The study of science identity has become crucial in the field of science education. It significantly influences students’ involvement, determination, and achievements in science areas (Carlone and Johnson, 2007). Science identity is about how individuals see themselves as understanding, engaging in, and appreciating science (Stets et al., 2017). A strong sense of science identity is linked to higher motivation, resilience, and success in science (Shanahan, 2009; Hazari et al., 2010; Chen et al., 2021).

Several factors work together to shape a student’s science identity. These include the learning environment and the broader historical and social context. This is especially true for young people from diverse backgrounds (Fraser et al., 2014). For educators in science, reflecting on their teaching methods and dealing with challenges can lead to a deeper understanding of their professional identity (Mansfield et al., 2022). Using an intersectionality approach, the importance of acknowledgment and emotional responses is highlighted. This approach also points out issues related to power, inequality, and exclusion (Avraamidou, 2020). Engaging in real science activities can help build a science identity. It does so by making individuals feel they belong and are recognized, interested, and skilled in science (Huffmyer et al., 2022). Thus, developing a science identity involves both personal and societal aspects. These include relationships, the broader context, self-reflection, acknowledgment, emotions, and professional viewpoints.

Research has pinpointed key elements that help form a strong science identity. One significant element is self-efficacy, or the belief in one’s ability to accomplish certain tasks (Kim, 2018; Syed et al., 2019; Miles and Naumann, 2021). Studies have emphasized various aspects of science self-efficacy. These aspects include understanding integrated science concepts, applying science practically, and effectively communicating scientific ideas (Britner and Pajares, 2006; Wang and Tsai, 2019; Hu et al., 2022).

Success in science topics also plays a vital role in forming science identity (Hazari et al., 2017; Chen et al., 2021). When students do well in science, they more strongly identify with it. Their successes make them feel capable and competent in this field (Carlone and Johnson, 2007; Shanahan, 2009; Master and Meltzoff, 2020). Moreover, studies have examined how factors like gender and socioeconomic status might influence the link between self-efficacy, academic success, and science identity (Robnett et al., 2015; Miles and Naumann, 2021). While some research shows clear gender differences in science identity formation (Desy et al., 2011; Wang and Yu, 2023), other studies suggest these differences may depend on various social and cultural factors (Sachdev, 2018; Chan et al., 2019). Having a strong science identity can greatly affect students’ academic and career paths. Those with a well-developed science identity are more likely to choose science-related fields, overcome obstacles, and take part in science activities and groups (Jackson et al., 2016; Stets et al., 2017).

Recognizing the vital role of science identity, experts have suggested specific teaching methods and interventions. These strategies aim to nurture a positive science identity among all students (Vincent-Ruz and Schunn, 2018; Rushton and Reiss, 2021; Sandrone, 2022). Possible actions include showing students a variety of scientist roles, offering real science experiences, and creating supportive educational settings. Such settings help students feel competent and valued in the science community (Meyer and Crawford, 2015; Sheffield et al., 2021). Research on science identity underscores its importance in encouraging students to engage, achieve, and persist in science fields. By understanding how self-efficacy, academic performance, and other factors interact, educators can design strategies and interventions. These efforts aim to build a strong science identity among diverse student groups. Strengthening science identity is key to widening participation in STEM fields and preparing future scientists and innovators.

Self-efficacy beliefs, which refer to an individual’s confidence in their ability to perform specific tasks or activities, have been extensively studied in the context of science education (Bandura, 1997; Britner and Pajares, 2006). A growing body of research has specifically examined the role of science learning self-efficacy, which encompasses an individual’s beliefs about their capabilities to learn and perform various science-related tasks and activities such as virtual reality and lab acitivities (Usher and Pajares, 2009; Glynn et al., 2011; Gungor et al., 2022).

Science learning self-efficacy has been conceptualized as a multidimensional construct, comprising distinct yet interrelated dimensions (Wang and Tsai, 2019; Tan et al., 2021). These dimensions include experimental science proficiency, which reflects an individual’s confidence in conducting scientific experiments and interpreting data; practical science application, which involves the ability to apply scientific knowledge and skills to real-world situations; and science communication efficacy, which encompasses the belief in one’s ability to effectively communicate scientific concepts and ideas (Lin and Tsai, 2013; Sezgintürk and Sungur, 2020).

