Introduction

Science, technology, engineering, and mathematics (STEM) play an important role in the various aspects of one’s life, from individual health to environmental issues (Nagle et al., 2009). Thus, STEM literacies are essential for every citizen in preparation for life in a modern society (National Research Council [NRC], 2015; NGSS, 2013; Sawyer, 2008). That being said, studies show a declining interest in science and other STEM areas throughout the school years, as well as negative perceptions regarding the social and personal relevance of science (Jarvis & Pell, 2002; Pellegrino & Hilton, 2012; Sturman & Rudduck, 2009). These findings create a challenge for science educators. Reversing this trend requires addressing the students’ voices, diverse students’ needs, and appropriate instructional methods. Meeting the needs of diverse students remains a source of frustration, misunderstanding, and distrust by some teachers, parents, and even students themselves (Fetters et al., 2003; Jarvis et al., 2017; Valiandes & Neophytou, 2018; Woodcock et al., 2019).

Student diversity in the science classroom poses a further challenge to science educators. Student diversity includes many variables such as: ethnicity, socioeconomic status, culture, abilities, and personal characteristics (Sanger, 2020; Schroeder-Davis, 2009). Personal characteristics determine the effectiveness of instruction and learning (Ahlfeldt et al., 2005; Tomlinson, 2015) and, therefore, the depth of learning. Although many studies and initiatives call for implementing appropriate instructional strategies to address the needs of all students in the science classroom, the students’ voices are hardly heard. Thus, the primary aim of this study was to narrow this gap by bringing forth the voices of middle school students regarding their science learning preferences and examining the dynamic interactions between the students’ science learning preferences and other personal characteristics: gender, having or not having a learning disability (i.e., with/without LD), and science achievements (i.e., their scientific knowledge and skills scores) in the context of science instruction and learning. To the best of our knowledge and based on the Literature Review we conducted, studies examining the dynamic interactions between these variables in the context of learning preferences in middle school science studies are scarce.

Students’ learning preferences

Students’ learning preferences are based on what they perceive as most effective for them, in other words, their preferred learning mode or style. When instruction accommodates learning style preferences, students are more likely to approach a learning situation, leading to better performance, achievements, and learning outcomes (Alp et al., 2018; Burden, 2010; Cassidy, 2004; Heacox, 2012; Hussain & Ayub, 2012; Stanton, 1974). Allowing students to learn according to their learning preferences makes them more likely to participate in learning, and feel respected and valued (Benny & Blonder, 2016; Stanton, 1974; Tomlinson & Jarvis, 2014). The field of students’ learning preferences gained much attention with the prevalence of the term learning styles. Learning styles theories are widely known in the field of educational research and have been studied since the 1970s. The term learning styles has been used in education to explain students’ different approaches to learning (Kozhevnikov, 2007). Over 60 different models of learning styles have been proposed (Coffield et al., 2004), focusing on various student characteristics, such as cognition, personality, information processing, or instructional preferences (e.g., Bruner, 1965; Dewey, 1986; Fleming, 1995; Gardner & Hatch, 1989; Piaget, 1976; Slater et al., 2007).

Nonetheless, there is ambivalence regarding the contribution of a learning styles-based pedagogy to learning gains. Setbacks in this area include a lack of solid explanatory theory; poor reliability and validity of constructs; and a failure to link learning styles-based instruction to achievement (An & Carr, 2017; Pashler et al., 2008). Apparently, while it is easy to categorize an individual as a specific type of learner, people do not learn in the same way in all situations. Indeed, most people are likely multimodal and multi-situational learners, switching learning strategies based on the context of the material to be learned (Mutua, 2015). Solid theoretical frameworks in cognitive and developmental psychology offer alternative explanations for student characteristics or styles in terms of differences in sensory-based representations, levels of expertise, self-regulation, perfectionism, and temperament (An & Carr, 2017). Despite the criticism, and although learning styles theories may have lost some of their luster, introducing students to their learning preferences is an important tool for teachers. Studying how students prefer to learn can provide valuable information for establishing a discourse about learners’ strengths and weaknesses (Adey et al., 2012), enhancing their participation, and empowering them to engage in their learning (Franquesa-Soler et al., 2019). Explicit discussions with students about their learning preferences can bring their voices into the classroom, thus giving them ownership of their learning processes (Dhiman, 2014; Spektor-Levy et al., 2019) and promising better student outcomes (An & Carr, 2017). This discourse can provide educators with insights into possible interventions to promote cognitive, developmental, fluency, self-regulation capabilities, effortful control, and other learning strategies (Dhiman, 2014; Spektor-Levy et al., 2019). Thus, the intention of offering all students equal opportunities for success should be reflected in an ongoing, metacognitive dialogue between the teachers and the students about learning outcomes, achievements, and learning processes. Without explicit discussions on the subject, providing learning choices may result in confusion, ending up with students who do not achieve their potential. In this study, we asked students about their science learning preferences, using a specially developed questionnaire (see the Research Tools section) based on Fleming’s learning styles theory (Fleming, 1995) and the Multi-Faceted Holistic Approach (MuFHA; Yifrach, 2008; Yifrach et al., 2015) for teaching in heterogeneous classrooms. Fleming’s theory and the Multi-Faceted Holistic Approach provide us with the theoretical foundation and the tools to explore students’ voices and their learning preferences with the understanding that multi-modal instruction has the potential to cater to a variety of learners.

Fleming’s learning styles theory

Fleming’s learning styles theory focuses on information about the sensory path, which supports individuals in learning effectively. This path includes visual, aural/audio, read/write, and kinesthetic (VARK; Fleming, 1995) modalities. Each of these learning modalities (i.e., learning modes) is expressed as follows: Visual—learning through graphic information including figures, maps, charts, etc.; Aural/Audio—learning through lectures, group discussions, self-talk, etc.; Read/Write—learners focus on text-based data such as reports, lecture notes, journals, etc.; and Kinesthetic—hands-on experiential learning, outdoor learning, demonstrating, simulating, etc. Fleming’s theory and the varied learning modalities, served as guidelines in developing the Learning Preferences Questionnaire for this study.

The multi-faceted holistic approach

The Multi-Faceted Holistic Approach (MuFHA) makes learning accessible and welcoming to all students. The MuFHA was developed after the researchers conducted an extensive literature review and mapped pedagogical methods recognized as contributing to learning (Spektor-Levy et al., 2019; Yifrach, 2008). The approach addresses a diverse range of students, whether they are typical learners, struggling, students with LD, gifted, talented, etc. The MuFHA situates the whole learner (McCombs, 2017; Rimm-Kaufman & Jodl, 2020) at the center of the instruction and learning while addressing four dimensions of a learner’s needs: cognitive, social, emotional, and sensorimotor. The cognitive dimension includes, for example, teaching methods based on constructivism, explicit instruction of knowledge and skills, self-regulation, reflective thinking, and formative assessment. The social dimension can include teaching strategies such as collaborative learning, peer teaching, assessment, and culturally responsive teaching. Approaches addressing the emotional dimension include, amongst others, positive corrective feedback, relevancy, humor, and self-advocacy. Finally, the sensorimotor dimension includes experiential learning, multimodality, gamification, maker pedagogy, and more. Utilizing auditory means, visual means, sensorimotor means, emotional strategies, and agency support throughout different learning situations allows each student to engage in learning at a pace and time and in a context that suits them. MuFHA-based instruction was found to advance students’ sense of self-efficacy and their achievements in science studies (Spektor-Levy et al., 2021; Yifrach et al., 2011). The Multi-Faceted Holistic Approach and its varied learning modalities served as guidelines in developing the Learning Preferences Questionnaire for this study.

