ArticlePDF Available

Abstract and Figures

Educational Impact and Implications Statement Load Reduction Instruction (LRI) is an instructional approach aimed at managing the cognitive demands placed on students as they learn. LRI encompasses five key principles: (a) reducing the difficulty of instruction during initial learning, as appropriate to learners’ levels of prior knowledge, (b) instructional support and scaffolding, (c) ample structured practice, (d) appropriate provision of instructional feedback-feedforward (combination of corrective information and specific improvement-oriented guidance), and (e) guided independent application. The present study explored the role of science teachers’ LRI in student- and classroom-level science engagement and the role of science engagement in student- and classroom-level science achievement. Findings revealed that student reports of their science teacher’s LRI was significantly associated with higher levels of self-reported engagement, and engagement was significantly associated with higher achievement. Drawing on the five key principles of LRI, the study suggests ways that educators can develop and deliver instruction that appropriately manages the cognitive burden on science students as they learn, and in doing so, enhance these students’ science engagement and in turn, their science achievement.
Content may be subject to copyright.
Load Reduction Instruction
Martin, A.J., Ginns, P., Burns, E., Kennett, R., & Pearson, J. (2020). Load reduction instruction in
science and students’ science engagement and science achievement. Journal of Educational
Psychology. DOI: http://dx.doi.org/10.1037/edu0000552
This article may not exactly replicate the authoritative document published in the journal. It is not
the copy of record. The exact copy of record can be accessed via the DOI:
http://dx.doi.org/10.1037/edu0000552
Load Reduction Instruction
2
Load Reduction Instruction in Science and Students’ Science Engagement and Science
Achievement
Andrew J. Martin1, Paul Ginns2, Emma C. Burns3, Roger Kennett1, Joel Pearson1
1University of New South Wales, Australia
2The University of Sydney, Australia
3Macquarie University, Australia
May 2020
Requests for further information about this investigation can be made to Professor Andrew J.
Martin, School of Education, University of New South Wales, NSW 2052, AUSTRALIA. E-Mail:
andrew.martin@unsw.edu.au. Phone: +61 2 9385 1952. Fax: +61 2 9385 1946.
Acknowledgements: The authors thank the participating schools for assisting with data collection
and Vera Munro-Smith, Carolyn Imre, Brad Papworth, Rebecca Collie, and Herb Marsh for advice
on study design and analysis. This study was funded by the Australian Research Council (Grant
#LP170100253) and The Future Project at The King’s School.
Load Reduction Instruction
3
Submitted May 2020
Abstract
Among a sample of 2,071 Australian high school students nested within 188 science classrooms, the
present study explored the role of science teachers’ “load reduction instruction (LRI; instruction
that seeks to reduce cognitive load by appropriately balancing explicit instruction with guided
autonomy) in student- and classroom-level science engagement and the role of engagement in
student- and classroom-level science achievement. Using doubly-latent multilevel structural
equation modelling, results showed that, at the student- and classroom-level, student reports of their
teacher’s LRI was significantly and positively associated with self-reported engagement, and
engagement was significantly and positively associated with achievement. Thus, (a) LRI was
associated with greater individual student engagement, that in turn was associated with greater
individual student achievement in science, and (b) beyond student-level effects, LRI was associated
with greater classroom engagement that in turn was associated with greater classroom achievement.
We also found that the association between LRI and achievement was mediated by engagement.
Implications for educational practice in science are discussed.
Keywords: load reduction instruction; cognitive load; engagement; achievement; science
Running head: Load Reduction Instruction
Load Reduction Instruction
4
Educational Impact and Implications Statement
Load Reduction Instruction (LRI) is an instructional approach aimed at managing the cognitive
demands placed on students as they learn. LRI encompasses five key principles: (1) reducing the
difficulty of instruction during initial learning, as appropriate to learners’ levels of prior knowledge,
(2) instructional support and scaffolding, (3) ample structured practice, (4) appropriate provision of
instructional feedback-feedforward (combination of corrective information and specific
improvement-oriented guidance), and (5) guided independent application. The present study
explored the role of science teachers’ LRI in student- and classroom-level science engagement and
the role of science engagement in student- and classroom-level science achievement. Findings
revealed that student reports of their science teacher’s LRI was significantly associated with higher
levels of self-reported engagement, and engagement was significantly associated with higher
achievement. Drawing on the five key principles of LRI, the study suggests ways that educators can
develop and deliver instruction that appropriately manages the cognitive burden on science students
as they learn, and in doing so, enhance these students’ science engagement and in turn, their science
achievement.
Load Reduction Instruction
5
Load Reduction Instruction in Science and Students’ Science Engagement and Science
Achievement
Introduction
Students experience substantial academic demands at school and these demands tend to
escalate as students move from one academic year to the next. For example, the frequency and
difficulty of homework, assignments, tests, and subject matter increases with each year of school,
especially at times of major transition such as from elementary school to middle school and then to
high school (Anderman, 2013; Anderman & Mueller, 2010; Graham & Hill, 2003; Martin, Way,
Bobis & Anderson, 2015). Notably, over much of the same time span, research has demonstrated
declines in students’ academic engagement (Burns, Martin, & Collie, 2018; Eccles & Roesser,
2009; Wang & Eccles, 2012). In response to both these challenges (escalating academic demands,
declining engagement), researchers are pointing to the need for instructional approaches that can
help manage the cognitive burden on students that not only assist their learning, but also their
engagement (Evans & Martin, 2020; Martin, 2016; Martin & Evans, 2018, 2019; Moreno, 2010;
Moreno & Mayer, 2007).
Cognitive load theory (CLT) has identified principles of instruction that are aimed at easing
the cognitive burden on students as they learn (Sweller, 2012; Sweller, Ayres, & Kalyuga, 2011).
Harnessing these principles to develop an instructional model, Martin (2016; Martin & Evans,
2018, 2019) proposed “load reduction instruction(LRI) as an approach to help manage the
cognitive demands placed on students as they learn. LRI encompasses five key principles: (1)
reducing the difficulty of instruction during initial learning, as appropriate to learners’ levels of
prior knowledge, (2) instructional support and scaffolding, (3) ample structured practice, (4)
appropriate provision of instructional feedback-feedforward (combination of corrective information
and specific improvement-oriented guidance), and (5) guided independent application (Martin,
2016; see also Martin & Evans, 2018, 2019).
