Introduction and Literature Review 

Many primary aged learners are limited in their capacity to apply the knowledge and skills gained in their formal science education experiences in ways that make meaning beyond the classroom. Evidence from the Trends in International Mathematics and Science Study (TIMSS) provides robust, if fallible, evidence that most year 4 students across the majority of participating nations have difficulty in generalising their science learning to new contexts (Martin et al., 1997; Thomson et al., 2020a). The most recent iteration of the Australian Sample Assessment in Science Literacy also indicated limitations among primary students’ science knowledge and capabilities (ACARA, 2019) that can persist after the primary years (Thomson & De Bortoli, 2008; Thomson et al., 2019). It can reasonably be argued that systems of science education could be improved in terms of developing learners’ scientific literacy: a central tenant of science education research (Roberts & Bybee, 2014; NASEM, 2016) that is embodied in science curricula globally (ACARA, 2021a; Eggleston, 2018; Kim et al., 2013; NGSS, 2013). This study adopts a broad definition of scientific literacy as a learners’ ability to apply science knowledge and skills to novel contexts whilst recognising the wide-ranging socio-cultural impacts of science advancements (Bybee, 1997; Roberts & Bybee, 2014).

Primary teachers have the most direct role in developing children’s scientific literacy, but they themselves can often face hurdles from their experiences as science learners (Harms & Yager, 1981; Howitt, 2007). Indeed, despite their willingness to pursue engaging, student-centred science teaching practices (ACARA, 2019; 2013; Banilower, 2019), primary teachers can still be limited by their science content knowledge (Appleton, 2003; Murphy & Smith, 2012) and low science teaching efficacy beliefs (STEBs) (Denessen et al., 2015). These issues have been associated with an overreliance on passive, disengaging practices such as note taking, lectures and teacher-driven investigations (Goodrum et al., 2001; Goodrum & Rennie, 2007; Tytler et al., 2008), in addition to insufficient classroom time to satisfy minimum science curricular requirements (Goodrum et al., 2001; Office of the Chief Scientist, 2012; Tytler & Griffiths, 2003; Tytler et al., 2008). Extant literature suggests that these issues are global in scope (Carlone et al., 2011; Osborne & Dillon, 2008; Roth, 2014; Weiss et al., 2003).

These challenges can be compounded by inequitable educational experiences and outcomes between metropolitan and non-metropolitan students (Cardak et al., 2017; Cooper et al., 2018; Cuervo & Acquaro, 2018; Halsey, 2018; OECD, 2013). For at least half a century, research has consistently shown that rural, regional and remote learners (hereafter referred to as non-metropolitan learners) experience greater disadvantage than learners enrolled in metropolitan schools (Human Rights & Equal Opportunity Commission, 2000). Non-metropolitan education can often be diminished by poorer teacher retention, more inexperienced and ‘out-of-field’ teachers, and less relevant curricula, which often results in lower levels of educational attainment and lower likelihood of pursuing higher education (Cardak et al., 2017; Cooper et al., 2018).

Science achievement is no different, with PISA and TIMSS data showing significant differences in the performance of metropolitan and non-metropolitan learners (Fraser et al., 2019; Sullivan et al., 2018). Science education gaps are likely related to staff capacity, resource availability and challenges to learning environments in non-metropolitan schools (Cuervo & Acquaro, 2018; Halsey, 2018; Sullivan et al., 2018). In accordance with trends in the field, the year 4 mathematics, year 8 mathematics and year 8 science assessments in the most recent iteration of the TIMSS all yielded substantial gaps between metropolitan and non-metropolitan schools (Thomson et al., 2020b). For example, the percentages of year 8 science students from regional (67%) and remote (50%) locations at or above the National Proficiency standard were below those from metropolitan centres (77%). Curiously, for the year 4 science assessment, the National Proficiency percentages of remote (74%), regional (75%) and metropolitan (80%) were similar as no statistically significant difference between metropolitan and non-metropolitan was detected (Thomson et al., 2020b). When considered in conjunction with the significant overall increase in Australian year 4 students’ science achievement (Thomson et al., 2020a), further research into these trends of improved educational attainment and geographical equity is warranted. It may be the case that long-term efforts, by a variety of stakeholders, are beginning to influence the quality of Australian primary science teaching (e.g. Deehan, 2021, 2022; Fraser et al., 2019; Fitzgerald et al., 2021; Skamp & Preston, 2021).