The focus of this study was to investigate how students’ science learning preferences (regarding four learning modalities: visual, auditory, sensorimotor, and agency support) are related to their characteristics: gender, whether or not they have learning disabilities (i.e., with/without LD), and their science achievements (i.e., scientific knowledge and skills scores). We decided to focus on these specific student characteristics for the following reasons: First, these characteristics typify a heterogeneous, mixed-gender classroom. Second, although these student characteristics (i.e., gender, with/without LD, and science achievements) have been studied in terms of learning science (Husband, 2012; OECD, 2016; Scruggs et al., 2012; Siegle et al., 2014), few studies have investigated the dynamic interactions between those variables in the context of learning preferences in science studies. The findings of such an investigation may shed light on students’ preferred learning modalities and suitable instructional methods to support the diversity of students and promote achievements in science studies.

Achievements and student diversity in science education

In terms of promoting achievements, scholarly literature is replete with studies exploring the best instructional strategies to support distinct groups of students with a common profile, such as girls (Ceci & Williams, 2011; Chen et al., 2019); under-achieving students (Thomas & Bogner, 2012; Zohar & Dori, 2003); students with learning disabilities (Asgar et al., 2017; Scruggs et al., 2013); gifted and talented students (Rasmussen & Lingard, 2018; Siegle et al., 2014). However, few studies have explored the learning modalities preferred by students in terms of improving science achievements among these groups of students. All students’ social, emotional, sensorimotor, and cognitive needs should be addressed to improve their achievements (Anderson et al., 2022; Ferreira, 2021; Mavilidi et al., 2021).

Gender differences and science learning preferences

Gender differences in science studies are widely documented in the scholarly literature (e.g., Hodgetts, 2010; Hong et al., 2013; Hoogerheide et al., 2016; Husband, 2012; Jeong & Davidson-Shivers, 2006; Liu & Wang, 2019; OECD, 2016). Gender differences are manifested as girls being less interested in science (Fortus & Vedder-Weiss, 2014; Miller et al., 2006) and the underrepresentation of girls and women in STEM fields (Burke & Mattis, 2007; Ceci & Willliams, 2011; Ceci et al., 2009; Cheryan et al., 2017; Hill, et al., 2010; Stoet & Geary, 2018), despite the superior performance of female students over male students in high school science studies and at other levels of education (Clark et al., 2008; Parker et al., 2018; Rahafar et al., 2017). Knowing that male and female graduate students have similar cognitive abilities (Jeong & Davidson-Shivers, 2006; Liu & Wang, 2019; Reuben et al., 2014) may indicate the need to adapt teaching and learning methods to the different needs of boys and girls. Studies indicate that boys and girls have different learning preferences and use different approaches to learning (Addabbo et al., 2016; Bayrak, 2012; Fredricks et al., 2018). For example, although both boys and girls prefer student-centered learning (Fredricks et al., 2018), which includes visual, sensorimotor, social, and emotional dimensions, this preference is expressed differently. Effective learning strategies for boys include project-based education that facilitates hands-on, kinesthetic learning; use of graphics, pictures, and storyboards (Gurian & Stevens, 2010); competitive learning opportunities; game-based learning; offering choices to study topics that appeal to them (Gurian & Stevens, 2010; Hawley & Reichert, 2010; King & Gurian, 2006; Stevens, 2006); and participation in critical discourse (Asterhan et al., 2012). Recommended teaching and learning methods (e.g., agency support, Fredricks et al., 2018) that may be particularly appropriate for girls include the provision of a guided and structured learning approach beforehand (Salminen et al., 2012); collaboration and teamwork (Asterhan et al., 2012; Meece et al., 2006; Miller et al., 2006); inclusion of real-world issues such as science news reports (Tsai et al., 2013); and learning in the context of everyday life (Lin et al., 2012).

Thus, to create a gender-friendly classroom, a variety of instructional methods should be implemented that address the needs of all genders: To capitalize on their strengths based on their gendered needs; to foster the development of positive study habits and learning strategies; and to maximize the instructional outcomes (Cevallos, 2017). Nevertheless, few studies have sought to explicitly ask students about their learning preferences in science studies.

Students with learning disabilities in science classrooms

Students with special educational needs (SEN) fall into three categories (Ruijs et al., 2010): a) cognitive problems, which include developmental delay, language and arithmetic difficulties, dyslexia and dyscalculia; b) behavioral problems, which include autism, problems with making an effort at school and externalizing problem behavior; and c) other problems, including physical disabilities, internalizing problem behavior, and gifted children. In this study, the term Learning Disabilities (LD) relates to cognitive problems. LDs stem from genetic and/or neurobiological factors or injuries that alter how the brain functions (Giofrè et al., 2017), resulting in difficulties in cognitive processing, organizational skills, social perception, social interaction, and perspective-taking (Asghar et al., 2017; Giofrè et al., 2017; Kavale & Forness, 1996; Sridhar & Vaughn, 2001). Learning difficulties are physiological manifestations of LD that persist throughout an individual’s lifetime. Individuals with LD may have average intelligence or above (Fiedorowicz et al., 2015). However, given the intricate interconnections within these individual capabilities, and the spectral nature of learning disabilities, individuals with LD cannot be perceived through a binary lense. Classroom activities, grades, and assessments can lead to significant pressure to achieve and conform, triggering signs of difficulty that signal one’s shortcomings both to one’s self and to others. In such an environment, challenges in the learning process can alter young individuals’ perceptions of themselves and their capabilities (Crosnoe et al., 2007). As in academics, performance is typically measured through skills exhibited by reading ability, mathematical abilities, written expression, and spelling (Burns, 2010; Hintze et al., 2006; Vaughan-Jensen et al., 2011). Students with LD perform academically at levels that are significantly below what would be expected, given their intellectual or cognitive potential. Students with LD who struggle to understand social interactions may also experience difficulties engaging in collaborative learning through inquiry processes (Brigham et al., 2011; Scruggs et al., 2012). The disparity between the intellectual potential exhibited by individuals with LD and their seeming inability to translate this intellectual capability into a consistent achievement level or expected performance is frustrating for the students, as well as for parents and teachers (Callinan et al., 2013). Unfortunately, this discrepancy is often misinterpreted by people in their surroundings as laziness or lack of motivation (Cortiella & Horowitz, 2014; Jacobs, 1978; Short et al., 1992), leading these students to feel inadequate when compared with their peers (Denhart, 2008; Renick & Harter, 1989). This can potentially cause a decline in self-esteem and even depression, anxiety, and dropping out of school (Korhonen et al., 2014; Morrison & Cosden, 1997; Van Ameringen et al., 2003). Appropriate support for students with cognitive LD should suit each individual’s learning disability subtype and include specific skill instruction, accommodations, compensatory strategies, and self-advocacy skills (Giofre et al., 2017). Providing such support requires that science teachers change their traditional teaching practices (Kolonich et al., 2018; Atanga et al., 2020; Spektor-Levy, et al., 2021; Spektor-Levy et al., 2019;) and verify the learning preferences of their students with LD.