Research has consistently demonstrated a positive relationship between science learning self-efficacy and academic achievement in science subjects (Britner and Pajares, 2006; Usher and Pajares, 2009; Alhadabi, 2021). Students with higher levels of science learning self-efficacy tend to exhibit greater persistence, resilience, and engagement in science-related tasks, leading to improved academic performance (Glynn et al., 2011; Zheng et al., 2018).

Moreover, science learning self-efficacy has been identified as a critical factor in shaping students’ science identity, which refers to an individual’s recognition of themselves as someone who understands, participates in, and values science (Trujillo and Tanner, 2014; Syed et al., 2019). Students with higher levels of science learning self-efficacy are more likely to develop a stronger science identity, as their confidence in their abilities to learn and perform science-related tasks reinforces their self-perceptions as capable and competent in the field (Hazari et al., 2010).

Studies have looked at how demographic factors, like gender and economic status, might affect confidence in science learning self-efficacy (Tan et al., 2023). However, the influence of gender on science self-efficacy is less clear, with some studies finding no difference between male and female students, while others suggest that male students may have higher levels of self-efficacy and better science learning outcomes (Trisnawati et al., 2020; Nurhasnah et al., 2022). Research has shown that men and women may have different levels of confidence in their science abilities (Zeldin et al., 2008; Robinson et al., 2022). However, other studies point out that these differences often depend on social and cultural backgrounds (Reuben et al., 2014; Robinson et al., 2022). Overall, these findings highlight the importance of considering demographic factors when examining confidence in science learning self-efficacy, as they can play a role in shaping students’ beliefs and motivation in science education.

Understanding that self-efficacy in science learning is important, experts suggest specific ways to help students believe more in their science abilities (Usher and Pajares, 2009; Glynn et al., 2011). These methods include hands-on experience, observing others, positive encouragement, and learning to handle stress (Bandura, 1997; Usher and Pajares, 2009). The study of students’ confidence in their science learning abilities is key in science education. Researchers are looking into how this confidence relates to academic success, science identity, and other influencing factors. Boosting this confidence is important for getting students involved, helping them to keep going, and making them successful in science fields. This work is vital for creating a skilled and varied group of STEM professionals.

Educational research has been paying close attention to how gender, science identity, and confidence in science learning intersect. This review looks at how gender affects the development of science identity and learning confidence in science. It points out differences between genders, examines the reasons for these differences, and discusses ways to overcome them. Science identity is about whether a person sees themselves as someone linked to science. It is important for getting students interested in science and helping them stick with it (Carlone and Johnson, 2007). Studies show that boys are more likely to consider themselves connected to science from an early age (Alexander et al., 2012; Archer et al., 2014). This trend is due to cultural ideas and gender roles that suggest science is mainly for men (Robinson et al., 2019; Al-Balushi et al., 2022). Girls encounter stereotypes that imply science is not suitable for them. This can affect how they see themselves in relation to science (Reuben et al., 2014; Master and Meltzoff, 2020). However, positive support from teachers and family can help girls feel more connected to science (Hazari et al., 2017; Wang et al., 2020).

Believing in one’s ability to succeed in science is crucial for motivation and staying with science studies (Bandura, 1997). There’s a noticeable difference in this belief between genders. Girls often feel less capable in science, especially in areas like physics and engineering (Britner and Pajares, 2006). Studies suggest girls feel less capable in science because they get less encouragement and have fewer role models (Herrmann et al., 2016). But mentoring and hands-on learning can greatly boost girls’ confidence in their science abilities (Zeldin et al., 2008; Jackson et al., 2016; Stets et al., 2017).

Having a strong science identity can increase a student’s confidence in their science abilities (Gubbels and Vitiello, 2018). If students see themselves as “science people,” they are more likely to believe they can achieve in science. Similarly, if they believe in their science abilities, they will likely develop a stronger science identity. Research shows that science identity and confidence support each other, leading to more interest and persistence in science for both boys and girls (Alexander et al., 2012). A positive classroom environment and inclusive teaching methods can help build both science identity and confidence, especially for students who are less represented in science (Trujillo and Tanner, 2014).