Load Reduction Instruction
6
Complementing the experimental work that is typical under CLT (and which has revealed
causal factors implicated in human learning), preliminary correlational LRI research has been
conducted (in mathematics). This found that LRI is linked with greater mathematics engagement
and achievement (Martin & Evans, 2018) and that LRI is associated with gains in mathematics
motivation, engagement, and achievement across the course of an academic year (Evans & Martin,
2020). With a new sample, the present multilevel study extends that preliminary LRI mathematics-
focused research by: modeling student-level (Level 1; L1) and classroom-level (Level 2; L2)
processes in high school science; using data from a different and much larger sample of students
and classrooms; and, investigating the role of LRI in students’ science achievement via enhanced
academic engagement in science (see Figure 1). Moreover, because we disentangle L2 from L1 LRI
effects, we can gain insight into the role of class-level LRI (L2) beyond variance at L1. Whereas
experimental designs (the predominant method in CLT) provide insight into causal relationships
between factors, this multilevel naturalistic correlational design provides insight into these
associations within and between the very classrooms in which these processes occur. Thus, whilst
not supporting causal conclusions, our design does shed important light on associations at student-
and classroom-levels.
Load Reduction Instruction
CLT identifies two kinds of cognitive load that teachers can impose on students that impede
learning: intrinsic and extraneous (Sweller et al., 2011). Intrinsic cognitive load refers to the
inherent difficulty of instructional material and activity. Intrinsic cognitive load is managed by
teachers when they present instructional material appropriate to students’ level of knowledge
(Sweller et al., 2011). Extraneous cognitive load results from how material is structured and
presented (Sweller et al., 2011). It can be presented clearly, sequentially, and explicitly to students,
so they are guided through learning in a linear and structured fashion, resulting in low extraneous
load. Or, it can be presented in a way that places more of the responsibility on students to work out
the structure of the information, navigate among a range of potential solutions, and/or draw on
Load Reduction Instruction
7
information about which they have relatively little prior knowledge, resulting in higher extraneous
load (Sweller et al., 2011). Extraneous cognitive load does not contribute to learning and is
therefore identified as an unnecessary burden on students (Sweller et al., 2011).
Drawing on major principles of CLT, LRI refers to a range of instructional strategies that
reduce extraneous cognitive load (as the primary yield)as well as some associated intrinsic
cognitive load (as a secondary yield) as these strategies are implemented (Martin, 2016). Reducing
these two types of load is especially important in the early stages of the learning process, when
students are novices (such as when they begin a new subject, a new unit of work, etc.). Importantly,
there is an appropriate time for guided discovery approaches as students acquire the necessary skill
and knowledge (Liem & Martin, 2013; see also Kalyuga, Ayres, Chandler, & Sweller, 2003). This
being the case, LRI contends that explicit and constructivist perspectives are not only compatible,
but essentially interconnectedthe success of one is dependent on the success of the other (Liem &
Martin, 2013; Martin, 2016; Martin & Evans, 2018, 2019). According to LRI, however, it is also the
case that when discovery and exploratory approaches are carried out prior to students having the
requisite skill and knowledge, extraneous cognitive load will impede the effectiveness of this
approach and impede students’ learning (Martin, 2016).
Working Memory, Long-term Memory, Fluency, and Automaticity
When seeking to develop instruction that reduces learners’ cognitive burden, the human
memory system is a key consideration. Working and long-term memory are primary components
for learning (Kirschner, Sweller, & Clark, 2006; Sweller, 2012; Winne & Nesbit, 2010). Working
memory is responsible for receiving and processing information (e.g., solving problems, performing
tasks, etc.), including new and unfamiliar information. Learning takes place when information is
successfully “moved” from working memory and encoded in long-term memory in a way that the
learner can successfully retrieve this information later (Kirschner et al., 2006; Martin & Evans,
2018, 2019; Sweller, 2012; Winne & Nesbit, 2010; see also Anderson & Pearson, 1984 for cognate
perspectives on information retrieval).
Load Reduction Instruction
8
However, working memory is very limited and this is a major challenge for educators when
teaching new material to students (Sweller, Ayres, & Kalyuga, 2011; Winne & Nesbit, 2010). This
is in contrast to long-term memory, which has vast capacity (Sweller, 1993). According to
Kirschner and colleagues: “Any instructional theory that ignores the limits of working memory
when dealing with novel information or ignores the disappearance of those limits when dealing with
familiar information is unlikely to be effective” (2006, p. 77). There is thus a need for instruction
that accommodates the reality of the challenges of working memory and helps students transfer
knowledge between working and long-term memory (Martin, 2015, 2016; Martin & Evans, 2018,
2019; Paas, Renkl, & Sweller, 2003; Sweller, 2003, 2004; Winne & Nesbit, 2010).
Developing students’ fluency and automaticity in skill and knowledge are key means by
which this can occur. Fluency and automaticity free up working memory resources, reduce
cognitive burden, and better enable students to transfer novel information into long-term memory
(Rosenshine, 2009). As Martin and Evans (2019) describe, fluency and automaticity are developed
through LRI’s first four principles: (principle #1) reducing the difficulty of instruction in the initial
stages of learning, as appropriate to the learner’s level of prior knowledge (see also Pollock,
Chandler, & Sweller, 2001; Mayer & Moreno, 2010); (#2) providing appropriate support and
scaffolding to learn the relevant skill and knowledge (see also Renkl, 2014; Renkl & Atkinson,
2010); (#3) allowing sufficient opportunity for practice (see also Nandagopal & Ericsson, 2012;
Purdie & Ellis, 2005; Rosenshine, 2009); and (#4) providing appropriate feedback-feedforward
(combination of corrective information and specific improvement-oriented guidance) as needed (see
also Hattie, 2009; Mayer & Moreno, 2010; Schute, 2008).
As the student’s working memory is freed up with increasing fluency and automaticity, they
are well positioned to apply their fluent and automated skill and knowledge to activities that
previously may have posed too great a cognitive burden on them. These activities include
application of their skill and knowledge to novel tasks, higher order reasoning and thinking,
Load Reduction Instruction
9
problem solving, and guided discovery (Martin, 2016; Martin & Evans, 2019). This represents
principle #5: guided independent learning.
As relevant to this latter principle, CLT researchers have shown that when learners develop
expertise (i.e., when they have sufficient fluency and automaticity in skill and knowledge), they can
be impeded by overly explicit and structured approaches and actually benefit from more open,
problem-solving approachesdemonstrated through the expertise reversal effect (e.g., Kalyuga,
2007; Kalyuga et al., 2003; Kalyuga, Chandler, Tuovinen, & Sweller, 2001). Thus, following
appropriate difficulty reduction, instructional support, guided practice, and feedback-feedforward
from the teacher (i.e., LRI principles #1-4), students now have the necessary prior knowledge,
fluency, and automaticity such that their working memory is not so burdened. It is then that more
demanding subject matter can be presented to them (Kalyuga et al., 2003). In sum, with adequate
explicit and structured foregrounding (i.e., LRI principles #1-4), there is a point in the learning
process when more independence (i.e., LRI principle #5) is vital (Liem & Martin, 2013; Martin,
2016; Martin & Evans, 2018; Mayer, 2004). As we discuss below, when the cognitive burden on
learners is appropriately managed and they develop the requisite fluency and automaticity, they are
likely to engage better with their schoolwork, and in turn this assists their achievement.