Therefore, this current study seeks to investigate approaches to reversing the declining interest in STEM subjects and students’ negative perception regarding the social and personal relevance of science (Jarvis & Pell, 2002; Pellegrino & Hilton, 2012; Sturman & Rudduck, 2009). To this end, the study brings forth the voices of middle school students regarding their science learning preferences through four modalities (based on Flemings’ theory and the MuFHA): visual, auditory, sensorimotor, and agency support to investigate how their personal characteristics—such as gender and having LD—affect those preferences. Another aim of this study was to explore whether there is a correlation between the students’ science achievements and their science learning preferences.

Research questions and hypotheses

In light of the literature review underscoring the need to support diverse students in the science classroom to foster their interest in science, coupled with the lack of studies exploring the voices of the students regarding their science learning preferences in middle school science studies, this study sought to explore the following questions:

  1. 1.

    Are there significant differences in the level of science learning preferences through each of the four learning modalities (visual, auditory, sensorimotor, and agency support) among middle school science students in a heterogeneous classroom?

  2. 2.

    Are there significant differences between boys and girls and between students with/without LD in the level of their science learning preferences through each of the four learning modalities (i.e., visual, auditory, sensorimotor, and agency support)?

  3. 3.

    Is there a significant correlation between the students’ science achievements and their science learning preferences?

  4. 4.

    Do students’ personal characteristics (i.e., gender, with/without LD, and science achievements) contribute to their science learning preferences?

  5. 5.

    Do students’ science learning preferences (i.e., learning modalities—visual, auditory, sensorimotor, and agency support) contribute to their science achievements beyond their gender and whether they have or do not have LD?

We formulated several hypotheses based on the Literature Review.

Regarding differences in the level of science learning preferences through each of the four learning modalities (i.e., visual, auditory, sensorimotor, and agency support) among middle school science students, we proposed Hypothesis 1: Different learning preferences will be expressed by different students. This hypothesis is based on scholarly literature indicating various factors influencing preferences in learning, such as the content studied, the learning location, and the level of achievement (Schulze & Bosman, 2018). Other studies indicated recommended learning methods for specific groups of learners, such as kinesthetic learning for boys (Gurian & Stevens, 2010), guided and structured learning for girls (Salminen et al., 2012), inquiry-based learning for students with LD (Scruggs et al., 2013), and so on. Therefore, as part of the current study, we expected specific groups of learners to prefer different learning modalities. Regarding differences between boys and girls and between students with/without LD in terms of the level of their science learning preferences, we proposed Hypothesis 2: The level of science learning preferences will differ by gender and whether students have or do not have LD (with/without). Previous studies found that girls prefer agency support (Fredricks et al., 2018) and guided and structured learning (Salminen et al., 2012); collaboration and teamwork (Asterhan et al., 2012; Meece et al., 2006; Miller et al., 2006); and learning in the context of real-world, everyday life issues (Lin et al., 2012; Tsai et al., 2013). Therefore, we assumed that of the four learning preferences modalities (i.e., visual, auditory, sensorimotor, and agency support) examined in this study, girls’ most preferred learning modality would be agency support, including intrapersonal and interpersonal support. As for boys, we surmised that sensorimotor learning would be preferred over other learning modalities, as suggested in the literature (Gurian & Stevens, 2010). In addition, we expected that science learning preferences levels would differ depending on whether students have or do not have LD. Difficulties in cognitive processing, organizational skills, social perception, etc., among students with LD may affect their learning preferences (e.g., An & Carr, 2017; Asghar et al., 2017). Thus, we hypothesized that most learning modalities would be challenging for students with LD. This would be expressed by a higher preference for agency support, which can scaffold the learning process.

Regarding the possible correlation between the students’ science achievements and their science learning preferences, we proposed Hypothesis 3: A significant correlation will be found between the students’ science achievements (i.e., their scientific knowledge and skills scores) and their level of science learning preferences. This aligns with previous studies showing that difficulties in cognitive processing, organizational skills, social perception, etc., among low-achieving students affect their preferences (e.g., Asghar et al., 2017), mainly favoring agency support.

Regarding whether students’ personal characteristics (e.g., gender, with/without LD, and science achievements) contribute to their science learning preferences: In Hypothesis 4, we proposed that based on Hypotheses 2 and 3, the students’ characteristics (gender, with/without LD, and science achievements) would contribute to the students’ science learning preferences (Jackman & Morrain-Webb, 2019; OECD, 2016).

Regarding whether students’ learning preferences (i.e., visual, auditory, sensorimotor, and agency support) would contribute to their science achievements beyond their gender and whether they have or do not have LD: In Hypothesis 5, we expected that students’ learning preferences (i.e., visual, auditory, sensorimotor, and agency support) would contribute to their science achievements beyond their gender and whether they have or do not have LD. As studies have shown that the most common form of instruction is lecturing (e.g., Kaya & Kaya, 2020; Schwerdt & Wuppermann, 2011), students might be accustomed to the auditory learning modality. Thus, we hypothesized that the students’ preferred learning modality would predict their science achievements beyond their gender and whether they have or do not have LD.

Methodology

Participants

Our sample comprised 305 students in the ninth grade (14–15 years old) from six urban schools of a middle socioeconomic status. There were 139 boys (45.6%) and 166 girls (54.4%). Of these, 48 students (27 boys and 21 girls) agreed to reveal (with parental permission) the fact that they were diagnosed with learning disabilities. All participants were of the same cultural heritage and spoke the same language. All the participants studied in heterogeneous mainstream classes, and instruction was mainly frontal. The study focused on ninth-grade middle school students who are about to decide how to continue their studies in high school. Some of them may not study science anymore. Therefore, it is precisely at this age that it is of great importance to address science learning—for students who are in their final stages of studying science in an educational/academic setting as well as for students who may want to pursue the subject and may go on to a professional path in the field.

Research tools

The study applied a quantitative methodological approach employing two questionnaires.

Scientific knowledge and skills questionnaire

The students’ science achievements (i.e., their scores on the Scientific Knowledge and Skills Test) were measured by a newly developed questionnaire, the Scientific Knowledge and Skills Questionnaire, based on items from the annual, nationally administered formal science assessment for the ninth grade, which represented typical scientific knowledge and skills expected to be acquired by ninth-grade middle school students.

