EPISTEMIC COGNITION …PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-19
metacognitive instruction: Interaction between the treat-
ment and epistemic cognition, Res. Sci. Educ. 51, 547
(2021).
[53] N. Yuruk, M. E. Beeth, and C. Andersen, Analyzing the
effect of metaconceptual teaching practices on students’
understanding of force and motion concepts, Res. Sci.
Educ. 39, 449 (2009).
[54] L. D. Bendixen and D. C. Rule, An integrative approach
to personal epistemology: A guiding model, Educ.
Psychol. 39, 69 (2004).
[55] L. D. Bendixen and F. C. Feucht, Personal epistemology
in the classroom: What does research and theory
tell us and where do we need to go next? in Personal
Epistemology in the Classroom: Theory, Research, and
Implications for Practice,editedbyL.D.Bendixenand
F. C. Feucht (Cambridge University Press, Cambridge,
England, 2010).
[56] A. Elby and D. Hammer, Epistemological resources
and framing: A cognitive framework for helping teach-
ers interpret and respond to their students’epistemol-
ogies, in Personal Epistemology in the Classroom:
Theory, Research, and Implications for Practice,edited
by L. D. Bendixen and F. C. Feucht (Cambridge
University Press, Cambridge, England, 2010),
pp. 409–434.
[57] K. R. Muis, The role of epistemic beliefs in self-regulated
learning, Educ. Psychol. 42, 173 (2007).
[58] D. C. Rule and L. D. Bendixen, The integrative model of
personal epistemology development: Theoretical under-
pinnings and implications for education, in Personal
Epistemology in the Classroom: Theory, Research, and
Implications for Practice, edited by L. D. Bendixen and
F. C. Feucht (Cambridge University Press, Cambridge,
England, 2010), pp. 94–123.
[59] J. Brownlee, N. Purdie, and G. Boulton-Lewis, Changing
epistemological beliefs in preservice teacher education
students, Teach. High. Educ. 6, 247 (2001).
[60] C. S. Kalman, M. Sobhanzadeh, R. Thompson, A.
Ibrahim, and X. Wang, Combination of interventions
can change students? epistemological beliefs, Phys. Rev.
ST Phys. Educ. Res. 11, 020136 (2015).
[61] S. Yerdelen-Damar and A. Eryılmaz, The impact of the
metacognitive 7E learning cycle on students’epistemo-
logical understandings, Kastamonu Educ. J. 24, 603
(2016).
[62] T. T. Moores, J. C. J. Chang, and D. K. Smith, Clarifying
the role of self-efficacy and metacognition as predictors of
performance: Construct development and test, ACM
SIGMIS Database: The DATABASE for Advances in
Information, Systems 37, 125 (2006).
[63] A. M. Schmidt and J. K. Ford, Learning within a learner
control training environment: The interactive effects of
goal orientation and metacognitive instruction on learning
outcomes, Pers. Psychol. 56, 405 (2003).
[64] K. J. Graham, C. M. Bohn-Gettler, and A. F. Raigoza,
Metacognitive training in chemistry tutor sessions in-
creases first year students’self-efficacy, J. Chem. Educ.
96, 1539 (2019).
[65] S. Yerdelen-Damar and A. Eryılmaz, The impact of the
metacognitive inquiry-based instruction on physics
self-efficacy, in Proceedings of the Applied Education
Congress (APPED), Ankara, Turkey (2012), https://fedu
.metu.edu.tr/tr/system/files/ReportsAndDocuments/aped
.pdf.
[66] J. L. Nietfeld, L. Cao, and J. W. Osborne, The effect of
distributed monitoring exercises and feedback on perfor-
mance, monitoring accuracy, and self-efficacy, Metacogn.
Learn. 1, 159 (2006).
[67] A. Taghani and M. R. Razavi, The effect of metacognitive
skills training of study strategies on academic self-
efficacy and academic engagement and performance of
female students in Taybad, Curr. Psychol. 41, 8784
(2022).
[68] S. A. Coutinho and G. Neuman, A model of metacog-
nition, achievement goal orientation, learning style
and self-efficacy, Learn. Environ. Res. 11, 131 (2008).
[69] R. Cera, M. Mancini, and A. Antonietti, Relationships
between metacognition, self-efficacy and self-
regulation in learning. J. Educ. Cult. Psychol. 7, 115
(2013).