Load Reduction Instruction and Academic Outcomes
Prior Related Research
Research has shown that the principles of LRI are positively associated with students’
academic outcomes. For example, meta-analysis by Haas (2005) found that explicit instruction was
the most effective method of teaching algebra, with its success attributable to factors central to LRI
such as scaffolding and some level of guided and independent practice. A meta-analysis by Alfieri
and colleagues (2011) also showed that principles emphasized under LRI are significantly
implicated in students’ achievement, including worked examples and feedback. Following from this
line of research (see also Cromley & Byrnes, 2012; Lee & Anderson, 2013; Liem & Martin, 2013;
Mayer, 2004), Martin (2016) drew numerous connections between LRI and student engagement.
Load Reduction Instruction
10
His review reported that key LRI factors (e.g., sufficient practice and feedback-feedforward)
underpin learning competence (Archer & Hughes, 2011) and that this is likely to enhance students’
academic engagement. Relatedly, research has found greater classroom engagement when
appropriate structure is in place (i.e., teachers provide clear directions, explicit plans for the lesson,
guidance, and feedback; Jang, Reeve, & Deci, 2010; Sierens, Vansteenkiste, Goossens, Soenens, &
Dochy, 2009). Martin and Evans (2018) further proposed that LRI factorssuch as reducing task
difficulty as appropriate to the learner’s level of prior knowledgemay lead to lower
disengagement (Ashcraft & Kirk, 2001; Martin, 2016).
LRI and Engagement in the Present Study
Engagement in the present study is considered from a tripartite perspective, comprising class
participation (behavioral engagement), positive intentions (cognitive engagement), and enjoyment
(emotional engagement; Fredricks, Blumenfeld, & Paris, 2004). This encompasses not only
observable behavioral phenomena such as participation (which has tended to be the dominant
perspective on engagement), but also less demonstrable phenomena such as emotions and cognition
(on these latter dimensions, see Gourlay, 2015 for critique of engagement perspectives that give
inadequate attention to more “private, silent, unobserved” practices). According to Martin (2016),
LRI enhances engagement in numerous ways. First, with regard to participation (behavioral
engagement), LRI is aimed at freeing cognitive resources so students are able to keep up with class
activities and learning material that is central to the lesson. When students are abreast of subject
matter and classroom activities, they are better able to meaningfully participate in class (Finn &
Zimmer, 2012). Second, with regard to emotional engagement (enjoyment), it has been found that
reducing cognitive load in a task enhances an individual’s flow experience in that task, largely
because flow relies on task efficacy and reducing load enables this to happen (Chang, Liang, Chou,
& Lin , 2017). Enjoyment is a construct highly aligned with flow (Csikszentmihalyi, 1990) and thus
we suggest LRI will be positively associated with it. Third, in terms of cognitive engagement
(intentions), reducing cognitive burden optimizes understanding and efficacy (Feldon, Franco,
Load Reduction Instruction
11
Chao, Peugh, & Maahs-Fladung, 2018), which are foundations for positive intentions with regard to
one’s academic future (Burns et al., 2018). On these bases, the present study investigates the link
between LRI in science classrooms and students’ engagement in science. In so doing, this study
expands on recent science-oriented classroom-based research demonstrating that situational factors
in science lessons can enhance science engagement (Inkinen, Klager, Schneider, Juuti, Krajcik,
Lavonen, & Salmela-Aro, 2019), the significant role of teachers’ beliefs in students’ science
engagement (Shumow & Schmidt, 2013), and the significant role of teacher-provided support in
science engagement (Strati, Schmidt, & Maier, 2017). We hypothesize that LRI in science will be
positively associated with engagement in science.
LRI, Engagement, and Achievement
The dominant finding from CLT research is that manipulating extraneous cognitive load can
impact learning. CLT researchers have identified a number of cognitive load effects (e.g., worked
example effect, etc.) that are associated with increases in learning (Sweller et al., 2011). In fact, the
bulk of CLT research has focused on learning as an outcome. There has been relatively little
attention given to engagement and its role in CLT. It is thus noteworthy that many psycho-
educational researchers identify engagement as an important means by which learning occurs, as
well as being a desirable outcome in itself. In terms of the tripartite engagement framework, for
example, researchers would suggest that participating in class helps students acquire better
understanding of a subject area (Credé, Roch, & Kieszczynka, 2010; Green, Liem, Martin, Colmar,
Marsh, & McInerney, 2012; Lysakowski & Walberg, 1982); enjoyment reflects an immersion in
and connection with subject matter that optimizes its acquisition (Martin & Jackson, 2008;
Nakamura & Csikszentmihalyi, 2009); and students’ conceptions of their academic futures impact
their present learning (e.g., Burns, Martin, & Collie, 2019; de Bilde, Vansteenkiste, & Lens, 2011;
Kauffman & Husman, 2004). We therefore hypothesize that engagement in science will be
positively associated with science achievement.
Load Reduction Instruction
12
This being the case, we explore the role of engagement in mediating the link between LRI and
achievement. Thus, whereas CLT (and similar) research tends to investigate learning as being
directly impacted by CLT strategies, we introduce research suggesting that LRI enhances
engagement that in turn enhances learning (as indicated by achievement). We therefore test the
extent to which the effect of LRI on achievement is predominantly mediated by engagement.
Indeed, engagement has been identified as a desirable educational end in itself (not only a means to
other educational ends; e.g., Burns et al., 2019; Fredricks et al., 2004) and thus understanding the
role of LRI in engagement has educational importance. Notwithstanding this, we leave open the
possibility that LRI will also be directly associated with achievement by first testing a “fully
forward” model in which LRI indirectly (via engagement) and directly is linked to achievement.
LRI in Science Education and the Acquisition of Science Concepts
LRI is proposed as an ongoing instructional process intended to appropriately manage
cognitive burden as learners develop. Ultimately this is aimed at building up the body of long-term
knowledge and skill, that leads to the development of more complex mental models and cognitive
schemas (Leppink, 2017)which is particularly important for science students, who will encounter
many concepts that are quite complex and abstract. As the body of knowledge, mental models, and
cognitive schemas develop in long-term memory, the student is capable of visualizing more abstract
ideas, including phenomena with which they may have no direct experience or cannot “see”
(Leppink, 2017). Notably, evidence-based strategies to do this align well with the principles of LRI.