The questionnaire comprised 51 items: 48 multiple-choice questions and three open-ended questions. Questions tested the participants’ scientific knowledge as per the national science curriculum in the following subjects: (a) nutrition and the digestive system; (b) the scientific inquiry process and skills; and c) linear graph reading skills. The items, examples of which are presented below, sought to estimate scientific competencies and knowledge based on scientific question categories in the OECD’s Programme for International Student Assessment (PISA; OECD, 2023):

  1. 1.

    Explain phenomena scientifically: What sequence of events led to a change in the blood glucose concentration?

  2. 2.

    Evaluate and design scientific inquiry: How was the dependent variable measured?

  3. 3.

    Interpret data and evidence scientifically: Why does the blood glucose concentration decrease after 1.5 h?

  4. 4.

    Content knowledge: What is the route that food takes in the body?

  5. 5.

    Procedural knowledge: What is the control in the experiment?

  6. 6.

    Epistemic knowledge: (a) Choose the data from one nutrient in the table that supports Ian’s claim. (b) Explain your choice based on the role of nutrients in the body.

The questionnaire was examined for content validity by four expert panelists, including two science education faculty researchers, one of whom is also a middle school science teacher, and two other middle school science teachers. Each question received between 0–3 points, and the maximum sum of the scores for all questions was 100. The open-ended responses were analyzed based on a scoring scheme, and another team of three science education researchers validated the coding and scoring. The interrater reliability (i.e., the intra-class correlation [ICC]) among the three science education researchers regarding the students’ performance on the three open-ended responses was high (0.90–0.98). These ICC values indicate that the three science education researchers highly agreed with the coding and scoring. Reliability measures of the entire questionnaire revealed a Cronbach’s alpha coefficient of 0.84. This high value of Cronbach’s alpha indicates a high level of internal consistency between the questionnaire items.

The learning preferences questionnaire

Learning preferences in this study were measured by the newly developed Learning Preferences Questionnaire based on Fleming’s (1995) model of visual, aural, reading/writing, and kinesthetic (VARK) modalities for presenting and processing information and on Yifrach, et al. (2015) Multi-Faceted Holistic Approach (MuFHA) for science instruction in heterogeneous, inclusive classrooms.

The questionnaire consisted of 29 items. The questionnaire aimed to measure the level of students’ science learning preferences regarding each of the four learning modalities: visual, auditory, sensorimotor, and agency support. The visual measure concerned the use of visual means in learning (e.g., graphic information, including figures, maps, charts, and so on). For example, “I understand scientific ideas better when I see pictures or movies”; “It’s easier for me to understand scientific information when it’s presented in illustrations, tables, and graphs”. The auditory measure concerned learning through listening (e.g., lectures, group discussions, self-talk, and so on). For example: “Scientific knowledge makes more sense when I listen to a lecture”; “I understand what I write better if I read it aloud to myself”. The sensorimotor measure concerned kinesthetics and embodied pedagogy (e.g., learning that joins body and mind in a physical and mental act of knowledge construction, like building models and learning through sensing; Nguyen & Larson, 2015). For example: “I manage to understand and learn scientific information better if I’m in motion (at my desk or moving around the classroom)”; “I understand scientific information better when I can feel and touch things, such as the phenomena/instruments/materials”. The agency supports measure concerned personal scaffolding that could facilitate learning. For example: “When a task needs to be performed, I understand the guidelines best when they are written in more detail”; “It’s easier for me when the task is divided into parts—like conducting an experiment or doing classwork step by step”; “When the teacher describes the agenda of the lesson at the beginning, it helps me understand what we’re learning”.

A series of tests were conducted to examine the reliability, content validity, and the construct validity of the questionnaire (see Supplementary Materials 1). Each item was measured on a 4-point Likert scale ranging from 1 (disagree) to 4 (strongly agree).

The internal consistency of the Cronbach’s alpha for the four factors (learning modalities) of the Learning Preferences Questionnaire were as follows: visual 0.78; auditory 0.72; sensorimotor 0.79; and agency support 0.80. The internal consistency of the Cronbach’s alpha for the overall learning preferences scale was 0.90. The correlation coefficients between the four factors (learning modalities) in the Learning Preferences Questionnaire were significant (r = 0.521–0.826).

Procedure and ethics

Ethical approval (reference number 9367) to conduct the study was granted by the senior monitoring and control coordinator from the Ministry of Education’s office of Chief Scientist. The questionnaires were administered during the science lessons in the ninth-grade classrooms of 22 science teachers from six participating schools. The questionnaires were administered online in the classroom at the beginning of a scheduled lesson. A questionnaire link was sent in a text message, and students were asked to respond by completing the online questionnaires anonymously. Students only participated in the study following parental consent but were offered a chance to refuse to participate.

Data analysis

Before analyzing the study questions and hypotheses, we conducted Shapiro–Wilk tests to determine whether the dependent variables were normally distributed. The dependent variables were the students’ science learning preferences levels and science achievements (i.e., scientific knowledge and skills scores). The results indicated that the distribution of the dependent variables deviated significantly from the normal distribution (p < 0.05). Therefore, we examined the research questions and hypotheses by conducting both parametric and non-parametric tests. The non-parametric analyses were Mann–Whitney, Friedman, and Wilcoxon tests. The Mann–Whitney test served as the non-parametric test instead of the one-way multivariate analyses of variance (MANOVAs) to examine differences in the dependent variables between students with and without LD and the differences between boys and girls. The Friedman test served as the non-parametric test instead of the one-way repeated measures analyses of variance (ANOVAs) to examine differences between the learning preferences for the whole sample, for each gender, and of students with/without LD. The Wilcoxon test served as the non-parametric test to examine the differences between pairs of learning preferences (post-hoc analyses instead of Bonferroni post-hoc analyses). The findings of the non-parametric analyses indicated the same significant level of differences as the parametric analyses. Therefore, only the findings of the parametric analyses are presented, with the means and standard deviations of the dependent variables, instead of reporting the mean or the sum ranks. We also conducted a Mauchly’s test to examine the sphericity assumption. The result of the Mauchly’s test indicated that the assumption of sphericity was rejected. As a result, we reported the adjusted degree of freedom (df) in decimal numbers.

Results

Students’ learning preferences in science studies

To examine the first research question regarding the differences in the level of science learning preferences per modality (i.e., visual means, auditory means, sensorimotor means, and agency support) among middle school science students in a heterogeneous, inclusive classroom, we conducted one-way repeated measures ANOVA. The independent variables were the four learning modalities (i.e., visual, auditory, sensorimotor, and agency support). The dependent variable was the learning preferences level (i.e., the degree of preferring each learning modality). As shown in Table 1, all measures of the science learning preferences levels scored above the mid-point (on a scale of 1–4). That is, the participants indicated that they preferred all four learning modalities. However, significant differences were found between the levels of learning preferences, F(2.784, 846.364) = 17.82, p < 0.001, ηp2 = 0.06 (see Table 1). A Bonferroni analysis indicated that the mean preference level for learning science through visual and sensorimotor modalities was significantly higher than the mean preference levels for learning science through the auditory modality and for agency support (p < 0.001).