[70] S. Yerdelen-Damar and H. Peşman, Relations of
gender and socioeconomic status to physics through
metacognition and self-efficacy. J. Educ. Res. 106, 280
(2013).
[71] M. C. Boyes and M. Chandler, Cognitive development,
epistemic doubt, and identity formation in adolescence, J.
Youth Adolesc. 21, 277 (1992).
[72] T. Krettenauer, The role of epistemic cognition in ado-
lescent identity formation: Further evidence, J. Youth
Adolesc. 34, 185 (2005).
[73] O. Levrini, P. Fantini, G. Tasquier, B. Pecori, and M.
Levin, Defining and operationalizing appropriation for
science learning, J. Learn. Sci. 24, 93 (2015).
[74] P. W. Irving and E. C. Sayre, Identity statuses in upper-
division physics students, Cult. Stud. Sci. Educ. 11, 1155
(2016).
[75] C. J. Faber, R. L. Kajfez, D. M. Lee, L. C. Benson, M. S.
Kennedy, and E. G.Creamer, A grounded theory model of
the dynamics of undergraduate engineering students’
researcher identity and epistemic thinking, J. Res. Sci.
Teach. 59, 529 (2022).
[76] X. Guo, X. Hao, W. Deng, X. Ji, S. Xiang, and W. Hu,
The relationship between epistemological beliefs, re-
flective thinking, and science identity: A structural
equation modeling analysis, Int. J. STEM Educ. 9,40
(2022).
[77] S. Kapucu and E. Bahçivan, High school students’
scientific epistemological beliefs, self-efficacy in learn-
ing physics and attitudes toward physics: A structural
equation model, Res. Sci. Technol. Educ. 33,252
(2015).
[78] J. A. Chen and F. Pajares, Implicit theories of ability of
Grade 6 science students: Relation to epistemological
beliefs and academic motivation and achievement in
science, Contemp. Educ. Psychol. 35, 75 (2010).
[79] D. Hammer, Epistemological beliefs in introductory
physics, Cognit. Instr. 12, 151 (1994).
[80] K. Hogan, Relating students’personal frameworks for
science learning to their cognition in collaborative con-
texts, Sci. Educ. 83, 1 (1999).
YAREN ULU and SEVDA YERDELEN-DAMAR PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-20
[81] M. H. Lee, R. E. Johanson, and C. C. Tsai, Exploring
Taiwanese high school students’conceptions of and
approaches to learning science through a structural
equation modeling analysis, Sci. Educ. 92, 191 (2008).
[82] S. A. Rosenberg, D. Hammer, and J. Phelan, Multiple
epistemological coherences in an eighth-grade discussion
of the rock cycle, J. Learn. Sci. 15, 261 (2006).
[83] J. Biggs, Individual differences in study processes and
the quality of learning outcomes, Higher Educ. 8, 381
(1979).
[84] C. Chin and D. E. Brown, Learning in science: A
comparison of deep and surface approaches, J. Res.
Sci. Teach. 37, 109 (2000).
[85] J. M. Case and R. F. Gunstone, Approaches to learning in
a second year chemical engineering course, Int. J. Sci.
Educ. 25, 801 (2003).
[86] E. Hazel, M. Prosser, and K. Trigwell, Variation in
learning orchestration in university biology courses,
Int. J. Sci. Educ. 24, 737 (2002).
[87] K. R. Muis and M. C. Duffy, Epistemic climate and
epistemic change: Instruction designed to change stu-
dents’beliefs and learning strategies and improve
achievement, J. Educ. Psychol. 105, 213 (2013).
[88] M. S. Khine, B. J. Fraser, and E. Afari, Structural relation-
ships between learning environments and students’non-
cognitive outcomes: Secondary analysis of PISA data,
Learn. Environ. Res. 23, 395 (2020).
[89] Y.H. Lee and H. Y. Hong, Preservice teachers’intention
for constructivist ICT integration: Implications from their
Internet epistemic beliefs and internet-based learning self-
efficacy, Interact. Learn. Environ. 32, 102 (2023).
[90] H. I. Strømsø and I. Bråten, Beliefs about knowledge and
knowing and multiple-text comprehension among upper
secondary students, Educ. Psychol. 29, 425 (2009).
