Examining the Structural Relationships Between Pre-Service Science Teachers ’ Intention to Teach and Perceptions of the Nature of Science and Attitudes

: The current study aimed to examine the structural relationships between pre-service science teachers ’ intention to teach and perceptions of the nature of science (NOS) and attitudes toward teaching science. The sample consisted of 206 pre-service science teachers from a major university in the United Arab Emirates who have bachelor ’ s degrees in science. Quantitative research methodology was used to study the structural relationships among pre-service science teachers ’ intention to teach, perceptions of the NOS, and attitudes. Structural equation modeling using the Lavaan package was used to test these structural relationships. The results show that the relationships between the indigenous constructs (NOS) and the mediator constructs (DAS) can predict pre-service science teachers ’ intention to teach science ( p < 0.05).


Introduction
Due to the visible results of detrimental science teaching practices, many researchers in teacher education have acknowledged the importance of examining teachers' behavioral intention to teach.Teaching science is considered a hard profession, and effective teaching is a challenge for teachers and researchers in the science education field.Many reasons and factors contribute to the effectiveness of science teaching at the school level.It requires a complicated matrix of teaching skills to implement all of these skills and tools in the classroom (Abd-El-Khalick, 2013).
Van Aalderen-Smeets et al. (2017) pointed to the necessity of improving science teachers' attitudes toward teaching as a crucial factor that predicts the teacher's intention to teach and use these skills in the teaching process, and at the same time, recent research studies have highlighted the importance of teachers understanding of the nature of science (NOS) on their ability and intention to teach science concepts and laws (American Association for the Advancement of Science (AAAS), 1990(AAAS), , 1993)).This enhances meaningful science learning (Rutherford, 2009).Teachers' attitudes and understanding of the NOS help inform students that science evolves but is not constantly upheaval, and scientists accept that knowledge is changeable and renewed (Koksal & Koseoglu, 2015;Lederman, 1992;National Research Council, 2000).
Therefore, students need to understand how science is organized, as scientific enterprise plays a crucial role in students' ability to influence public support for science in their societies (Liu & Lederman, 2007;NRC, 2000).Likewise, Cobo, Abril, and Romero-Ariza (2022) pointed out that teachers who have inappropriate conceptions of the NOS have a low quality of teaching science.However, Abd-El-Khalick (2013) and McComas and Olson (2000) stated that science teachers' adequate perceptions of NOS help transfer accurate scientific knowledge and may have the intention to teach.
Attitudes are described as a degree of preference for something or object (Skott, 2015).Baumert and Kunter (2013) asserted that teachers may exhibit specific attitudes toward teaching and learning in the school context.However, there is a lack of conceptual clarity when it comes to precisely defining attitudes meaning, as they are rarely differentiated from beliefs or subjective theories (Fives & Buehl, 2012).Considering that attitude is a complex, multidimensional, and non-unified concept.Many studies did not distinguish attitudes from opinion or motivation, while other studies did not capture the overall attitude concept (Ajzen, 2001).This problem makes the task of interpreting the results of previous related studies and linking them with the results of recent studies somewhat difficult due to the ambiguity of the concept of attitude (Pardo & Calvo, 2002).
To overcome such problems in science education, Van Aalderen-Smeets and Van der Molen (2013) and Van Aalderen-Smeets, Walma van der Molen, and Asma (2012) work developed an instrument to measure teachers' attitudes toward teaching science called the Dimensions of Attitude toward Science (DAS) instrument.Therefore, this instrument was used in the current study.Also, it is necessary to clarify the difference between attitudes toward science that describe individuals' beliefs, feelings, and values about science (Tytler & Osborne, 2012) and the scientific attitudes as cognitive attributes characterizing the scientific method of thinking like search for evidence (Osborne, Simon, & Collins, 2003).
Pre-service teachers' beliefs and attitudes are important for understanding and improving the teaching-learning process, and they are also closely linked to teachers' strategies for dealing with challenges in their daily professional lives (OECD, 2009).That is why both theory and research concur that effective teachers extend beyond merely possessing pedagogical content knowledge.In other words, the attitudes of teachers play a crucial role in their professional competence (Baumert & Kunter, 2013).