For example, complex and abstract cognitive schemas are better and more readily developed when
novice learners begin with worked examples, then progress to partially worked examples, and then
autonomous application of more complex and abstract concepts (Leppink & van den Heuvel, 2015).
Task fidelity is another factor that can be manipulated. Here, novices might begin by reading about
a scientific phenomenon, then they may escalate to a simulated task, and then they may perform a
real task that is naturalistically more complex and requires abstract reasoning or visualization
(Leppink & van den Heuvel, 2015). These are just two aspects of learning that are relevant to
Load Reduction Instruction
13
managing cognitive load and which are well aligned with LRI principles aimed at helping novice
science students to become more expert in complex tasks and abstract reasoning. To the extent this
is the case, a systematic application of LRI (or similar) over time means that when the science
student encounters complex concepts or is required to engage in abstract reasoning, they have the
cognitive schemas available for them to do so.
We further contend that what is proposed under LRI can be integrated with science education
principles that are widely advocated in schools and school systems. For example, in the present
study’s context (New South Wales; NSW, Australia), science students are required to engage with
real objects, address real problems that occur in the world, and collaborate (Board of Studies NSW,
2012). These can be readily applied under LRI: the focus of students’ practice and independent
application can be conducted with real objects, the science topic or unit within which LRI is
implemented can quite credibly map onto real problems in the world, and collaborative learning has
potential to increase working memory capacity by virtue of the “collective working memory effect”
(Sweller, van Merriënboer, & Paas, 2019; see also Zambrano, Kirschner, & Kirschner, 2019). Thus,
we suggest the principles of LRI are sufficiently broad and applicable enough that they can
accommodate quite diverse tasks, topics, and aims of major science syllabus.
The Load Reduction Instruction Scale (LRIS): Introducing a Short Form
Martin and Evans (2018) developed the Load Reduction Instruction Scale (LRIS), comprising
five factors, to assess the five LRI principles. Each factor is composed of five items (thus, a 25-item
instrument) administered to students, who rate their teacher’s instructional practice. Martin and
Evans (2018) asked students in 40 high school mathematics classrooms to complete the LRIS and
found each factor to be reliable, the overall factor structure to be sound, and significant bivariate
correlations with mathematics motivation, engagement, and achievement. A subsequent study
linking that data with a previous survey (thus, controlling for auto-regression) found LRI was
associated with gains in mathematics motivation, engagement, and achievement (Evans & Martin,
2020).
Load Reduction Instruction
14
The present study extends that initial work in three ways. First, it is conducted in science
classrooms. Science is considered a difficult and stressful school subject by many students (Coe,
Searle, Barmby, Jones, & Higgins, 2008) and there are documented declines in science participation
in this study’s context (Australia; Office of the Chief Scientist, 2014). Science thus represents an
important domain in which to explore instructional practices aimed at reducing the burden on
students as they learn. Second, this study substantially expands the number of classrooms involved
(to 188 classrooms in the present study). Being a study of instructional practice, its validity
demands we get as much coverage across classrooms as possible. Indeed, Evans and Martin (2020)
noted 40 classrooms was a limitation, risking bias in the estimates of standard errors (Maas & Hox,
2005) and recommended future research collect data from more classrooms. Third, we introduce
and explore a brief form of the LRIS (described in Materials). Although long forms are the “gold
standard” in many cases, there are also situations where shorter forms are desirable (Gogol,
Brunner, Goetz, Martin, Ugen, Keller ... & Preckel, 2014); we thus introduce the LRIS-Short
(LRIS-S) as a tool for researchers who may want to assess LRI, but need to do so with a briefer
form. Martin and Evans (2018) verified a reliable and valid higher order LRIS factor in their
research and so we extend that notion to now test a brief unidimensional measure.
Aims of the Present Study
LRI is an instructional approach that seeks to reduce cognitive load on students by
appropriately balancing explicit instruction and guided independence. Preliminary student- and
classroom-level studies have shown that LRI is associated with student motivation, engagement,
and achievement in mathematics. The present study extends that research by collecting data from a
much larger number of classrooms, focusing on science, and testing a new short form of the LRIS.
Specifically, it explored the role of science teachers’ LRI (as reported by students) in science
engagement (also reported by students) and the role of engagement in science achievement (an
objective test). In addition, because we disentangle L2 (classroom) from L1 (student) LRI effects,
we can gain insight into the role of class-level LRI processes (L2) beyond L1 processes. We
Load Reduction Instruction
15
hypothesized that: (a) LRI will be associated with engagement, (b) engagement will be associated
with achievement, and (c) beyond these associations at the student-level (L1), associations among
these factors will also be significant at the classroom-level (L2). Although we expect engagement to
mediate the link between LRI and achievement, we leave open the possibility that LRI will also be
directly associated with achievement. Figure 1 demonstrates the model to be tested.
Method
Participants
The sample for this study comprised 2,071 Australian high school students nested within 188
science classrooms from 8 schools. The schools were from the independent school sector, located in
a major capital city of New South Wales (NSW), on the east coast of Australia. The original sample
comprised 2,199 students, however, we removed students who did not identify their classroom (90
students) as this information was critical for our multilevel modelling. We also removed any classes
where there were less than 3 students in a class, as we deemed these class sizes too small to yield
reliable estimates (38 students from 25 classes; McNeish, 2014). The average class had
approximately 11 students (not unduly disproportionate to the staff-to-student ratio for high schools
in the independent school sector, taking into account non-teaching staff numbers, non-participation,
and student absences; Australian Bureau of Statistics, 2019). Of the 8 schools, 4 were single-sex
boys’ schools and 4 were single-sex girls’ schools. Sixty percent of students were girls. Students
were in Year 7 (29%), Year 8 (22%), Year 9 (24%), and Year 10 (25%)the first 4 years of high
school in Australia. The mean age of students was 14.02 years (SD = 1.27 years). Eight percent of
students spoke a language other than English at home. Students varied in SES (range 846 to 1181,
M = 1138, SD = 41, on the Australian Bureau of Statistics Index of Relative Socio-Economic
Advantage and Disadvantage classification; higher scores indicating higher SES), but in aggregate
were higher than the Australian average of 1000.