[91] M. M. Williams and C. George-Jackson, Using and doing
science: Gender, self-efficacy, and science identity of
undergraduate students in STEM, J. Women Minorities
Sci. Eng. 20, 99 (2014).
[92] P. Vincent-Ruz and C. D. Schunn, The nature of science
identity and its role as the driver of student choices, Int. J.
STEM Educ. 5, 48 (2018).
[93] Z. Hazari, P. M. Sadler, and G. Sonnert, The science
identity of college students: Exploring the intersection of
gender, race, and ethnicity, J. Coll. Sci. Teach. 42,82
(2013).
[94] C. Monsalve, Z. Hazari, D. McPadden, G. Sonnert, and P.
Sadler, presented at PER Conf. 2016, Sacramento, CA,
https://www.compadre.org/Repository/document/
ServeFile.cfm?ID=14235&DocID=4589.
[95] V. Seyranian, A. Madva, N. Duong, N. Abramzon, Y.
Tibbetts, and J. M. Harackiewicz, The longitudinal effects
of stem identity and gender on flourishing and achieve-
ment in college physics, Int. J. STEM Educ. 5, 40 (2018).
[96] S. Cwik and C. Singh, Not feeling recognized as a physics
person by instructors and teaching assistants is correlated
with female students’lower grades, Phys. Rev. Phys.
Educ. Res. 18, 010138 (2022).
[97] S. Yerdelen-Damar and A. Eryılmaz, Questions about
physics: The case of a Turkish ‘Ask a Scientist’website,
Res. Sci. Educ. 40, 223 (2010).
[98] R. Trumper, Factors affecting junior high school students’
interest in physics, J. Sci. Educ. Technol. 15, 47 (2006).
[99] J. M. Nissen, Gender differences in self-efficacy states in
high school physics, Phys. Rev. Phys. Educ. Res. 15,
013102 (2019).
[100] S. Cwik and C. Singh, Damage caused by societal
stereotypes: Women have lower physics self-efficacy
controlling for grade even in courses in which they
outnumber men, Phys. Rev. Phys. Educ. Res. 17,
020138 (2021).
[101] M. S. Topçu and Ö. Yılmaz Tüzün, Elementary students’
metacognition and epistemological beliefs considering
science achievement, gender and socioeconomic status,
Elementary Educ. Online 8, 676, 2009.
[102] F. Kurt, Investigating students’epistemological beliefs
through gender, grade level, and fields of study, Master’s
thesis, Middle East Technical University, 2009.
[103] K. Ozkal, C. Tekkaya, S. Sungur, J. Cakiroglu, and E.
Cakiroglu, Elementary students’scientific epistemologi-
cal beliefs in relation to socioeconomic status and gender,
J. Sci. Teach. Educ. 21, 873 (2010).
[104] S. Ozkan and C. Tekkaya, How do epistemological beliefs
differ by gender and socioeconomic status, Hacettepe
Univ. J. Educ 41, 339 (2011).
[105] X. Guo, W. Deng, K. Hu, W. Lei, S. Xiang, and W. Hu,
The effect of metacognition on students’chemistry
identity: The chain mediating role of chemistry learning
burnout and chemistry learning flow, Chem. Educ. Res.
Pract. 23, 408 (2022).
[106] D. Hammer and A. Elby, On the form of a personal
epistemology, in Personal Epistemology: The Psychology
of Beliefs about Knowledge and Knowing, edited by B. K.
Hofer and P. R. Pintrich (Lawrence Erlbaum Associates
Publishers, Hillsdale, NJ, 2002), pp. 169–190.
[107] B. K. Hofer, Dimensionality and disciplinary differences
in personal epistemology. Contemp. Educ. Psychol. 25,
378 (2000).
[108] S. Chen, B. Wei, and H. Zhang, Exploring high school
students’disciplinary science identities and their
differences, Int. J. Sci. Math. Educ. 21, 377 (2022).
[109] Y. Ulu and S. Yerdelen-Damar, Fizik Benlik Ölçeğinin
Türkçeye Uyarlama Çalışması, V. Ulusal Fizik Eğitimi
Kongresi, 2022, Istanbul, Online (2022).
[110] A. Akın, R. Abacıve, and B. Çetin, Bilişötesi farkındalık
envanteri’nin türkçe formunun geçerlik ve güvenirlik
çalışması, Kuram Uygul. Egit. Bil. 7, 655 (2007).