A recent study on science education in Russia has brought attention to many obstacles and inequalities associated with gender. Gender inequalities continue to exist in Russian science education, as women are not adequately represented in research output and scientific influence across different fields of study. Throughout history, women in Russia have had limited representation in science (Kataeva et al., 2023). Their research output and scientific influence in these domains have consistently been lower compared to men (Paul-Hus et al., 2015; Loyalka et al., 2021). The discrepancy is apparent from the early stages of education, as males are more commonly represented among the top 20% of high-achieving students in science at both the fourth and eighth-grade levels. This pattern has been documented in other nations, including Russia (Meinck and Brese, 2019). Notwithstanding these difficulties, girls in Russia are more inclined to take academic paths in secondary education in comparison to boys. This tendency is impacted by family money and parental education. However, no further impacts of gender on this transition have been discovered (Bessudnov and Malik, 2016). Nevertheless, the collective proficiency of Russian students, encompassing both males and females, falls behind that of nations such as China, highlighting the necessity for enhanced pretertiary education to adequately equip students for science and technology programs at the university level (Loyalka et al., 2021). These findings indicate that although there have been improvements in certain aspects, there are still notable differences between genders in Russian science education. This highlights the need for specific initiatives aimed at promoting gender equality and improving the overall quality of science education.

To reduce the gender gap in science identity and confidence, there are interventions like mentoring programs, fair teaching methods, and introducing students to a variety of science role models. Studies have found that female science role models can make a big difference for female students. They help build both science identity and confidence (Conner and Danielson, 2016). Mentorship from female scientists has been shown to boost female undergraduates’ confidence and connection with science (Dennehy and Dasgupta, 2017). In conclusion, gender has a big influence on science identity and confidence in science learning. Although there are noticeable differences, targeted efforts and inclusive teaching can lessen these differences. They can make science education more equal. Future research should keep looking at how gender, identity, and confidence in science are related. It should focus on interventions that create a science learning environment that is welcoming and fair for everyone.

Understanding the link between doing well in science, feeling connected to science, and believing in one’s science abilities is very important in the study of science education. Knowing how success in science shapes the way people view themselves as science learners and doers is helpful. It can lead to more interest and lasting involvement in the science field. When people feel they are “science people” and are seen that way by others, we say they have a science identity. Doing well in science helps strengthen this identity. This includes getting good results, being acknowledged, and getting positive feedback. Important research by Avraamidou (2020) and Bryan et al. (2011) shows that success in science activities and recognition from others are crucial for developing a strong science identity.

Believing you can succeed in science is known as self-efficacy in science learning. Success in science boosts this belief by showing you can do well and are skilled in science. Studies by Britner and Pajares (2006) and Usher and Pajares (2009) highlight that achievements and overcoming challenging tasks are essential for building this self-confidence. The development of a science identity and the belief in your science abilities are interconnected. A strong science identity can drive you to attempt difficult science tasks, and succeeding in these can increase your confidence in your abilities. Trujillo and Tanner (2014) and Sandrone (2022) have found that educational strategies that improve science identity and self-efficacy lead to better academic performance in science.

Educational interventions aimed at enhancing success in science can have a positive effect on students’ science identity and self-efficacy. These can include inquiry-based learning, feedback, and peer collaboration, as suggested by Sheffield et al. (2021) and Eagan et al. (2023), who argue that connecting science education to students’ lives and self-regulated learning are beneficial. In conclusion, positive experiences and achievements in science education are vital. They encourage success and support teaching practices, recognition, and opportunities for mastering experiences. This strengthens students’ identification with science and belief in their science capabilities. Future research should continue to explore these relationships and the effectiveness of interventions designed to enhance all three components in diverse educational contexts.

In this study, the effect of science learning self-efficacy on science identity was investigated, along with whether gender and science success play a moderating role in this relationship. A quantitative approach was employed, and a theoretical model, as depicted in Figure 1, was developed for the analysis. The Partial Least Squares Structural Equation Modeling (PLS-SEM) approach was utilized to test the model. The choice of PLS-SEM was strategic, due to the complexity of the model which includes multiple sub-dimensions of science learning self-efficacy. Additionally, the preference for PLS-SEM was reinforced by its suitability for data that do not necessarily follow a normal distribution.