Procedure
Load Reduction Instruction
16
Human ethics approval was received from the lead researcher’s university. School principals then
provided approval for their school to participate. Following this, parents/carers and participating students
provided consent. The online survey and science test were administered during a regularly scheduled
science class in the second term (of four school terms) of 2018. Students were asked to complete the
survey and test on their own. Teachers were instructed that they could provide some assistance with
procedural aspects of the survey and test, but not provide students with help in answering specific test
questions.
Materials
Load-Reduction Instruction Scale – Short (LRIS-S). The LRIS-S was developed to
measure student perceptions of their teacher’s use of various strategies known to reduce extraneous
cognitive load (and in the course of this, some intrinsic cognitive load). It was adapted from the full
25-item LRIS that has previously been validated in mathematics (Martin & Evans, 2018). The
LRIS-S is a scale where the five LRI factors are represented by a single item (the full LRIS has 5
items for each of the 5 factors). Each item was drawn from the full LRIS, selected on the basis of
that item being deemed as the most direct single-item reflection of the target factor and it being the
highest loading item for its target factor in the Martin and Evans (2018) study. In two cases an item
was minimally adjusted from the original LRIS (for the practice factor, we removed the words
“many times over” from the end of the item; for the feedback-feedforward factor, the word
“frequent” was used instead of “constructive”). The factors and items (oriented to science) are as
follows: difficulty reduction (“When we learn new things in this science class, the teacher makes it
easy at first); support (“In this science class, the teacher is available for help when we need it);
practice (“In this science class, the teacher makes sure we practice important things we learn);
feedback-feedforward (“In this science class, the teacher provides frequent feedback that helps us
learn); and independence (“Once we know what we’re doing in this science class, the teacher gives
us a chance to work independently). Students responded on a seven-point scale (1 = strongly
disagree to 7 = strongly agree).
Load Reduction Instruction
17
Given the five LRIS factors are highly correlated and the single global LRIS factor has
previously demonstrated strong psychometric fit, a single factor was used, as previously
recommended, to avoid suppression effects and collinearity in modeling (Martin & Evans, 2018).
Indeed, in a study of effective motivational factors, Gogol et al. (2014) found the patterns of
correlations between single-item measures and several outcome measures aligned with correlations
derived with corresponding long scales. Table 1 shows skewness and kurtosis for this 5-item LRIS-
S factor. Table 1 also shows that reliability at L1 and L2 was high (L1ωh = .85; L2ωh = .96). A
preliminary one-factor doubly-latent multilevel congeneric model demonstrated sound fit, 2 (14) =
11.52, p = .65, RMSEA < .001, CFI = 1.00, within-SRMR = .007, between-SRMR = .033, and
yielded acceptable factor loadings (see Table 1).
Student report methodologies have known limitations, such as the potential for students to
misinterpret items or to under- or over-report on a given phenomenon (Karabenick, Woolley,
Friedel, Ammon, Blazevski, Bonney, ... & Kelly, 2007). It is therefore important that where
possible we can provide some support for the use of this methodology. With regard to the LRIS,
because there is an emphasis on reduction of cognitive load in LRI, it is important that the LRIS can
be cross-validated with measures that formally assess key elements of cognitive load, namely
intrinsic load (load referring to task difficulty and complexity) and extraneous load (load referring
to difficulty and complexity of instruction; Chandler & Sweller, 1991). Martin and Evans (2018)
demonstrated that scores on the LRIS were negatively associated with both. Thus, the LRIS does
seem to be accessing aspects of instruction that are meant to impact distinct elements of cognitive
load. In addition, recent research into high school students’ evaluation of teaching suggests these
are reliable and valid means of measuring actual classroom activity (Marsh, Dicke, & Pfeiffer,
2019). As a further point, the multilevel modelling itself accounts for individual student perceptions
of instruction (Level 1) and shared class-level aspects of instruction (Level 2). The latter may be
considered something of a group-level convergence on actual instructional practices (indeed, the
acceptable ICC2 value in Table 1 confirms this). We therefore suggest that student reported LRIS
Load Reduction Instruction
18
and class-level estimations of this offer valid insights into the instructional practices under focus in
this study.
Engagement. We adopted the Fredricks et al. (2004) tripartite approach for our measure of
science engagement, using previously validated scales (Martin & Liem, 2010). This comprised
positive intentions for cognitive engagement (4 items; e.g., “I look forward to continuing with
science at school; L1ωh = .92, L2ωh = .99), class participation for behavioral engagement (4 items;
e.g., “I participate in science class activities; L1ωh = .87, L2ωh = .98), and enjoyment for emotional
engagement (4 items; e.g., “I enjoy science; L1ωh = .94, L2ωh = .99). Students responded on a
seven-point scale (1 = strongly disagree to 7 = strongly agree). Prior research has identified high
correlations among these three factors and suggested modeling them as one engagement factor to
avoid collinearity or suppression effects (Burns, Martin, & Collie, 2018). We therefore created a
scale score for each of the three reliable engagement measures (the mean of the 4 items for each
measure) and used these as indicators of a latent science engagement factor (we did this rather than
estimate a higher order factor in order to reduce the parameters to estimate, especially for the
modeling at L2 where power can be an issue; Kreft & De Leeuw, 1998). Table 1 shows skewness
and kurtosis for this study’s science engagement factor. We found reliability at L1 and L2 (L1ωh =
.81; L2ωh = .92). A preliminary one-factor doubly-latent multilevel congeneric model demonstrated
sound fit, 2 (2) = 3.01, p = .22, RMSEA = .016, CFI = .998, within-SRMR = .010, between-SRMR
= .020, and yielded acceptable factor loadings (see Table 1). As this is a student reported measure,
there are similar validity issues as there are with the student reported LRIS (discussed above).
Major reviews of self-reported engagement suggest that this can be a valid means of assessing
actual engagement, as indicated by its associations with teacher reports of engagement and actual
attendance (Fredricks & McColskey, 2012). Its validity is further supported through many studies
showing students’ self-reported engagement significantly associated with objective measures of
achievement, as would be hypothesized under major engagement frameworks (Fredricks &
McColskey, 2012). In addition, the ICC2 value in Table 1 indicates it is also defensible to model as
Load Reduction Instruction
19
a Level 2 (classroom) construct. We therefore suggest that student reports offer valid insights into
the engagement factors under focus in this study.