Table 1 Mean and SD of Students’ Learning Preferences Levels and Correlations Between Students’ Learning Preferences Levels in each of the Four Learning Modalities
Full size table

Pearson correlation analyses were conducted to examine the correlations between the levels of learning preferences in each of the four learning modalities. As Table 1 shows, significant positive correlations were found among all levels of the learning preferences. These results revealed that students who exhibited a higher level of learning preferences for using visual means also tended to indicate a preference for learning science using other means (e.g., sensorimotor).

In order to strengthen these results and to determine the optimal number of groups (K), the elbow technique (“elbow plot” or “elbow curve”; Hartigan & Wong, 1979) was used. The criterion states that the optimal number of clusters is the one in which adding another cluster does not add significant information (Baños et al., 2022). For learning science, two preference groups were the ideal number. The first group consisted of 135 students who exhibited lower preference levels in all four learning modalities (visual: M = 2.34, auditory: M = 2.20, sensorimotor: M = 2.17; and agency support: M = 2.23). The second group consisted of 170 students who exhibited higher preference levels in all four learning modalities (visual: M = 3.17; auditory: M = 2.84; sensorimotor: M = 3.22; and agency support: M = 2.96; SD = 3.22). These cluster patterns strengthened the results of the Pearson correlation analyses, which indicated that students who exhibited a higher level of learning preferences for learning science using visual means tended to indicate a preference for using other modalities as well.

Differences in science learning preferences levels through the four learning modalities

To examine the second research question regarding the differences in science learning preference levels according to gender and with/without LD, two one-way MANOVAs were conducted. The independent variables were the students’ personal characteristics (gender, with/without LD). The dependent variable was the learning preferences level (i.e., the degree of preference of each learning modality).

Differences in science learning preferences levels by gender

We hypothesized that there would be differences in learning preferences between girls and boys and that girls would show more preference for agency support while boys would prefer sensorimotor learning. Our hypothesis was partially confirmed.

The independent variable in the MANOVA was the students’ gender. The dependent variables were the science learning preference levels. The results of the MANOVA analysis indicated that the students’ science learning preferences levels differed significantly by gender, F(4,300) = 5.40, p < 0.001, ηp2 = 0.07. Examining the differences in the level of learning preferences between boys and girls for each learning modality indicated that the students’ preferences for learning science through visual means, F(1,303) = 8.40, p = 0.004, ηp2 = 0.03, and their preferences for agency support, F(1,303) = 7.70, p = 0.006, ηp2 = 0.03, differed significantly by gender. The results revealed that girls exhibited higher levels of learning preferences for visuals and agency support than boys. The effect sizes of the differences, measured by partial eta squared (ηp2), were low-moderate. No significant differences were found between boys and girls in the level of learning preferences for auditory and sensorimotor modalities. Table 2 presents the mean and SD of science learning preference levels by gender.

Table 2 Mean and SD of science learning preferences levels by gender
Full size table

Differences in science learning preferences levels by with/without ld

We hypothesized that students with LD would express a higher preference for agency support. Due to the vastly different sizes of the groups of students with and without LD (n = 48, n = 257, respectively), before conducting the MANOVA analyses, Levene’s tests (Kelter, 2021) were conducted to examine whether the variances of the science learning preferences levels differed significantly between the groups. The results of the Levene’s tests indicated no significant differences between the two groups in variances (p > 0.05).

The independent variable in the MANOVA was whether the students had or did not have a learning disability (with/without LD). The dependent variables were the science learning preferences levels. The results of the MANOVA analysis indicated that the science learning preferences levels did not differ significantly by with/without LD, F(4,300) = 0.16, p = 0.960, ηp2 = 0.00 (see Table 3). Thus, our hypothesis was disproved.

Table 3 Mean and SD of students’ science learning preferences levels by with/without LD
Full size table

Correlation between science achievements and science learning preferences levels

Pearson correlation analyses were conducted to examine the correlation between the science achievements (i.e., scientific knowledge and skills scores) and the students’ science learning preferences levels. We hypothesized that students’ science achievements would correlate significantly with their level of learning preferences. We also hypothesized that a correlation would be found between students with low science achievements and learning preferences for agency support. Our hypothesis was partially confirmed. With the exception of preferences for learning science through sensorimotor means, significant positive correlations were found between the students’ science achievements (i.e., scientific knowledge and skills scores) and their science learning preferences levels for visual, auditory, and agency support: visual: r(303) = 0.13, p = 0.020; auditory: r(303) = 0.18, p < 0.001; agency support: r(303) = 0.14, p = 0.017, respectively. These results indicated that the more the students preferred to learn science through visual, auditory, and agency support means, the higher their scientific knowledge and skills score. It should be noted that the correlations were significant but with low coefficients.

Contribution of gender, with/without LD, and science achievements to science learning preferences levels

To examine the fourth research question, regarding the contribution of the students’ personal characteristics to the explained variance of their science learning preferences levels, four multiple regression analyses were conducted to investigate the contribution of gender, with/without LD, and the science achievements to the science learning preferences levels: one analysis for each of the learning modalities. We conducted the regression analyses in addition to the MANOVA and ANOVA to examine the combined contribution of the three student characteristics to the students’ science learning preferences levels (see Table 4).

Table 4 Multiple regression analyses for science learning preferences levels (dependent variable) by gender, with/without LD, and science achievements
Full size table

As Table 4 shows, the contribution of the student characteristics measure was significant on three of the four science learning modalities (visual, auditory, and agency support). The contribution of the student characteristics to most of the preferred learning modalities was significant. Although we conducted four multiple regression analyses, which may have increased the risk for Type I error, a Bonferroni correction for this multiple comparison was also conducted (α = 0.05/4 = 0.0125). Additionally, it should be noted that there was a modest (all p < 0.05) contribution of the significance level of the students’ characteristics to the science learning preferences level measures (4%, 3.7%, and 4.1% for visual, auditory, and agency support, respectively). Specifically, the students’ gender significantly contributed to the science learning preferences level through visual means and with agency support. The positive β coefficients indicated that girls exhibited higher preference levels for visual learning and agency support than boys. In addition, science achievements (i.e., scientific knowledge and skills scores) significantly contributed to the science learning preferences levels through auditory means and for agency support. The positive β coefficients indicated that students with greater scientific knowledge exhibited higher preference levels for learning science through auditory means and with agency support. With/without LD did not contribute significantly to any of the four regression analyses.