As previously mentioned, the original DAS Instrument is based on a theoretical framework that describes the teachers' professional attitudes and consists of three major dimensions "Cognitive Beliefs, Affective State, and Perceived Control" (Van Aalderen-Smeets et al., 2012).In this article, the instrument is modified with some editing in its subcomponents to suit our study.Cognitive beliefs about teaching science comprise relevance, perceived difficulty, and gender-stereotypical beliefs, "teachers' perceptions about differences between these groups."However, the affective state includes enjoyment and anxiety in teaching science (Van Aalderen-Smeets & Van der Molen, 2013).Perceived control is determined by two subcomponents: self-efficacy and context dependency.Self-efficacy could be described as a teacher's beliefs about his or her ability to act, depending on internal factors such as knowledge, confidence, and skills (Bandura, 1997;Van Aalderen-Smeets et al., 2012).On the other hand, context dependency could express pre-service beliefs and feelings about the influence of external factors on their teaching, such as time limits and students' abilities (Van Aalderen-Smeets & Van der Molen, 2013).
In the same context, some science teachers may be reluctant to teach science due to perceptions of their selfefficacy and ability to control some contextual factors, reducing their confidence in their ability to teach (King, Shumow, & Lietz, 2001).Reviewing previous research indicates that teacher education programs, mainly science programs, have the potential to construct appropriate beliefs, increase self-confidence, and enhance positive attitudes toward teaching science, leading to more effective instructional practices (Morrell & Carroll, 2003;Tosun, 2000).Furthermore, negative impressions of pre-service science teachers' abilities could affect their interest in teaching science, affecting their beliefs about teaching science and teaching practices in classrooms (Ginns, Tulip, Watters, and Lucas, 1995;Kazempour, 2014;Putman, 2012).
Relatedly, in science education, beliefs have often been linked with teaching practices (Luft & Roehrig, 2007;Savasci-Acikalin, 2009), while Fives and Buehl (2012) and Priestley, Biesta, and Robinson (2015) emphasize that the relationship between beliefs and teaching practices is not straightforward.Several studies showed that quality science education programs lead to preparing highly trained teachers to teach science, as it affects their attitudes toward teaching science (Velthuis, Fisser, & Pieters, 2014;Worch, Li, & Herman, 2012).
In the same context, field training is one of the factors influencing the formation of pre-service science teachers' attitudes because it includes many activities such as observations, microteaching, self-reflection, and feedback which may cause a struggle for them regarding its twofold effect.Field training leads to enhancing pre-service science teachers' self-confidence because of gaining teaching experience on the one hand.On the other hand, it may cause anxiety because of a feeling of teaching complexity, their ability to manage many tasks simultaneously, and the extent of external factors' effects (Carrier, 2009;Hoy, Davis, & Pape, 2006;Putman, 2012).Accordingly, Skott (2009) indicated that context and constraints must be considered to enhance pre-service teachers' intention to teach.
Given that, it is inferred that pre-service teachers' attitudes are influenced by receiving constructive feedback and appropriate teaching experiences during their field training (Carrier, 2009;Soprano & Yang, 2013).Furthermore, effective mentoring strategies and experiences solidify teachers' intention to teach (Campbell & Kovar, 1994).One of the basic characteristics of teachers that affect science teaching competencies is a sense of self-efficacy, which reflects their attitudes toward teaching science effectively (Zaruba, Westphal, Gutmann, & Vock, 2021).At the beginning of teachers' careers, their attitudes can potentially guide classroom settings, the quality of instruction (Voss & Kunter, 2020), students' achievement (Wilcox-Herzog, Ward, Wong, & McLaren, 2015), and increasing students' learning (Putman, 2012).Avraamidou (2017) explained that many new teachers struggle at the beginning as a result of a lack of scientific knowledge (Chalufour, 2010), a lack of NOS in science teaching (Abd-El-Khalick & Lederman, 2000), and feeling anxiety and low self-efficacy in teaching science (Gerde, Pierce, Lee, & Van Egeren, 2018;Hamlin & Wisneski, 2012).This leads us to determine the main factor controlling preservice teachers' intention to teach.
In the same context, the literature review supports that successful intention to teach pre-service science teachers requires factors beyond the NOS (Abd-El-Khalick & Lederman, 2000), and Shahali, Halim, Treagust, Won, and Chandrasegaran (2017) highlighted the importance of other factors called knowledge of specific contexts, such as science content, beliefs, and self-efficacy, as a clear road map that guides the future instructional decisions a science teacher makes about teaching.
Several research studies were conducted to investigate the relationships between teachers' perceptions of the NOS, intention to teach science, and attitudes toward Science (Abd-El-Khalick & Lederman, 2000;Corrigan, 2007;Schofield et al., 2023;Schoon, & Boone, 1998;Takriti et al., 2023), but these studies did not investigate whether the relationships are direct or indirect.