The questionnaire in this study was mainly distributed to high schools in Russia. Six high schools (2 schools in each city) were in Moscow, Almetyevsk, Khabarovsk cities. All the participants were asked to read and approve the ethical consent form before participating in this study. The schools were randomly selected in consultation with local administrators to ensure a diverse sample. Before participating, all students were required to read and approve an ethical consent form. While specific age data was not collected, Russian high school students typically range from 14 to 18 years old. Initially, 600 questionnaires were distributed. After removing incomplete responses, the final sample consisted of 519 participants. The gender distribution was nearly equal, with 263 females and 256 males.

In this research, the team collected data using two different questionnaires. We used the Science Identity Scale (Chen and Wei, 2022) and the Science Learning Self-Efficacy Scale (Lin and Tsai, 2013). Both scales (in appendix) were applied to the linguistic adaptation process. For linguistic adaptation, three experts were enlisted. The first expert was responsible for translating the scales from English into Russian. Subsequently, the second expert retranslated the Russian-translated scales back into English. The third expert then compared both translations along with the original scales, thereby validating the Russian version of the scales.

Chen and Wei (2022) confirmed that the Science Identity Scale is both valid and reliable. It was made for students in the later years of high school. The scale has 24 items. The Cronbach’s alpha for the scale was 0.95. This means the questions are very consistent within the scale.

The Science Learning Self-Efficacy Scale, as assessed by Lin and Tsai (2013), is designed for students in their later years of high school. It consists of 25 items spread across five sub-dimensions. The scale has a Cronbach’s alpha of 0.97, indicating it is a very reliable tool for measuring students’ belief in their science abilities. A test of the model including the previously identified sub-dimensions was conducted. However, the HTMT value turned out to be very high, suggesting that there might be overlapping items. Therefore, an exploratory factor analysis was carried out on a randomly selected group of 260 people from the main sample. To determine the number of factors, a parallel analysis was used. A Varimax rotation was applied for factor loadings (Williams et al., 2010). As a result of the factor analysis, items that overlapped were removed (items 1, 2, 4, 7, 13, 17, 23). Bartlett’s Test of Sphericity showed a chi-square value of 5,619, which is significant (p < 0.001), and the KMO was calculated to be 0.953. According to the results of the factor analysis, the scale items were divided into four different sub-dimensions. The first sub-dimension includes 8 items and has been named “Integrative Science Competence.” The second sub-dimension contains 6 items and is called “Practical Science Application.” The third has 5 items, named “Experimental Science Proficiency.” The fourth sub-dimension includes 2 items and is termed “Science Communication Efficacy.” Altogether, these dimensions account for 75% of the variance in the scale.

The gender of the participants was determined based on their self-reports. In the evaluation of science success, success in Physics, Chemistry and Biology courses was calculated based on grading between (1–5). The average success of 3 subjects was recalculated as Science success.

The researchers used a statistical method called Partial Least Squares Structural Equation Modeling (PLS-SEM) to analyze the data. This method is useful for testing theoretical models with multiple factors. First, the researchers checked that the survey questions used to measure each factor were reliable and valid. The factor loadings, Cronbach’s alpha, and average variance extracted values all met the recommended thresholds, indicating good reliability and validity. Factor loadings above 0.70 were deemed acceptable. Internal reliability was established based on composite reliability above 0.70. Average variance extracted (AVE) above 0.50 confirmed convergent validity. Discriminant validity was verified using the Fornell-Larcker criterion and Heterotrait-Monotrait (HTMT) ratio, ensuring constructs differed from others (Henseler et al., 2017; Hair et al., 2019; Wong, 2019). Next, the researchers examined the direct effects of the dimensions of science learning self-efficacy on science identity using path analysis. Bootstrapping with 5,000 resamples was used to calculate the t-statistics and p-values for assessing the significance of path estimates. Finally, moderating roles for gender and science success were tested. All PLS-SEM analyses were performed using the SmartPLS v 4.0 (Ringle et al., 2022).

The model in which all the items were included was tested. The loading values of the 3rd and 5th items of science identity were removed because they were lower than 0.70. Then the model was tested again. Starting with factor loadings, which measure the strength of the relationship between each item and its supposed underlying construct, we observe that all listed items demonstrate loadings above the 0.70, signifying robust correlations.