It is possible that the recent evidence of more equitable primary science performance between metropolitan and non-metropolitan learners (Thomson et al., 2020a, 2020b) could be reflective of the dissemination of more evidence based (Aubusson et al., 2015, 2019; Deehan et al., 2022), conceptually clear science teaching practices (Harlen, 2015; Roth, 2014) through Initial Teacher Education (ITE) programmes (Deehan, 2021, 2022; Fitzgerald et al., 2021). Such a claim cannot be investigated without an explicit understanding of what may constitute ‘best practice’ in primary science education. Thus, Appendix 1 presents a list of 38 evidence based primary science teaching approaches (Aubusson et al., 2015, 2019; Deehan, 2017, 2022) which serves as a foundation for the investigation of primary science education across metropolitan and non-metropolitan contexts in this paper. The following question will be answered in the paper:

  • Are there differences in the reported science teaching practices and science teaching efficacy beliefs of a sample Australian primary teachers based on school location (metropolitan and non-metropolitan)?

Theoretical Framework

The concept of teacher efficacy has been utilised as a major theoretical underpinning for this study due to its close association with classroom practices and outcomes, which cannot be directly investigated through distal data, and its well-established literature base spanning science education research (Enochs & Riggs, 1990; Riggs & Enochs, 1990), education research (Gibson & Dembo, 1984; Goddard et al., 2000) and psychology research (Bandura, 1977, 1986). Bandura’s seminal unifying theory of behavioural change emphasised the importance of self-efficacy in influencing coping behaviours, exertion of effort and resilience in the performance of desirable actions (Bandura, 1977). Self-efficacy is not an immutable characteristic as it has been shown to vary based on positive or negative influences, with Bandura (1997a, 1997b) himself citing mastery experiences (ME), vicarious experiences, verbal persuasion and emotional arousal (EA) as being the strongest influences on self-efficacy. It is clear such factors would be having complex impacts on teachers’ self-efficacy in all educational settings. Bandura also transitioned the concept of efficacy to teaching practice where he found that ME had the strong impact on teachers’ self-efficacy (Bandura, 1997a, 1997b). This connection between self-efficacy and firsthand MEs provides a reasonable theoretical justification for examining teacher efficacy and reported teaching practices. At the very least, Bandura’s extensive body of work (e.g. Bandura, 1977, 1986, 1997a, 1997b) shows that self-efficacy is a construct that is both associated with desirable actions and malleable, which enhances its conceptual value in educational research. Teacher efficacy (TE) is a measure of an educator’s beliefs in their own and/or their profession’s ability to enhance student learning outcomes (Gibson & Dembo, 1984). TE is a valuable proxy measure in education because higher TE can be indicative of more committed teachers (Chesnut & Burley, 2015; Høigaard et al., 2012) who adopt more effective teaching practice (Klassen & Tze, 2014; Nie et al., 2013) that often result in strong outcomes for their learners (Çoğaltay, & Karadağ, 2017; Goddard et al., 2000). TE has also been consistently operationalised through valid and reliable measures (e.g. Bandura, 1997a, 1997b; Humphries et al., 2012; Lumpe et al., 2000).

The Science Teaching Efficacy Belief Instruments (STEBI-A and STEBI-B) have been consistently employed as measures of inservice and preservice teachers’ STEBs for over three decades (Enochs & Riggs, 1990; Riggs & Enochs, 1990) through a wide variation of research contexts (Deehan, 2017). Like the TE research, the STEBI research has linked primary teachers’ STEBs to desirable outcomes for both teachers and students (Deehan et al, 2020; Clark, 2009). In particular, teachers with higher STEBs are more likely to use more active teaching (Burton, 1996; Lardy, 2011), feel more positively towards their school leaders (Clark, 2009) and, most importantly, improve the science learning outcomes of their students (Angle and Moseley, 2009). There is also a clear gap in addressing the metropolitan and non-metropolitan divides in the STEBI literature as none of the 257 articles considered in a recent meta-analysis addressed this important area of educational equity (Deehan, 2017).

Methodology

A digital survey was used to investigate the potential teaching and efficacy differences between metropolitan and non-metropolitan educators in a jurisdiction of close to 30,000 inservice primary teachers employed across slightly more than 1500 public schools. Quantitative data were collected from an online survey of educators’ primary science teaching practices and efficacy beliefs from mid-to-late 2021.