Research Questions
The main goal of this study is to explore the structural relationship among Pre-service Science Teachers' Intention to Teach, perceptions of the NOS, and Attitudes toward Science.This goal is achieved by answering the following two research questions: 1. Are the relationships between the Indigenous Variable (Intention to teach) and Mediator variable components (Dimensions of Attitudes toward Science) statistically significant?2. Are the relationships between the Mediator variable components (Dimensions of Attitudes toward Science) and Exogenous variable components (Teachers' perceptions of the NOS) statistically significant?

Data Collection: Instruments and Procedures
The study used three instruments to collect data to investigate the structural relationships between Teachers' perceptions of the NOS and Attitudes toward Science (DAS) and Pre-service Science Teachers' Intention to Teach.The first instrument was the Myths of Science Questionnaire (MOSQ).The MOSQ was used to measure the pre-service science teachers' conceptions of NOS as suggested by Doungpaen and Buaraphan (2012) since its psychometrics properties were established and found within the accepted levels.MOSQ is composed of ( 14) items that were distributed to four constructs.A panel of experts in science education was asked to review the instrument, and they suggested distributing the items into three constructs, namely, scientific worldview, scientific inquiry, and the scientific enterprise, after revising all the items and aligning with the project 2061 documents (American Association for the Advancement of Science (AAAS), 1990, 1993; Koksal & Koseoglu, 2015).Cronbach Alpha, Item to Total correlation range, and the number of items of each construct of MOSQ are shown in Table 1 as evidence of scale reliability.
The second instrument was the Dimensions of Attitudes toward Science (DAS) developed by Van Aalderen-Smeets and Van der Molen (2013) and was reused in the study by Mills, Whiteford, Brown, and Tomas (2023) and Wendt and Rockinson-Szapkiw (2018).Also, DAS was reviewed by a panel of experts, and a revised version of DAS was used in the current study.The revised DAS consists of 21 items distributed to five constructs: Perceived relevance, Perceived difficulty, Affective state Self-efficacy, and Context dependency.Table 1 shows Cronbach Alpha, Item to Total correlation range, and the number of items of each construct of DAS as evidence of scale reliability.
The third instrument used in this study is the Behavioral intention to teach science (Mills et al., 2023;van Aalderink & Walma van der Molen, 2015).The construct "gender beliefs" was excluded because the sample consists of female pre-service teachers.The "enjoyment" and "anxiety" constructs were merged into one construct named " Affective state" (Larsen & Diener, 1992).Table 1 shows Cronbach Alpha, Item to Total correlation range, and the number of items of intention to teach construct as evidence of scale reliability.
Structural equation modeling (SEM) was used in this study.This analysis consists of two parts in the same analysis.The first one called the measurement model adds evidence of the construct validity of all components of the tools used to collect the data in this study as shown in the results section of this study.The second one is called the structural model which is used to answer the current study research questions.

Sampling
The current study sample consisted of 206 Pre-service science teachers from Al Ain University -United Arab Emirates.All the participants have bachelor's degrees in science and are enrolled in their first semester in the Postgraduate Professional Diploma in Teaching Program.About 92% of the sample is from the Sultanate of Oman as AL Ain University located at the borders between the Sultanate of Oman and the United Arab Emirates.The Ministry of Education in the Sultanate of Oman requires having a Professional Diploma in Teaching for all teachers before hiring them.Only 8% of the participants in this study are UAE citizens or UAE residents; 18 participants were excluded from the analysis as they reported that they were teachers (In-service Teachers).Sample distribution according to majors in bachelor degrees is shown in Table 2.
The sample voluntary response to the instrument was done using Google Forms.An email was sent to all students at the Postgraduate Professional Diploma in Teaching Program who have bachelor's degrees in science and the professors who teach them were asked to encourage the students to participate in this study.The total number of participants was 206 Preservice science teachers.

Data Analysis
SEM was used to measure and analyze the relationships between all the variables shown in Figure 1.One of the most common packages used to run this analysis is the lavaan package (Latent Variable Analysis).One of the main attractions of lavaan is its intuitive and easy-to-use model syntax, fairly complete, and contains most of the features that applied researchers are looking for in a modern SEM package (Rosseel, 2012).The sem function under the lavaan package consists of two parts.The first is the measurement model, which is a form of confirmatory factor analysis (CFA), which has the items' loadings on their constructs and their significance.The measurement model tests the relationships between the items (Observed variables) and the measured construct (Latent variable), which provide evidence for the psychometric properties of the surveys.The second part is the structural model.This model represents the relationships between the constructs or the latent variables and their significance.Four global fit indices were used to assess the goodness of fit of the suggested model.These indices and their cutoff scores are presented in Table 3.