Contribution of gender, with/without LD, and the science learning preferences levels to science achievements

In order to examine the fifth research question, regarding the unique contribution of the students’ science learning preferences levels beyond their characteristics (i.e., gender and with/without LD) to their science achievements, a hierarchical regression analysis was conducted. Hierarchical regression can evaluate the contributions of predictors above and beyond previously entered predictors, as a means of statistical control, and for examining incremental validity (Hunsley & Meyer, 2003). The hierarchical regression analysis conducted in this study examined the contribution of gender, with/without LD, and the science learning preferences level to the explained variance of their science achievements. In the first step of the regression model, the gender and with/without LD variables were entered in a stepwise manner. Only variables that contributed significantly to the explained variance were entered in this manner. The variables were entered by their level of significance. In the second step of the regression model, the science learning preferences levels were also entered in a stepwise manner. Note that these variables were only entered into the regression model in the second step to examine the unique contribution of the science learning preferences levels beyond students’ gender and whether they have or do not have LD (see Table 5).

Table 5 Hierarchical regression analyses for science achievements (dependent variables), by gender, with/without LD, and the science learning preferences levels
Full size table

As Table 5 shows, gender and with/without LD contributed significantly (6.4%) to the students’ science achievements. Gender contributed 4.8% with positive β coefficients, while with/without LD contributed 1.6%, beyond the contribution of gender, with negative β coefficients. Finally, following our hypothesis, preference levels for learning science through auditory means contributed significantly (3.1%) to the students’ science achievements beyond other student characteristics. The positive β coefficients indicated that as the preference level for learning science through auditory means increased, the science achievements (i.e., the scientific knowledge and skills scores) increased, respectively.

Discussion

As part of the literature review, we described the challenges faced by science educators to reverse the trend of students’ declining interest in science and other STEM areas throughout the school years. Reversing this trend requires addressing the students’ voices, addressing the diverse students’ needs, fostering interest in science, and developing appropriate instructional methods. Moreover, students’ learning preferences must be understood in order to counter this tendency and enhance science achievements. There is a need to initiate a discourse on learners’ strengths and weaknesses and to facilitate dialogue between teachers and students regarding this matter. Therefore, the main aim of this study was to explore what students’ science learning preferences are and what the dynamic interactions are between their science learning preferences and three of their personal characteristics (i.e., gender, having or not having a learning disability, and level of scientific knowledge and skills). Below, we elaborate on our findings and the implications derived from these findings.

students’ learning preferences in science studies

The first research question sought to explore the learning preferences of middle school science students in heterogeneous, inclusive classrooms. To address this question, we developed the Learning Preferences Questionnaire, which was validated and tested for reliability. Findings showed that students’ learning preferences levels of all four learning modalities (visual, sensorimotor, auditory, and agency support) scored above the mid-point, meaning that the majority of students preferred each of the four learning modalities to a certain extent. A correlation test between the four learning preferences levels showed that all four learning modalities presented to the students were correlated. This result suggests that students in inclusive heterogeneous science classrooms expect to encounter a variety of teaching and learning methods throughout their school years, as was underscored in other studies (Roy et al., 2013; Tomlinson, 2014). However, further analyses showed a significantly higher preference for learning science through visual and sensorimotor means compared to auditory means and agency support. These findings are in line with other evidence indicating that the implementation of multiple representations, including visual and verbal representations, is linked to better learning in science, mathematics, and reading (DeStefano & LeFevre, 2004; Jornet & Roth, 2015; Mastropieri & Scruggs, 1997).

Nevertheless, this does not necessarily imply that students prefer multiple representations simultaneously. Rather, various representations should be implemented in different learning situations as part of a learning unit. The highly visual nature of human cognition (Coleman & Willis, 2015; Shanahan et al., 2016; Storbeck et al., 2006) may explain the students’ preference for visual means in learning.

Further support for a preference for visual means may be related to Mayer’s Cognitive Theory of Multimedia Learning, according to which a combination of text and pictures promotes comprehension and problem-solving transfer (Hoeffler et al., 2010). It may also be related to Lamb and Etopio’s (2019) study, which recommended that learning environments offer a combination of visual presentations of information along with text to promote the development of cognitive attributes such as critical thinking and memory.

Students’ preferences for sensorimotor learning, such as hands-on and experiential learning, may reflect their inclination to be active learners. The positive effects of active learning (i.e., student-centered learning) were revealed in a number of studies, indicating a variety of cognitive and affective outcomes (Freeman et al., 2014; Haak et al., 2011), retention of students in the sciences (Graham et al., 2013), contribution to increased interest, appreciation of opportunities to think creatively, increased accountability, and the creation of a community of learners. Student-centered learning is also associated with a reduced sense of isolation compared to teacher-centered learning (Epstein, 2006; Owens et al., 2020; Seymour & Hewitt, 1997). In science learning, hands-on and experiential learning are considered beneficial for achieving a deeper understanding of scientific content, the nature of science, and inquiry practices (Bidarra & Rusman, 2017; Kang & Keinonen, 2018; McComas, 2015; Minner et al., 2010; Yifrach, et al., 2015). Students’ preferences in the current study seem to align well with research findings regarding the benefits of visual and sensorimotor means for learning in general and science learning in particular.

On the other hand, the significantly low preferences for auditory means and agency support in learning contradict evidentiary findings from observations in classrooms showing that the most common method of transmitting information is lecture and verbal explanation (Akkus et al., 2007; Kaya & Kaya, 2020; Kolonich et al., 2018; NRC, 2012; Schwerdt & Wuppermann, 2011). Given that students do not really prefer lecture-based teaching, these findings are disconcerting, as the most common teaching strategy is not aligned with students’ preferences.

Differences in science learning preferences levels through the four learning modalities

Differences in science learning preferences levels by gender

The second research question sought to examine the extent to which gender and having or not having LD affect middle school students’ learning preferences levels in science studies. Our hypothesis was confirmed; the results revealed that girls exhibited higher levels of learning preferences for visual means and agency support than boys. Based on Wrigley-Asante et al. (2023), boys apparently prefer challenges, taking further steps to memorize the scientific principles they learned or use them straight away, while girls require a deeper understanding of the principles and rationale. Therefore, girls may seek more visual support to decrease their cognitive load (Haslam & Hamilton, 2010) and promote learning (Tan et al., 2020). This potential explanation should be investigated in future studies.

The significance of agency support, as expressed in responses to the questionnaire, concerns the teacher’s involvement in task mediation and explicit instruction. Girls’ preferences may be interpreted as their need for more personal teacher attention. This interpretation is supported by studies indicating bias among teachers and in school curricula, giving boys more positive and negative attention in the classroom (Sadker & Sadker, 2010), undermines girls’ self-esteem and discourages them from pursuing non-traditional female careers, such as those involving math and science (Cheryan et al., 2015; Cundiff et al., 2018; Diekman et al., 2017; Liu & Wang, 2019). Another explanation for girls’ preferences for agency support is perhaps related to girls’ unique learning traits. For example, a meta-analysis on women’s leadership in STEM fields (Mccullough, 2011) showed that while men prefer to work with objects, women prefer to work with other people. Female students study for more hours, ask more questions, and seek more feedback on assignments than male students. Other studies (e.g., Asterhan et al., 2012; Chen et al., 2019; Salminen et al., 2012) have indicated that girls prefer agency support in the form of cognitive support and teamwork to promote learning of argumentation and science, with a guided-inquiry approach for girls in middle school (Kim, 2016).