Achievement. Science achievement was measured using an online test consisting of 12
questions developed by the head of the science department at a large Sydney school. We opted to
develop our own test in order to avoid potential differences in achievement scores as a function of
differences in school-based assessment tasks (e.g., science course grades). However, we recognize
actual science course grades offer important information and so we developed our test to accord
with the science syllabus applicable to our sample. Thus, two forms were devised, one based on the
Stage 4 (years 7 and 8) NSW science syllabus and the other based on the Stage 5 (years 9 and 10)
syllabus. This being the case, the questions were set within the contexts of Physical World, Earth
and Space, Living World, and Chemical World (NSW science syllabus). This approach ensured that
the science questions aligned with students’ skill level and with material that had been taught. The
test was thus designed to gauge a snapshot of the student’s scientific literacy. The questions were
similar in design to the PISA Science test in that they provided a stimulus and asked questions
arising from the stimulus. Questions were arranged in “units” around each stimulus to reduce the
literacy load and each stimulus contained graphical material to increase access by students of lower
reading comprehension. Each unit contained questions which ranged in difficulty from simple to
complex with key scientific skills featured in each unit. The scientific skills addressed included:
questioning and predicting, planning investigations, conducting investigations, processing and
analyzing data and information, and problem solving. After initial item development, language
accessibility was first evaluated by the head of the languages department at the same school. The
tests were then reviewed by five experienced NSW science teachers who assessed each item based
on (a) alignment with the NSW science syllabus, (b) language and cultural accessibility of question
text and associated graphics, and (c) the estimated proportion of students likely to answer correctly
(response options: 25%, 50%, or 75% of students). All science achievement responses were recoded
as dichotomous (0 = incorrect; 1 = correct). The number of correct answers was summed to create a
Load Reduction Instruction
20
total achievement score (as a continuous scale), and thus reflected something of a formative
construct. Achievement scores were then standardized by year level (M = 0; SD = 1).
Table 1 shows skewness and kurtosis for this study’s science achievement factor. We found
acceptable reliability at L1 and L2 (L1ωh = .79; L2ωh = .86). Table 1 shows acceptable ICC2 to
support its use as a classroom-level construct as well. Being a formative construct comprised of
numerous test questions we did not model this factor by way of its 12 indicators. Instead, we did so
as a total science achievement score and adjusted for unreliability by modeling it as an error-
adjusted variable (not a mean scale score). Error-adjusted scoring was used as it is considered a
more robust alternative to mean scale scores and it helps to avoid inflated or unreliable standard
errors (Kline, 2016). We estimated a student- (L1) and classroom-level (L2) achievement error-
adjusted score using the following equation: σh2 * (1 - ωh), where σh2 was the estimated variance of
achievement at L1 and L2 and ωh was the reliability estimate of achievement (h) at each of L1 and
L2. This yielded a standardized L1 loading of .74 and L2 loading of .93 (see Table 1).
Background attributes. For background attributes, participants reported age (a continuous
measure), gender (0 = male, 1 = female), language background (0 = English speaking, 1 = non-
English speaking), and socioeconomic status (SES) based on home postcode which was then
matched to Australian Bureau of Statistics Index of Relative Socio-Economic Advantage and
Disadvantage (a continuous score, ranging from relatively greater socio-economic disadvantage to
relatively greater socio-economic advantage, national M = 1000).
Data Analysis
The multilevel analyses for this study first comprised calculation of intra-class correlations.
ICC1 was calculated for the average correlation among students in the class (or, between-class
differences). ICC2 was calculated for the reliability of the class average. Multilevel confirmatory
factor analysis (CFA) and multilevel structural equation modelling (SEM) were then conducted
with Mplus version 8 (Muthén & Muthén, 2017). We used the MLR (maximum likelihood robust to
non-normality) estimator for multilevel modeling that provides parameter estimates with standard
Load Reduction Instruction
21
errors and a chi-square test statistic that are robust to non-normality (Muthén & Muthén, 2017). To
assess model fit, a Comparative Fit Index (CFI) greater than .90, a Root Mean Square Error of
Approximation (RMSEA) less than .08, and L1 and L2 Standardized Root Mean Square Residual
(SRMR) less than .10 indicated acceptable fit (Hsu, Kwok, Lin, & Acosta, 2015; Hu & Bentler,
1999; Kline, 2016; Marsh, Hau, & Wen, 2004). Sometimes L2 SRMR can be > .10 and to check if
this is due to measurement problems or inadequate power, the fit of multilevel one-factor
congeneric models are taken into account (Morin, Marsh, Nagengast, & Scalas, 2014). As we
demonstrated in Materials (above), the L2 SRMRs are acceptable in these one-factor congeneric
multilevel models and thus we conclude that any L2 SRMRs > .10 in our main analyses reflect
power issues, not inadequate measurement properties. In line with Morin et al. (2014), in order to
simplify interpretations and reduce non-essential multicollinearity, we standardized the indicators
that were modeled at L1 and L2. A p-value of < .05 was used for tests of statistical significance.
Missing data were dealt with using the Mplus default, Full Information Maximum Likelihood
(FIML; Arbuckle, 1996), an appropriate approach given that the substantive variables were not
missing completely at random, Little's MCAR test: 2 (19) = 17.03, p = .588. To account for the
hierarchical nature of the data, the ‘two-level complex’ and ‘cluster’ commands were used in
Mplus. The ‘two-level’ component enabled us to explicitly model Level 1 (student) and Level 2
(classroom) parameters and to obtain results at these levels; the ‘complex’ component allowed us to
appropriately adjust standard errors due to the clustering of students and classes within schools. We
did not have enough schools to run a three-level model (with schools at Level 3), but we could use
the ‘complex’ command to take into account the sampling structure that causes non-independence
between the observations within a school.
Doubly-latent multilevel CFA was conducted. Doubly-latent models unconfound and control
for measurement error due to sampling of items at Levels 1 and 2. They also adjust for error that is
due to the sampling of students in the aggregation of L1 attributes to form L2 constructs (Marsh,
Lüdtke, Robitzsch, Trautwein, Asparouhov, Muthén, & Nagengast, 2009). For the multilevel CFA,
Load Reduction Instruction
22
at L1 the following factors were included: LRIS-S, engagement, achievement, gender, age,
language background, and SES. At L2, the following factors were included: LRIS-S, engagement,
and achievement (modelled using the same indicators as L1, thus enabling Mplus defaults to
disentangle L1 from L2 variance). As noted in Materials, LRIS-S and engagement were modelled as
latent factors (indicated by component items), achievement was modeled as an error-adjusted score
(adjusted for unreliability), and background attributes were single-item indicators modeled with the
item loading fixed to 1 and uniqueness fixed to 0. We fixed the parallel loadings of L1 and L2 LRIS
and engagement factors (e.g., we fixed the loading of L1 LRIS-S item 1 to equal the loading of L2
LRIS-S item 1). This equates the factor metric across L1 and L2 and enhances construct
comparability across levels. It also enables a more parsimonious model that increases the stability
and accuracy of estimates (Lüdtke, Marsh, Robitzsch, & Trautwein, 2011; Morin et al., 2014).