Table 1 presents factor loading and reliability coefficients for dimensions related to Science Learning Self-Efficacy Scale and Science Identity Scale. Generally, factor loadings above 0.7 are considered good, indicating that the items are likely to be a good measure of the underlying construct they are intended to represent. Also, it displays the Cronbach’s alpha, Composite reliability (rho a and rho c), and the Average Variance Extracted (AVE). Cronbach’s alpha values are used to assess the internal consistency of the items within each latent variable. The values provided are high (all above 0.9), suggesting excellent internal consistency. The Composite Reliabilities (both rho_and rho_c) are similarly high, which further supports the reliability of the constructs. The AVE values are all above 0.5, the commonly recommended threshold, indicating a satisfactory level of convergent validity. Table 1 indicates that the items used to measure each construct in the study are reliable and that the constructs are well-defined. The high-reliability coefficients suggest that the latent variables are likely to be stable and consistent across different samples of respondents.

Table 2 presents the Heterotrait-Monotrait (HTMT) ratio of correlations, which is an index used to assess the discriminant validity in constructs within a model. A value of HTMT less than 0.90 is often considered indicative of adequate discriminant validity. The values shown suggest that all constructs in the science learning self-efficacy pairs exhibit discriminant validity, as most HTMT values are below the 0.90 threshold. The value between science identity and Integrative Science Competence is higher than the determined threshold value 0.90. But since science identity and Integrative Science Competence are from different scales and Science identity is a dependent variable, it does not pose a problem.

Table 3 presents the Fornell-Larcker criteria matrix. The diagonal elements of the table (bolded) represent the square root of the Average Variance Extracted (AVE) for each construct. The AVE square root should be higher than the off-diagonal elements in its corresponding row and column, which represent the correlations between constructs. Based on the Fornell-Larcker criterion, for discriminant validity to be established, the square root of the AVE (diagonal elements) for each construct must be greater than the correlations between that construct and any other construct in the model (off-diagonal elements). It would indicate that each construct is distinct and captures phenomena that are not captured by the other constructs, fulfilling the Fornell-Larcker criterion for discriminant validity. This suggests that the constructs in the model are sufficiently different from each other, which is an important aspect of building a sound theoretical model (Figure 2).

Table 4 summarizes statistical findings on the relationship between various science educational factors and the construct of “identity.” Each factor’s influence is measured by the original sample’s path coefficient, its sample mean, and standard deviation, followed by a calculation of t statistics to determine significance. The results for Experimental Science Proficiency show a path coefficient of 0.091, suggesting a slight positive effect on identity, with a standard deviation of 0.037. The t statistic of 2.460 and a p value of 0.007 indicate that this relationship is statistically significant, though the effect size is relatively small.

Integrative Science Competence has a notably stronger association with identity, evidenced by a path coefficient of 0.603. The consistency of this effect is underscored by an identical sample mean and a low standard deviation. The very high t statistic of 14.294 and a p value of practically zero strongly suggest that this is a robust and highly significant finding. The Practical Science Application factor shows a moderate positive influence on identity, with a path coefficient of 0.185. The t statistic of 4.807 and a negligible p value confirm the statistical significance of this relationship. Lastly, Science Communication Efficacy also has a small yet significant positive effect on identity, as indicated by a path coefficient of 0.089 and a t statistic of 3.470. Again, the p value of 0.000 affirms the statistical significance of this effect (Table 5).

For Experimental Science Proficiency, with an f² value of 0.013, the effect size of Experimental Science Proficiency on “identity” is very small. Cohen’s f² values of 0.02, 0.15, and 0.35 are typically considered small, medium, and large effect sizes, respectively. Hence, Experimental Science Proficiency has a negligible influence on “identity.” For Integrative Science Competence, the f² value of 0.455 indicates a large effect size, suggesting that Integrative Science Competence has a substantial impact on “identity.” It is the most influential predictor among the ones listed. For Practical Science Application, the f² value of 0.059 suggests a small to medium effect size. This indicates that Practical Science Application has a moderate influence on “identity.” For Science Communication Efficacy, with an f² value of 0.021, the effect of Science Communication Efficacy on “identity” is small, indicating it has some but limited influence (Table 5).

R-square value of 0.798 [0.772–0.824] for “identity” suggests that approximately 79.8% of the variance in “identity” can be explained by the independent variables included in the model. This is a high value, indicating a strong model that provides a good fit to the data. Adjusted R-square: The adjusted R-square value of 0.796 [0.770–0.823] is very close to the R-square value, which implies that the number of predictors in the model is appropriate, and the model is not overfitted with unnecessary variables (Figure 3).