Nevertheless, these insights need to be further examined as the effect size of the differences in this study was moderate. Future research should involve in-depth interviews with students to better understand girls’ perspectives of the four learning modalities.

Furthermore, as this study only addressed a binary system of gender and did not consider how students who do not identify as male or female students might respond to the Learning Preferences Questionnaire, future studies should include views of students who identify as other genders in order to expand our understanding of the learning preferences and needs of a broad diversity of middle school science students.

Differences in science learning preferences levels by with/without ld

No differences in learning preferences levels were found between students with and without LD. Contrary to our hypothesis, students with LD did not have a greater preference for agency support in learning than those without LD. These findings may strengthen researchers’ claims regarding the benefits of multi-faceted methods of instruction to reach all students (Florian & Spratt, 2013; Meyer et al., 2016; Rose et al., 2014; Tomlinson, 2015). Studies have shown that people can encode and represent information in multiple ways, including five sensory-based codes (visual, auditory, tactile, olfactory, and gustatory; e.g., Barsalou, 2008, 2016; Lyman & McDaniel, 1990; Richardson et al., 2003). Activating multiple representations increases memory, learning, and achievements. Students with learning disabilities often have difficulty representing information using one or more modalities (Andersson & Lyxell, 2007; Bull & Scerif, 2001; Geary et al., 2007; Mabbott & Bisanz, 2008; Passolunghi & Siegel, 2004). Thus, as expressed in the students’ preferences, we recommend implementing visual, sensorimotor, and auditory means of instruction and making agency support available. This conclusion is consistent with the findings of other studies showing that activating multiple representations in different situations throughout the learning of a subject, as well as cognitive support, is necessary for all students, particularly those with LD, as the more representations are activated, the better the learning (An & Carr, 2017).

Correlation between science achievements and science learning preferences levels

The variance in learning preferences levels by science achievements was partially in line with Hypothesis 3, with the exception of sensorimotor learning. In other words, the greater the preference for learning through visual and auditory means and with the help of agency support, the higher the achievements in science studies. However, these correlations were low and perhaps were due to students’ preferences for all four modalities. The correlation between preferring visual learning and science achievements can be explained by the highly visual nature of human cognition (Coleman & Willis, 2015; Shanahan et al., 2016; Storbeck et al., 2006) and the contribution of visual representations to learning (Hoeffler et al., 2010). Thus, we may posit that the more the students learn effectively with the help of visual representations in their study materials, the higher their achievements. A possible explanation for the correlation between auditory preferences and science achievements is that science teachers tend to lecture in class (Wendorf, 2018). This type of instruction may benefit students who prefer auditory means or who manage with the instructional norms in class and attain success independently. Learning through auditory means may be less effective for students who have difficulties with verbal representations (i.e., talking, reading, and writing). Thus, multimodal representation and instruction could be an effective instructional approach for those students who find auditory means less favorable. The tendency of teachers to lecture may be an outcome of the national pen-and-paper assessment tests still used in many countries to measure students’ knowledge in the sciences and to assess indicators of school quality and the effectiveness of instruction in achieving curricular goals (Childs & Baird, 2020). The prominent textual component of these exams causes a bias in favor of using more verbal and auditory instructional means due to teachers’ efforts to prepare students to succeed on these tests (Kolonich et al., 2018). Bearing this in mind, the fact that the Scientific Knowledge and Skills Questionnaire was based on a national pen-and-paper assessment test may indicate that these kinds of texts inherently favor students who prefer auditory learning, thus, giving them an advantage. In turn, this advantage may have been reflected in this study in the significant differences in the preferences for auditory learning among students with high scientific knowledge and skills scores relative to students with low science achievements. In fact, many studies have investigated approaches to accommodate learners’ diversity, concluding that by including a variety of alternative assessment options, students can demonstrate their actual knowledge and skills in a manner most appropriate for them (Fu et al., 2019; Hsia et al., 2016; Li et al., 2012; Maker, 2020; Mulder et al., 2014; Tenorio et al., 2016; Topping, 1998; van Popta et al., 2017). Alternative assessment is rooted in educational policy that addresses learner diversity (Schleicher, 2016; UNESCO, 2017).

The correlation found in this study between the preference for agency support in learning (e.g., detailed guidelines, dividing a task into parts, note taking, and so on) and the students’ science achievements can be explained by the effectiveness of these strategies for students with LD (Brigham et al., 2011). This finding supports other studies (Darling-Hammond et al., 2020; Paas et al., 2016), suggesting that it is not only students with LD who can benefit from these strategies; rather, all the students in the class can benefit. However, to indicate the effectiveness of a specific modality, future studies should investigate whether or not students’ achievements are affected by learning modalities when a teacher emphasizes a particular modality in instruction.

Contribution of gender, with/without LD, and science achievements to science learning preferences levels

The multiple regression analyses strengthened our findings regarding the variations in learning preferences levels based on student characteristics (i.e., gender, with/without LD, and science achievements). The combined contribution of the student characteristics to learning preferences levels reveals a significant contribution of girls’ preferences for learning science through visual means and with agency support, as well as a significant contribution of higher science achievements to learning science through auditory means and agency support. However, the contribution of these student characteristics to learning preferences was modest. Therefore, these findings should be further examined in future studies.

Nevertheless, the modest contribution of student characteristics to learning preferences can be interpreted as supporting students’ preferences for all four modalities. The lack of one or more specific preferred learning modalities may be interpreted as a modest correlation.

Moreover, other factors may influence students’ learning preferences, such as culture (Liu & Wang, 2019); teachers’ instructional methods and school norms (Wrigley-Asante et al., 2023); and differences in sensory-based representations, levels of expertise, and self-regulation (An & Carr, 2017). These factors should be given more attention in future studies.

The student characteristic with/without LD (independent variable) did not contribute significantly to any of the four regression analyses. That is to say, having or not having learning disabilities did not contribute to students’ science learning preferences levels in any of the four learning modalities. Although this finding did not confirm our hypothesis, it can be explained. In almost all cases, individuals with LDs are thought to have at least average intelligence or above (Fiedorowicz et al., 2015). Thus, given their intellectual or cognitive potential, a variety of learning modalities may support students with LD similarly to students without LD. For example, structuring learning activities in ways that minimize working memory overload through external representations and memory aids may help in reducing cognitive processing difficulties and storage loads (Asghar et al., 2017) among all students. Therefore, having or not having LD does not necessarily indicate a variance in learning preferences. Future research should involve a qualitative approach and in-depth interviews to better understand how students with LD perceive their learning preferences.