Results
lavaan package (Latent Variable Analysis), which works under the environment, was used to test the model using SEM analysis with latent variables.The global fit indices of testing the model are summarized in Table 3 There are different points of view among scholars and authors about interpreting and assessing the cutoff criteria.Therefore, it is recommended to use several fit indices at the same time (Devlieger et al., 2019).Model complexity and Sample size should be taken into account when evaluating the model fit indices (Hu & Bentler, 1999;Schumacker & Lomax, 2016).Hu and Bentler (1999) recommended using RSMEA of 0.06 or lower and SRMR of 0.09 or lower as a two-index criterion to evaluate model fit.Based on that, the fit indices shown in Table 3 indicate a poor fit of the model.This means that the model does not fit the collected data.Accordingly, the "modification indices" option that is available on the lavaan package was used to find the modifications that could be added to the suggested model to enhance the global fit indices.The suggestions were making a few items from the same construct correlate.The suggested modifications were analyzed quantitatively and qualitatively and added to the model accordingly.
The suggested modifications were added to the suggested model, and the analysis was conducted another time.The results of the global indices are presented in Table 4.
Based on the cutoff criteria presented in the table and the recommendation of Hu and Bentler (1999) to use RSMEA of 0.06 or lower and SRMR of 0.09 or lower as a two-index criterion to evaluate model fit, we can say that the suggested model has a good.Since the modified model fits the data, the path coefficients and relationships between the constructs can be evaluated.
The coefficients of this model are presented in two separate tables for simplifying purposes only.Table 5 shows the measurement model, which is a very important part of the model where the psychometrics of the used data collection tools can be evaluated.The coefficients in the measurement model represent loadings of the items on  the measured domain (subscales).The results presented in Table 5 added evidence of the construct validity of the tools used to collect the data as all the loadings of the items on its factors are statistically significant.2).Table 5 shows the results of the measurement models.Each part of the table is a confirmatory factor analysis.All the loadings in the table are statistically significant.This provides evidence of the construct validity of the instruments used to collect the data in this study.
Table 6 shows the SEM standardized regression coefficients (PE.est), their standard errors (PE.se), and their pvalues (PE.pvalue) to test the significance of the path coefficients between the constructs (null hypothesis: coefficient equals zero or no relationship between the constructs).These path coefficients are partial regression coefficients, and it is interpreted as the volume of change in the indigenous construct given a unit change in the exogenous construct when controlling for the effects of other exogenous constructs in the model.The path coefficients show the strength and the direction of the relationships between the Intention to Teach construct (indigenous variable or dependent variables) and the five DAS (Moderator variables: Perceived relevance, Perceived difficulty, Affective state Self-efficacy; and Context dependency) and the three constructs of Teachers perceptions of the NOS (Exogenous variables: Scientific world view, Science inquiry, and Scientific enterprise).
The suggested and final models have three types of constructs.The type is the indigenous construct (Intention to teach), Mediator construct (Dimensions of Attitudes toward Science), and Exogenous construct (Teachers' perceptions of the nature of science).The results shown in Table 6 are presented in two parts.The first represents the relationship between the Indigenous and Mediator constructs, while the second represents the relationship between the Mediator and Exogenous constructs.The relationships between the indigenous constructs and the mediator construct, as shown in Table 6, indicate that pre-service science teachers' intention to teach science has statistically significant path coefficients (p < 0.05) with four constructs of the Dimensions of Attitudes toward Science: Perceived relevance, Perceived difficulty, Affective State, and Self-efficacy.As the items that measure Affective state dimensions are negatively worded, the path coefficients are negative, indicating a negative relationship between these dimensions and the intention to teach science in the future.On the other hand, the Context dependency construct does not have a statistically significant path coefficient (p < 0.05) with pre-service science teachers' intention to teach science.
The relationships between the Mediator constructs and Exogenous constructs, as shown in Table 6, indicate that the three constructs of the nature of science have statistically significant path coefficients (p < 0.05) with Perceived relevance and Self-Efficacy DAS.The affective state dimension has a statistically significant path coefficient with only the Scientific enterprise construct from the nature of science constructs, while perceived difficulty has a statistically significant path coefficient with Science inquiry.Also, context dependency has a statistically significant path coefficient with Science inquiry.