[111] T. Kline, Psychological Testing: A Practical Approach to
Design and Evaluation (Sage Publications, Inc., Thou-
sand Oaks, CA, 2005).
[112] K. A Bollen, Total, direct, and indirect effects in structural
equation models, Sociol. Methodol. 17, 37 (1987).
[113] J. Pearl, Direct and indirect effects, in Proceedings of the
Seventeenth Conference on Uncertainty in Artificial
Intelligence (Morgan Kaufmann, San Francisco, CA,
2001), p. 411.
[114] K. A. Pituch and J. P. Stevens, Applied Multivariate
Statistics for the Social Sciences: Analyses with SAS
and IBM’s SPSS, 6th ed. (Routledge, New York, NY,
2016).
METACOGNITION AND EPISTEMIC COGNITION …PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-21
[115] J. Cohen and P. Cohen, Applied Multiple Regression/
Correlation Analysis for the Behavioral Sciences (Erl-
baum, Hillsdale, NJ, 1983).
[116] S. J. Finney and C. DiStefano, Nonnormal and categorical
data in structural equation modeling, Structural Equation
Modeling: A Second Course, 2nd ed., edited by G. R.
Hancock and R. O. Mueller (Information Age Publishing,
Charlotte, NC, 2013), pp. 439–492.
[117] L. K. Muth´en and B. Muth´en, Mplus User’s Guide
(Muth´en & Muth´en, Los Angeles, CA, 2017).
[118] L. T. Hu and P. M. Bentler, Cutoff criteria for fit indexes
in covariance structure analysis: Conventional criteria
versus new alternatives, Struct. Equ. Model. 6,1
(1999).
[119] H. W. Marsh, K. T. Hau, and Z. Wen, In search of golden
rules: Comment on hypothesis-testing approaches to
setting cutoff values for fit indexes and dangers in
overgeneralizing Hu and Bentler’s (1999) findings,
Struct. Equ. Model. 11, 320 (2004).
[120] K Schermelleh-Engel, H. Moosbrugger, and H. Müller,
Evaluating the fit of structural equation models: Tests of
significance and descriptive goodness-of-fit measures,
Methods Psychol. Res. Online 8, 23 (2003).
[121] J. Pallant, SPSS Survival Manual: A Step by Step Guide to
Data Analysis Using the SPSS Program, 4th ed. (Allen &
Unwin, Berkshire, 2011).
[122] S. R. Briggs and J. M. Cheek, The role of factor analysis
in the development and evaluation of personality scales, J.
Pers. 54, 106 (1986).
[123] C. K. Enders, Applied Missing Data Analysis (Guilford
Press, New York, 2010).
[124] J. C. Anderson and D. W. Gerbing, Structural equation
modeling in practice: A review and recommended two-
step approach, Psychol. Bull. 103, 411 (1988).
[125] W. A. Sandoval and K. Morrison, High school students’
ideas about theories and theory change after a biological
inquiry unit, J. Res. Sci. Teach. 40, 369 (2003).
[126] N. D. Finkelstein and S. J. Pollock, Replicating, and
understanding successful innovations: Implementing tu-
torials in introductory physics, Phys. Rev. ST Phys. Educ.
Res. 1, 010101 (2005).
[127] R. F. Moll and M. Milner-Bolotin, The effect of inter-
active lecture experiments on student academic achieve-
ment and attitudes towards physics, Can. J. Phys. 87, 917
(2009).
[128] V. L. Akerson and M. L. Volrich, Teaching nature of
science explicitly in a first-grade internship setting, J. Res.
Sci. Teach. 43, 377 (2006).
[129] W. A. Sandoval and B. J. Reiser, Explanation-driven
inquiry: Integrating conceptual and epistemic scaffolds
for scientific inquiry, Sci. Educ. 88, 345 (2004).
[130] R. S. Schwartz, N. G. Lederman, and B. A. Crawford,
Developing views of nature of science in an authentic
context: An explicit approach to bridging the gap between
nature of science and scientific inquiry, Sci. Educ. 88, 610
(2004).
[131] E. F. Redish and D. Hammer, Reinventing college physics
for biologists: Explicating an epistemological curriculum,
Am. J. Phys. 77, 629 (2009).