Contribution of gender, with/without LD, and level of science learning preferences to science achievements

The fifth research question sought to examine the contribution of gender, with/without LD, and learning preferences levels to science achievements. The hierarchical regression analysis revealed that girls are likelier to achieve better science scores than boys. This finding is supported by other studies (Deary et al., 2007; Ellis et al., 2008; Fischer et al., 2013; Givord, 2020; Mostafa, 2019; Rahafar et al., 2017) indicating that girls outperform boys in science achievements in school. This accomplishment was found to stem from achievement motivation and self-perceived academic achievement. Controlling by cognitive ability level revealed an even greater achievement difference between the genders in favor of the girls (Fischer et al., 2013).

Findings showed a significant negative contribution of having LD to science achievements. Given that students with learning disabilities have average intelligence (Fiedorowicz et al., 2015) and that all students, especially those with LD, should experience multiple teaching strategies (Kaya & Kaya, 2020; Spektor-Levy et al., 2019), the current situation in which teaching occurs mainly through lectures (Wendorf, 2018), may explain the vulnerability of students with LD (Finson et al., 1997; Hwang et al., 2018; Steele, 2004; Zembylas & Isenbarger, 2002).

Finally, our analysis revealed that learning preferences for auditory means contribute significantly and positively to science achievements beyond the contribution of the other student characteristics (i.e., gender and with/without LD). Our result is reinforced by Utami (2020), who investigated the relationship between learning styles and mathematics, finding that high school mathematics students with auditory learning preferences had greater achievements than other students, even though visual learning was the most dominant learning style. This result sheds light on the possible association between the prevalent verbal teaching strategy implemented in science lessons (Wendorf, 2018) and the higher scores of students who favor auditory learning. This relative advantage should be further explored.

Limitations and further research

This study examined levels of students’ learning preferences in science studies and the personal characteristics that influence those learning preferences, including gender, having learning disabilities, and science achievements (i.e., scientific knowledge and skills). We could only gather data about learning disabilities in certain schools, as information about these conditions is confidential. In some cases, the school administrations did not approve the transfer of this information, even anonymously. In other cases, partial information was submitted by the schools. Thus, the findings regarding the effects of having learning disabilities on learning preferences are based on a sample of 48 students who (with parental permission) volunteered to disclose this information to us. As only 48 of the participants reported having been diagnosed with learning disabilities, dividing this group according to the type of learning disability would have interfered with the statistical power, making the derivation of reliable, valid information unfeasible. While heterogeneous classes comprise students with and without LD, educators tend to treat students with LD differently, perceiving them as low-achievers and offering them modified and accessible assignments. Due to this prevalence of students with LD in mainstream classrooms, we chose to include their preferences in our study; however, we maintained the sample of the diagnosed students as one group. In fact, no differences were found in preferences between students with and without LD, reinforcing the premise regarding implementing multimodal instructional strategies to address the needs of diverse students in the science classroom. Future studies should include a larger sample of students with LD to further explore and verify their learning preferences in accordance with their specific LD type and compare them to learners without LD.

Furthermore, as 48 students disclosed information about learning disabilities, there were perhaps more students with diagnosed learning disabilities who did not provide that information. Additionally, perhaps there was a relatively low proportion of students with LD who were unaware of any disabilities (i.e., they were undiagnosed). In future studies, it is worthwhile considering having students declare a lack of LD or relying on confidential school information after appropriate ethical approval.

This study employed a quantitative methodology. Future research should involve a mixed method (quantitative and qualitative) approach to better understand students with LD and gender-based perspectives of learning preferences through in-depth interviews with students and utilizing more qualitative research tools.

Regarding the effect of the science achievements (i.e., the scientific knowledge and skills scores) on the students’ learning preferences: The participants were rated by their scores on the Scientific Knowledge and Skills Questionnaire developed for this study in line with the national science and technology curriculum. However, the actual topics studied in participating classrooms may have differed from the standard curriculum, which could have affected the students’ scores. Be as it may, the high reliability of this tool, together with the Learning Preferences Questionnaire, the various data analyses we conducted, and the fact that the various analyses revealed similar results strengthen the reliability of the findings in this study. Further examination of learning preferences in science studies and the dynamic interactions between this and other variables of student characteristics should be undertaken with a larger sample and a more diverse student population.

Finally, students’ learning preferences may have been influenced by their exposure to their teachers’ instructional methods, as teachers may favor different modalities to engage students. Therefore, further research can contribute to a better understanding of the relationship between students’ learning preferences and teaching methods that focus on particular learning modalities and how this relationship affects students’ achievements and sense of being heard. Qualitative research tools, like interviews and written self-reports, may reveal such emotional perceptions.

Conclusions

This study examined science learning preferences levels in terms of four learning modalities (visual, sensorimotor, auditory, and agency support) among middle school students. Pearson correlation analyses, analyses of variance, multiple regressions, and hierarchical regressions revealed that the participants favored all four learning modalities, with a significant preference for learning via visual and sensorimotor means. The findings are in line with other studies. For example, Mutua (2015) claimed that most people are multimodal and multi-situational learners. Additionally, psychology and neuroeducation theories contend that multiple representations contribute to better learning, specifically by integrating visual and sensorimotor means into instruction and learning (Lamb, 2016; Lamb et al., 2018; Moriña, 2020).

The implications of this study are twofold: First, teaching methods must be diversified to address the needs of the variety of learners (i.e., gender, learning disabilities, and achievement levels) in the classroom. Thus, in addition to auditory learning, instruction should also integrate visual representations, scaffolding for learning in the learning materials, agency support provided by the teachers, and sensorimotor learning methods, such as building models, playing games, outdoor learning, and so on.

One may argue that just because students exhibit a preference for particular modalities does not necessarily imply that these modalities are universally beneficial. The Universal Design Learning (UDL) approach addresses this claim, advocating that each individual learns differently, with this diversity being the norm rather than the exception. Teaching and learning are highly complex processes. Even the best teacher cannot address the prevalent heterogeneity in the classroom and adapt their teaching to suit everyone. Therefore, building diverse, flexible, and accessible learning pathways that do not rely solely on reading, is required (Glass et al., 2013) to ensure accessibility and participation of all students without lowering expectations and standards from the required accomplishments. However, such multimodal instruction must be accompanied by explicit metacognitive dialogue to facilitate students’ awareness of their learning preferences and to foster self-management of their learning (Dhiman, 2014; Franquesa-Soler et al., 2019; Spektor-Levy et al., 2019). The MuFHA combines UDL multimodal instruction and metacognitive processes to situate the whole learner, and all learners, at the center of the instruction and learning (Spektor-Levy, et al., 2021; Yifrach, et al., 2011).

The second implication is that curriculum developers can view this study as a recommendation to implement explicit agency support and provide more opportunities for sensorimotor engagement in learning.

Above all, our study results underscore the need for educators and policymakers to be attentive to the voices of the students in order to provide all students with equal opportunities and improve achievements in science studies and retain students in the sciences.