Context

There were 173 schools (> 10%) from the target population represented in this project. The Australian Curriculum and Reporting Authority (ACARA, 2021b) designates each Australian school’s geolocation based on the five levels of the Australian Statistic Geography Standard (ASGS) Remoteness Structure (ABS, 2016): (1) Major Cities, (2) Inner Regional, (3) Outer Regional, (4) Remote and (5) Very Remote. Table 1 shows the geolocation distribution of the schools sampled versus the target population. The geographical locations are reasonably similar, although the sample is more skewed towards regional locations than the target population. There was no statistically significant difference in the mean Index of Community Socio-Educational Advantage (ACARA, 2016) scores of the sampled and non-sampled schools, t(1594) = 1.845, p = 0.418, meaning that the sampled schools did not vary considerably from the target population in terms of parental occupation, parental education, school remoteness and Indigenous student enrolment percentages.

Table 1 Geolocation distributions of the sample schools (n = 173) and the target population (n > 1500) 

Participants

A series of recruitment approaches were employed as part of a non-probabilistic, purposive sampling of the target population of primary teachers. Two email invitations were sent to each school in the final two semesters of the 2021 school year. Physical mailouts, with QR codes to access the online survey, were sent to each school between email invitations. These primary recruitment strategies were supplemented by opportunistic snowball sampling and sharing across online platforms, both professional and social.

The final sample of 206 primary teachers, representing 0.67% of the target population, was strong in statistical terms (VanVoorhis & Morgan, 2007), and the sampling ratio compares favourably to seminal work in this space (1:150) (Goodrum et al., 2001). Whilst sampling remains ungeneralisable, the characteristics of the sample afford some broader speculative interpretations of the findings. Table 2 summarises the demographic data for all 206 participants.

Table 2 Participant demographic data (N = 206)

Quantitative Survey

The online survey was comprised of three key areas: the STEBI-A, science teaching approaches and curriculum coverage.

The STEBI-A

A selection of 16 5-point Likert scale items comprised the Personal Science Teaching Efficacy (PSTE) beliefs and Science Teaching Outcome Expectancies (STOE) scales (Riggs & Enochs, 1990). The eight PSTE items (e.g. ‘I generally teach science effectively’) are added together to measure participants’ beliefs about their personal effectiveness in science teaching. The eight STOE items (e.g. ‘The inadequacy of a student’s science background can be overcome by good teaching’) measure participants’ more general beliefs about science teaching to affect student learning positively. Upon initial publication, both the PSTE (α = 0.92) and STOE (α = 0.74) scales were valid, with the discrepancy being replicated (e.g. Moslemi & Mousavi, 2019) and discussed in the STEBI literature (Deehan, 2017; Unfried et al., 2022). The PSTE (α = 0.89) and STOE (α = 0.74) scales were reliable in this project (Pallant, 2020).

Science Teaching Approaches

The 38 science teaching approaches from Appendix 1 were presented to participants to dichotomously identify which they utilised in their science teaching practice. They were also afforded the opportunity to identify any additional approaches from their science teaching repertoires that they felt were not represented in the framework. The research teams coded the open responses to remove inappropriate (e.g. ‘Just do it’) and redundant (e.g. ‘Project-based learning’) responses. Open responses deemed appropriate by the research team included resource suites (e.g. ‘Inquisitive’), science fairs and integrated STEM approaches. Each participant was assigned a metric ‘Total Approaches’ score based on the number of relevant approaches selected or otherwise identified in their response.

Curriculum Coverage

Respondents were asked to identify all of the areas of the science curriculum they had addressed in their teaching during the past year. Eleven dichotomous items covered the strands and sub strands of the current Australian K-10 Science Curriculum (ACARA, 2021a). Table 3 organises the 11 curriculum areas presented to participants under the Science Understanding, Science as a Human Endeavour and Science Inquiry Skills. The maximum score on the Curriculum Coverage measure was 11.

Table 3 Australian K-10 science strands and sub strands

Data Analyses

Initially, descriptive statistics were calculated for the measures of science teacher efficacy (i.e. PSTE and STOE) and reported science teaching approaches (i.e. total approaches and curriculum coverage) for the groups of metropolitan or non-metropolitan-based teachers. To determine the difference between the two groups, a one-way ANOVA was computed on the PSTE, STOE, Total Approaches and Curriculum Coverage variables (Pallant, 2020). For additional detail, the magnitude of differences was measured through Hedge’s G to account for the different group sizes. The context, data and variance assumptions for the ANOVA were not violated, at least in part due to the resilience afforded by the large total sample size (VanVoorhis & Morgan, 2007). Non-parametric chi-squares were conducted for the sake of a more thorough interrogation of any between group differences on the reported use of the 38 separate science teaching approaches (Appendix 1).