The hypothesized structural model was tested using doubly-latent multilevel SEM (Marsh et
al., 2009; Morin et al., 2014). The same multilevel measurement model and its constraints
(described above) were the foundation for this hypothesized structural model. At L1, LRI predicts
(the term ‘predicts’ is used to connote a regression path) engagement and engagement predicts
achievement (background attributes also predict LRI, engagement, and achievement). At L2, LRI
predicts engagement and engagement predicts achievement. Modeling in this way enables us to
ascertain class-level effects beyond the effects at the student level. Put another way, because we
disentangle L2 (classroom) from L1 (student) LRI effects, we can gain insight into the role of class-
level LRI processes (L2) beyond L1 processes. This series of paths also enables tests of multilevel
indirect (mediation) effects which were conducted in subsidiary analyses using Sobel (Sobel, 1982)
and Baron and Kenny (1986) methods. Figure 1 shows the multilevel SEM to be tested. When
interpreting model results, we give some emphasis to the L2 (classroom) effects because they are
disentangled from student and residual variance (Marsh, Lüdtke, Nagengast, Trautwein, Morin,
Abduljabbar, & Köller, 2012). Also, in line with Marsh et al. (2012), because LRIS items have the
teacher as the referent, we can interpret it as a climate variable at L2 and because engagement and
Load Reduction Instruction
23
achievement have the student as the referent, we can interpret them as a context variable at L2. The
latter being the case, we can also subtract the L1 engagement → achievement path from the L2
engagement → achievement path to identify the unique contextual effect of engagement on
students’ achievement (see Marsh et al., 2012 for full discussion).
Results
Intra-class Correlations, Multilevel CFA, and Latent Correlations
We first calculated ICC1 for each substantive construct, with results as follows: LRI in
science ICC1 = .16; science engagement ICC1 = .18; science achievement ICC1 = .37. ICC2 for
each construct was as follows: LRI in science ICC2 = .72; science engagement ICC2 = .71; science
achievement ICC2 = .80. With more than 10% of the variance in each factor explained at Level 2
(as per ICC1) and the acceptable ICC2 values, multilevel modelling was justified (Byrne, 2012).
We thus proceeded to the multilevel measurement model that underlies the hypothesized structural
model (Figure 1). This involved doubly-latent multilevel CFA of all constructs (L1 and L2
substantive factors and background attributes). This multilevel CFA also generates latent bivariate
correlations that are the first insight into the relationships hypothesized in Figure 1.
This yielded an acceptable fit to the data, 2 (80) = 474.21, p < .001, RMSEA = .049, CFI =
.938, within-SRMR = .046, between-SRMR = .132 (in Materials we showed that the somewhat
higher between-SRMR is a power issue, not a problem with measurement properties). Latent
correlations are presented in Table 2. Here we summarize only significant correlations among L2
and L1 substantive factors (all non-significant correlations and all correlations with background
attributes are in Table 2). At the classroom-level (L2), LRI is significantly and positively correlated
with engagement (r = .73, p < .001), and to a lesser extent, with achievement (r = .27, p < .05). Also
at L2, engagement is significantly and positively correlated with achievement (r = .74, p < .001). At
the student-level (L1), LRI in science is significantly and positively correlated with science
engagement (r = .61, p < .001) and science achievement (r = .24, p < .001). Also at L1, engagement
is significantly positively correlated with achievement (r = .40, p < .001).
Load Reduction Instruction
24
Multilevel SEM
We then tested the multilevel model in Figure 1 using doubly-latent multilevel SEM.
Although we hypothesized a mediated model (LRI predicting achievement, via engagement), for
completeness we first tested a model with both direct and indirect effects of LRI to achievement (a
“fully-forward” model). However, the direct path from LRI to achievement at L1 was not
significant ( = -.01, p = .93), nor was it significant at L2 ( = -.58, p = .09; note a large standard
error of .35 for this parameter and hence its non-significance); therefore, subsequent analyses
removed this direct path and focused on the hypothesized mediated model. This model yielded an
acceptable fit to the data, 2 (82) = 475.68, p < .001, RMSEA = .048, CFI = .938, within-SRMR =
.046, between-SRMR = .148 (again, in Materials we showed that the somewhat higher between-
SRMR is a power issue, not a problem with measurement properties). Table 3 and Figure 2 show
results. Here we summarize only significant paths among L1 and L2 substantive factors and again
give relatively greater weight to findings at L2, which are purged of L1 variance. All non-
significant paths and all paths associated with background attributes are in Table 3.
At the classroom-level (L2), LRI was significantly and positively associated with engagement
( = .69, p < .001) and engagement was significantly and positively associated with achievement (
= .68, p < .001). Beyond variance attributable to background attributes (gender etc.), at L1 LRI was
significantly and positively associated with engagement ( = .59, p < .001) and engagement was
significantly and positively associated with achievement ( = .33, p < .001). As noted in Methods
and following Marsh et al. (2012), because engagement and achievement can be considered context
factors at L2 we can also ascertain the unique effect of engagement context on students’
achievement as .35 (the difference between L2 and L1 engagement → achievement paths).
We tested the indirect paths from LRI to achievement via engagement. Both were significant:
L2 LRI → engagement → achievement, = .48, p < .001; L1 LRI → engagement achievement,
= .19, p < .001. Because these represent somewhat complex indirect paths as they are among
climate (LRI) and context (engagement, achievement) factors (and are thus to be interpreted
Load Reduction Instruction
25
accordingly), we also applied the Baron and Kenny (1986) method to test mediation. In a first
model, LRI predicted achievement directly and engagement was omitted from the model. This
yielded significant direct LRI → achievement paths at L2 ( = .27, p < .01) and L1 ( = .19, p <
.001). However, when engagement was added as a mediator (thus, LRI → engagement →
achievement), the L2 and L1 LRI → achievement paths were no longer significant, while the L2
and L1 LRI → engagement paths and the L2 and L1 engagement → achievement paths were
significant—leading to our final mediated model (see L1 and L2 βs in Table 3 and Figure 2). The
total variance in engagement explained by all predictors was 49% at L2 (classrooms) and 39% at L1
(students). The total variance in achievement explained by all predictors was 47% at L2 and 27% at
L1.
Given our cross-sectional data, for completeness we also tested alternative model orderings.