Summary and Discussion
The main goal of this study is to explore the structural relationship among Pre-service Science Teachers' Intention to Teach, perceptions of the nature of science, and Attitudes toward Science.The main noticeable result of this study is that the NOS mediates the relationship between Pre-service Science Teachers' Attitudes and Intention to Teach.This means that developing positive attitudes toward teaching can be established by developing students' perceptions of the nature of science, which increases Pre-service Science Teachers' intention to teach science in the future.The previous studies showed that understanding the nature of science increases the teachers' intention to teach (Abd-El-Khalick & Lederman, 2000;Corrigan, 2007), and attitude toward teaching science affects the teachers' intention to teach (Schoon & Boone, 1998) so as a total this result covers the gap that previous studies missing, and it finds that there is a relation between the three factors, and according to this result, pre-services science teacher understanding of the nature of science effect on their attitude toward teaching science so on their intention to teach increase.
Another noticeable result is the statistically significant paths between the Self-efficacy and Perceived relevance constructs and all NOS constructs.This means that developing teachers' self-efficacy and Perceived relevance can be done by developing teachers' perceptions of the scientific worldview, scientific inquiry, and scientific enterprise.This result could direct teachers' preparation programs to focus on developing pre-service science teachers' perceptions of the NOS.
Van Aalderen-Smeets and Van der Molen (2013) concluded that at least two of the three attitude dimensions (i.e., affective states and perceived controlspecifically, the selfefficacy, context-dependency, enjoyment, and anxiety constructs) show predictive value for intended science-teaching behavior.The results of the current study showed that the relationships between the indigenous constructs and the mediator construct indicate that pre-service science teachers' intention to teach science has statistically significant path coefficients (p > 0.05) with four constructs of the Dimensions of Attitudes toward Science: Perceived relevance, Perceived difficulty, Affective State, and Self-efficacy.This means that they can contribute to forming the intention of pre-service teachers to teach science in the future.As evidenced by previous research, teachers with a high level of self-efficacy in teaching science will have a high attitude toward science (Kazempour, 2014;Senler, 2016;Shahali & Halim, 2023).Also, the relevance and enjoyment of teaching science strongly predicted their intention to teach.This indicates the significance of establishing the relevance and importance of science education (Mills et al., 2023).
The affective state dimension has a statistically significant path coefficient with only the Scientific enterprise construct from the nature of science constructs, which means that the affective state can be predicted by the Scientific enterprise.This result needs further research to stand on the relation between Scientific enterprise and the affective state.
The perceived difficulty has a statistically significant path coefficient with Science inquiry, which means that the perceived difficulty can be predicted by Science inquiry.Maybe because science inquiry involves the formulation of a question that can be answered through investigation and is one of the basics of scientific practices, also, science inquiry shifted from inquiry for conceptual understanding and development of process skills to inquiry for understanding the nature of science (Lee, Lee, Lam, Kwok, & So, 2018;NGSS Lead States, 2013), so the pre-service science teachers may have difficulty with it.
• Doing qualitative research to deeply understand students' perceptions of the nature of science for pre-service teachers.• More research might be needed to know how the different undergraduate program courses cover the nature of science and present it to the students.• The results of this study suggest covering the nature of science to pre-service teachers to raise their level of scientific knowledge related to understanding the nature of science.• Investigating the effect of Professional Diploma in Teaching programs on attitudes toward teaching and intention to teach using pre\post experimental research design.

Table 1 :
Data collection tools and their psychometric properties

Table 2 :
Sample distribution according to majors in bachelor's degrees

Table 3 :
Global fit indices of the suggested model Table 6, the second part of the model, has path coefficients (regression coefficients) between the constructs in the model and its p-values.Table 6 answers the two research questions; the top part of the table shows the relationships between the Indigenous Variable (Intention to teach) and Mediator variable components (Dimensions of Attitudes toward Science) while the rest of the table shows relationships between the Mediator variable components (Dimensions of Attitudes toward Science) and Exogenous variable components (Teachers' perceptions of the NOS; Figure

Table 4 :
Global fit indices of the modified model

Table 5 :
The loading of each item on its construct

Table 6 :
The path regression coefficients between the constructs for the final model The final model.Relationships Between Pre-Service Science Teachers' Intention and Perceptions  7