Results—Are There Differences in the Reported Science Teaching Practices and Science Teaching Efficacy Beliefs of a Sample Australian Primary Teachers Based on School Location (Metropolitan and Non-metropolitan)?

There are very few observable differences between the metropolitan and non-metropolitan groups on the PSTE, STOE, Total Approaches and Curriculum Coverage measures. Table 4 presents the descriptive statistics for the non-metropolitan and metropolitan teachers’ PSTE, STOE, Total Approaches and Curriculum Coverage scores. On both efficacy scales, the metropolitan and non-metropolitan educators displayed similar means and standard deviations, with both groups falling clearly into the ‘somewhat efficacious’ category (i.e. > 24 and < 32). In accordance with much of the existing STEBI literature (Deehan, 2017), the PSTE scores were higher than the STOE scores for both groups. The PSTE scale was the only measure with a between group mean difference greater than one point (1.06). The mean score differences between the metropolitan and non-metropolitan groups on the STOE subscale (0.18) and the Total Approaches measure (0.25) appeared negligible. In fact, Curriculum Coverage did not differ at all between the two geolocation groups.

Table 4 Descriptive statistics for the STEBs of metropolitan- and non-metropolitan-based teachers

There were no statistically significant differences detected between the metropolitan and non-metropolitan groups on the PSTE, STOE, Total Approaches and Curriculum Coverage measures. Table 5 presents the output for a one-way ANOVA on the four dependent variables between the geolocation groups. Despite the effect size (g = 0.194) indicating a small advantage to the metropolitan teachers on the PSTE scale, the difference was not significant (p = 0.162). Additionally, there were no statistically significant between-group differences on the STOE subscale (p = 0.720), the Total Approaches reported (p = 0.791) and the Curriculum Coverage scores (p = 0.992).

Table 5 One-way ANOVA for STEBs, Total Approaches and Curriculum Coverage by geographical location (metropolitan versus non-metropolitan)

More precise chi-square analyses were conducted on participants’ responses to the 38 specific framework approaches, and the number of ‘other’ approaches identified to account for the lack of sensitivity inherent in the broad Total Approaches measure. Table 6 summarises the statistical output for the series of chi-square tests computed to ascertain the differences between the metropolitan and non-metropolitan teachers on the reported use of specific science teaching approaches. In an extension of the results of the one-way ANOVA presented above, there were no significant differences between the groups in the reported uptake of 95% of the specific approaches, including the rate of identification of ‘other’ approaches, a possible indication of similar approaches to science teaching in metropolitan and non-metropolitan primary schools. There were only two teaching approaches that differed in frequency between the two groups. The non-metropolitan educators were more likely to include peer tutoring in their science teaching repertoires, X2 (1, N = 200) = 5.518, p = 0.019, whereas their metropolitan counterparts reported using Debate strategies with comparatively greater frequency, X2 (1, N = 200) = 4.765, p = 0.029. However, the significance of these findings in practice is contestable due to the lower overall frequency of use for Peer Tutoring and Debate approaches.

Table 6 Chi-square analyses of teaching approaches by the metropolitan (n = 91) and non-metropolitan (109) groups (counts and percentages)

Discussion

The findings presented in this paper seem to align with the most recent Australian TIMSS results, which showed no statistically significant differences in metropolitan, regional and rural year 4 students’ science achievement (Thomson et al., 2020b), as there were no substantial differences between metropolitan and non-metropolitan primary teachers’ PSTE, STOE, Total Approach and Curriculum Coverage scores. Even further interrogation of between-group differences for specific teaching practices revealed that reported use rates were similar for 95% of the approaches presented in the framework. A tentative interpretation may be that the similar STEBs and reported science teaching practices of metropolitan and non-metropolitan primary teachers may be influencing more equitable year 4 science outcomes by school location according to the TIMSS (Thomson et al., 2020b). Perhaps, such findings reflect the efforts of teachers, researchers and other educational stakeholders to improve the quality and equity of primary science education (e.g. Aubusson et al., 2015; Aubusson et al., 2019; Fitzgerald et al., 2021; Skamp & Preston, 2021). Indeed, interviews and surveys from a group of 17 primary science academics, supplemented by analysis of publicly available documents, indicated that authentic, accessible and student-centred practices are central in Australian pre-service primary science education (Deehan, 2022). However, any interpretation should be made with a high degree of caution as there is no classroom or student data to confirm reported practice or elucidate how teaching practices relate to student outcomes. Cautious optimism is the best way to interpret these findings as they contradict much of the educational research signalling rural and regional disadvantage. As outlined in the introduction, metropolitan learners have long experienced better short- and long-term educational outcomes than their peers in the regional and rural areas. Recent trends towards equity in Australian primary science warrant further investigation. At the very least, it appears that teachers may be resilient to issues surrounding place, and thus have tremendous potential to contribute to the long-term closing of metropolitan and non-metropolitan educational divides.