Table 6 presents original sample path coefficients, sample means, standard deviations, t statistics, and p values, all of which help assess the significance and strength of the relationships. Experimental Science Proficiency has a small positive effect on “identity” (path coefficient of 0.089) which is statistically significant (p = 0.042). Integrative Science Competence shows a strong positive effect on “identity” (path coefficient of 0.450), with high statistical significance (p < 0.001). Practical Science Application and Science Communication Efficacy both demonstrate moderate positive effects on “identity” (path coefficients of 0.154 and 0.153, respectively) and are statistically significant (p = 0.002 and p < 0.001, respectively). Science success also positively influences “identity” with a path coefficient of 0.148 and is statistically significant (p < 0.001).

In Interaction effects, the interaction of gender with Integrative Science Competence, Practical Science Application, and Science Communication Efficacy indicates a mix of small positive and negative influences on “identity.” However, none of these interactions are statistically significant, given that their p values exceed the conventional threshold of 0.05 for significance. Notably, the interaction between science success and Science Communication Efficacy is negative (path coefficient of −0.089) and is statistically significant (p < 0.001). This suggests that the combined effect of science success and science communication efficacy on identity is different than when these factors are considered alone.

From the data, we can observe the direct effects of science learning self-efficacy (Experimental Science Proficiency, Integrative Science Competence, Practical Science Application, and Science Communication Efficacy) on science identity. Integrative Science Competence stands out with the strongest positive direct effect on science identity, followed by Practical Science Application and Science Communication Efficacy, which also show significant positive relationships. Experimental Science Proficiency presents a significant relationship as well, but with a smaller effect size. Science success also positively contributes to science identity, highlighting its direct role in shaping a person’s identification with science.

When introducing gender as a potential moderating variable, the effects on science identity are nuanced. The interaction terms (gender with the different self-efficacy factors) do not show statistically significant effects on science identity, as their p values are not within the threshold for significance. This indicates that gender does not significantly alter the impact of science self-efficacy on science identity in the sample examined.

Similarly, science success as a moderator displays mostly non-significant interactions with the various components of learning science self-efficacy in relation to science identity. This suggests that the level of science success does not significantly change the way these self-efficacy components influence one’s science identity, with one notable exception. The interaction between science success and Science Communication Efficacy shows a negative and significant effect on science identity. This implies that although Science Communication Efficacy positively affects science identity, when combined with science success, the effect becomes negative. This could indicate a complex dynamic where students who are successful in science but do not feel efficacious in communicating science may experience a conflict that negatively impacts their science identity.

In summary, while learning science self-efficacy components have a strong standalone impact on science identity, the moderating roles of gender and science success are not straightforward. Gender does not appear to influence the relationship significantly. In contrast, science success does seem to moderate the relationship between Science Communication Efficacy and science identity, albeit in a negative way, warranting further investigation to understand this interaction fully.

Figure 4 displays three lines, each representing the relationship between Science Communication Efficacy and science identity at different levels of science success: 1 standard deviation below the mean (red line), at the mean, and 1 standard deviation above the mean (green line).

The slope of the line for participants with science success at −1 SD (below average) is notably steeper and negative, suggesting that for these individuals, as Science Communication Efficacy increases, their science identity significantly decreases. This could indicate that lower levels of science success might amplify negative feelings or perceptions about science identity as the ability to communicate about science increases. For participants at the mean level of science success, the slope is flatter and appears to be more neutral or slightly positive, indicating that at average levels of science success, increases in Science Communication Efficacy have a less pronounced or slightly positive effect on science identity. Lastly, the line for participants with science success at +1 SD (above average) shows a positive relationship. This suggests that individuals with higher levels of science success experience an increase in their science identity as their Science Communication Efficacy improves.

In essence, this simple slopes analysis indicates that the effect of Science Communication Efficacy on science identity is contingent upon the level of science success. The moderating effect of science success is such that individuals with lower success in science may feel a decrease in science identity with better communication efficacy, whereas those with higher success may feel their identity bolstered by the same. This could reflect a potential discrepancy between perceived self-efficacy in communication and actual success in science activities, which could influence how one’s science identity is formed or maintained.