[132] D. Cheung, The combined effects of classroom teaching
and learning strategy use on students’chemistry self-
efficacy, Res. Sci. Educ. 45, 101 (2015).
[133] X. Huang, R. E. Mayer, and E. L. Usher, Better together:
Effects of four self-efficacy-building strategies on online
statistical learning, Contemp. Educ. Psychol. 63, 101924
(2020).
[134] F. Ogan-Bekiroglu and M. Aydeniz, Enhancing preser-
vice physics teachers’perceived self-efficacy of argu-
mentation-based pedagogy through modelling and
mastery experiences, Eurasia J. Math. Sci. Technol. Educ.
9, 233 (2013).
[135] V. Sawtelle, E. Brewe, and L. Kramer, Positive impacts of
Modeling Instruction on self-efficacy, AIP Conf. Proc.
1289, 289 (2010).
[136] M. Yough, Tapping the sources of self-efficacy: Promot-
ing preservice teachers’sense of efficacy for instructing
English language learners, Teach. Teach. Educ. 54, 206
(2019).
[137] J. C. Y. Sun and K.Y.C. Hsu, A smart eye-tracking
feedback scaffolding approach to improving students’
learning self-efficacy and performance in a C program-
ming course, Comput. Hum. Behav. 95, 66 (2019).
[138] N. Valencia-Vallejo, O. López-Vargas, and L. Sanabria-
Rodríguez, Effect of a metacognitive scaffolding on self-
efficacy, metacognition, and achievement in E-Learning
environments, Knowl. Manage. E-Learn. 11, 1 (2019).
[139] J. H. Zhang, B. Meng, L. C. Zou, Y. Zhu, and G. J.
Hwang, Progressive flowchart development scaffolding to
improve university students’computational thinking and
programming self-efficacy, Interact. Learn. Environ. 31,
3792 (2021).
[140] R. Henderson, V. Sawtelle, and J. M. Nissen, Gender &
self-efficacy: A call to physics educator, Phys. Teach. 58,
345 (2020).
[141] T. Espinosa, K. Miller, I. Araujo, and E. Mazur, Reducing
the gender gap in students’physics self-efficacy in a team-
and project-based introductory physics class, Phys. Rev.
Phys. Educ. Res. 15, 010132 (2019).
[142] Z. Hazari, D. Chari, G. Potvin, and E. Brewe, The context
dependence of physics identity: Examining the role of
performance/competence, recognition, interest, and sense
of belonging for lower and upper female physics under-
graduates, J. Res. Sci. Teach. 57, 1583 (2020).
[143] Z. Y. Kalender, E. Marshman, C. Schunn, T. N. Nokes-
Malach, and C. Singh, Why female science, technology,
engineering, and mathematics majors do not identify with
physics: They do not think others see them that way,Phys.
Rev. Phys. Educ. Res. 15, 020146 (2019).
[144] Y. Li and C. Singh, Statistically equivalent models with
different causal structures: An example from physics
identity, Phys. Rev. Phys. Educ. Res. 20, 010101 (2024).
[145] V. Adlakha and E. Kuo, Critical issues in statistical causal
inference for observational physics education research,
Phys. Rev. Phys. Educ. Res. 19, 020160 (2023).
[146] J. R. Fraenkel, N. E. Wallen, and H. H. Hyun, Internal
Validity. How to Design and Evaluate Research in
Education (McGraw-Hill, New York, 2012), pp. 166–
183.
YAREN ULU and SEVDA YERDELEN-DAMAR PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-22
[147] A. Traxler, X. Cid, J. Blue, and R. Barthelemy, Enriching
gender in physics education research: A binary past and a
complex future, Phys. Rev. Phys. Educ. Res. 12, 020114
(2016).
[148] B. M. Byrne, Structural Equation Modeling with Mplus:
Basic Concepts, Applications, and Programming (Rout-
ledge, New York, 2012).
[149] F. F. Chen, Sensitivity of goodness of fit indexes to lack of
measurement invariance, Struct. Equ. Model. 14, 464
(2007).
[150] G. W. Cheung and R. B. Rensvold, Evaluating goodness-
of-fit indexes for testing measurement invariance, Struct.
Equ. Model. 9, 233 (2002).
METACOGNITION AND EPISTEMIC COGNITION …PHYS. REV. PHYS. EDUC. RES. 20, 010130 (2024)
010130-23