There are a number of viable research pathways that could build on this study. First, the absence of significant differences in primary teachers’ STEBs and reported practices between metropolitan and non-metropolitan locations, particularly when considered alongside the equitable rural, regional and metropolitan year 4 science achievement levels in the most recent TIMSS (Thomson et al., 2020b), merits deeper investigation to determine if these findings are aberrations or could inform discourse and decisions surrounding long-standing gaps in metropolitan and non-metropolitan educational outcomes. In particular, the experiences and perspectives of primary science educators and students alike could help to further clarify the nature of geographical location as it relates to primary science education. Second, the data presented in this project represents a single public education jurisdiction in Australia. This means that research at a national scale is needed to determine if these findings are part of a larger pattern of bridging rural, regional or metropolitan divides, or whether the findings are an aberration related to other educational factors, such as teacher traits, funding, resources and others. Similar research should also be pursued to position these findings within a global context. Third, school and classroom level data are vital to addressing the issue of ecological validity commonly associated with large scale quantitative research projects in education (Gorur, 2017) by providing more nuanced, detailed information.

Any interpretation of the findings presented in this manuscript should be tempered by a full understanding of the methodological limitations. Despite providing some useful and methodological defensible insights, the quantitative operationalisation of school locations unavoidably fails to capture the complexity of the lived experiences of those who live and learn in both metropolitan and non-metropolitan communities. Although the ASGS remoteness structure (ABS, 2016) is widely adopted, it cannot cater for issues such as for individual movement between jurisdictions and requires categorisation that may not have any tangible meaning in practice, thus creating an artificial sense of accuracy. Indeed, many educators are likely to have completed their ITE in metropolitan areas and thus may have rural and regional experiences and perspectives that differ from those with more long-term connections to their communities. The importance of the metropolitan- and non-metropolitan-focused analyses presented in this paper should not be overstated as place inherently intersects with factors such as socio-economic status and gender, in ways that were not considered. Additionally, the absence of students as a data source prevents a clear link between the educators’ STEBs and teaching practices, and student outcomes from being established without relying heavily on the theoretical framework (Efficacy). Also, despite the rigour of the framework of approaches (Appendix 1) underpinning this paper, it can neither be a complete reflection of all the approaches that may comprise a primary science teachers’ professional and pedagogical experience repertoire (Loughran et al., 2001, 2004) nor can it capture the complex ways that approaches are instigated and altered in classroom settings. Finally, despite the relative strength of the participant sample, the non-probabilistic recruitment techniques prevent full generalisation of the findings.

Conclusion

The findings presented in this paper contradict much of the existing literature that has described educational divides between metropolitan and non-metropolitan learners. When considered alongside the outlying equitable TIMSS science achievement for Australian year 4 students across regional, rural and metropolitan centres (Thomson et al., 2020a, 2020b), the similar STEBs and reported science teaching between metropolitan and non-metropolitan educators could be indicative of an emerging trend of geographic equity in Australian primary science education. It is important for students to be supported by efficacious teachers who, regardless of school location and status, can overcome general and localised challenges to the provision of high-quality science education (Bandura, 1977, 1986, 1997a, 1997b). This research has shown that both metropolitan and non-metropolitan learners may be equally likely to experience the benefits associated with higher teacher efficacy, including stronger teaching practice (Burton, 1996; Klassen & Tze, 2014; Lardy, 2011; Nie et al., 2013) and better student-outcomes (Angle and Moseley, 2009; Çoğaltay, & Karadağ, 2017; Goddard et al., 2000). The STEB findings presented in this paper are particularly important as non-metropolitan teachers have historically faced considerable challenges in the provision of high quality education (Cardak et al., 2017; Cooper et al., 2018; Cuervo & Acquaro, 2018; Halsey, 2018; OECD, 2013). Whilst the non-probabilistic sampling and reliance on distal data prevent any definitive statements from being made at this time, there is a clear need for further research in this space as there may be insights relevant to addressing the wicked problem of geographical educational disparity.