Three models were plausible alternatives to the hypothesized model: one where achievement
precedes engagement (perhaps more engaged students achieve more highly); one where
achievement precedes LRI (perhaps teachers use LRI in lower [or higher] achieving classes); and
one where engagement precedes LRI (perhaps teachers use LRI in low [or high] engagement
classes). The first alternative model (LRI → achievement → engagement) yielded a poor fit to the
data, 2 (82) = 1144.22, p < .001, RMSEA = .079, CFI = .834, within-SRMR = .068, between-
SRMR = .287. The second alternative model (achievement → LRI → engagement) yielded an
acceptable fit to the data, but did not surpass that of the hypothesized model, 2 (82) = 570.40, p <
.001, RMSEA = .054, CFI = .924, within-SRMR = .053, between-SRMR = .174. The third
alternative model (engagement → LRI → achievement) yielded an acceptable fit to the data, but did
not surpass that of the hypothesized model, 2 (82) = 570.40, p < .001, RMSEA = .054, CFI = .924,
within-SRMR = .053, between-SRMR = .174. We therefore deferred to the hypothesized ordering
and the findings above.
Discussion
Load Reduction Instruction
26
The present study explored the role of science teachers’ LRI (as reported by students) in
science engagement (also reported by students) and the role of science engagement in science
achievement (an objective test). As hypothesized, at the student-level (L1) we found that LRI was
significantly and positively associated with engagement, and engagement was significantly and
positively associated with achievement. Notably, beyond these student effects at L1, LRI was
associated with greater classroom (L2) engagement that in turn was associated with greater
classroom achievement. Also, at Levels 1 and 2, the link between LRI and achievement was
mediated by engagement.
Findings of Note
Links between LRI, engagement, and achievement.
The finding that engagement mediated the link between LRI and achievement is an important
one. It was also illuminating to find there was no significant direct path between LRI and
achievement (at both L1 and L2), beyond the variance in achievement explained by L1 and L2
engagement (although there was a modest significant positive correlation; Table 2). The bulk of
CLT research has focused on learning outcomes (typically assessed via task performance) when
testing for core effects (e.g., worked example effect, split attention effect, expertise reversal effect,
etc.). Relatively little CLT research has explored the engagement factors that may be implicated in
the learning process (but see Lambert, Kalyuga, & Capan, 2009; Swann, 2013, for examples).
Because LRI is focused on teachers’ classroom instructional practices, we have suggested it is
important to investigate effects known to be implicated in these classroom processesin the case of
this study, academic engagement. The significant role of engagement in the present study might
suggest the need for LRI to more explicitly acknowledge the role of engagement in its processes.
Adopting the tripartite engagement perspective (Fredricks et al., 2004), we investigated the
mediating role of engagement by way of students’ behavioral (classroom participation), cognitive
(positive intentions), and emotional (enjoyment) application in science. We suggest that this
tripartite perspective on engagement can help us better understand how LRI is connected to
Load Reduction Instruction
27
achievement. In relation to behavioral engagement (class participation), freeing cognitive resources
may mean that students are able to keep up with class activities and meaningfully participate;
participation is known to be associated with achievement outcomes (Voelkl, 1995). For cognitive
engagement (intentions), key principles of LRI seek to make instruction more accessible to students
and this may instill in students more positive self-conceptions and more positive orientations to
their academic future; positive intentions are known to be connected to enhanced achievement
(Khattab, 2015). In relation to emotional engagement (enjoyment), alleviating unnecessary
cognitive burden as students learn can help them better enjoy and immerse in what they are doing;
enhanced enjoyment and immersion in one’s subject matter is known to optimize its acquisition
(Martin & Jackson, 2008; Nakamura & Csikszentmihalyi, 2009). More broadly, the salient role of
engagement in achievement is not inconsistent with major perspectives on engagement and its place
in the learning process (Pekrun & Linnenbrink-Garcia, 2012). For example, the agentic elements of
academic engagement in which students exert control over their learning are key to understanding
how learning and achievement occur (Reeve, 2012; Schunk & Mullen, 2012).
However, given the mediating role of engagement is a somewhat novel finding, there is a
need for research to further investigate this. Also, the absence of a direct path between LRI and
achievement should not be misinterpreted. There is experimental evidence that certain
interventions, like worked examples, impact the learning and understanding of targeted concepts
and procedures (Sweller, 2012; Sweller et al., 2011). We suggest that our finding of no direct
relationship is not necessarily contrary to the experimental evidence. For example, we cannot be
sure of the extent to which ideas and procedures targeted by our achievement test were ones
targeted by specific teachers with these specific LRI techniques. We therefore propose that a main
contribution of our study is to highlight the role of LRI in students’ engagement, which is in turn
associated with achievement.
Contributions to debates about instruction.
Load Reduction Instruction
28
Over the decades there has been some tension between predominantly constructivist (and
post-modernist) instructional approaches and predominantly explicit (positivist and post-positivist)
instructional approaches (Martin, 2016; see also Tobias & Duffy, 2009). Our findings suggest that
framing the two as mutually exclusive establishes something of a false dichotomy (Martin, 2016;
Martin & Evans, 2018). Under LRI both are compatible when, having developed requisite skill and
knowledge and having reduced the burden on working memory (LRI principles #1 to #4), learners
are also well placed to apply the acquired skill and knowledge in independent, novel, and creative
ways (LRI principle #5; Martin, 2016; Martin & Evans, 2018). In fact, following research into the
expertise reversal effect (Kalyuga, 2007; Kalyuga et al., 2001; 2003), Martin and Evans (2018)
suggested that learners are “short changed” if they are not given the opportunity to pursue guided
independent application when they have the necessary skill and knowledge. Thus, LRI subsumes
both constructivist and explicit approaches by applying them as appropriate for each individual
learner as they refine their schemas and progress from novice to expert.
In saying this, however, it is important to consider situations where potentially different
approaches may be implementedand how these approaches compare with LRI. As a point of
illustration, we briefly consider three approaches. The first relates to “productive failure” strategies
which involve the design of conditions for novice learners to persist in generating and exploring
solution methods for solving novel, complex problems (Kapur, 2008). Although this approach and
LRI diverge in terms of what would generally be recommended in most initial learning situations,
LRI has recognized that productive failure may be appropriate for tasks or activities where minimal
guidance for novices is desirable. For example, Martin emphasized “the importance of clear
guidance and structure for novices in most learning conditions; however, on occasions where
relatively little guidance for novices is intended, productive failure research might be helpful to set
the conditions that optimize learning in these minimally guided situations (2016, p. 49). The
second relates to “facets-based instruction” that begins by eliciting students’ ideas before instruction
takes place and involves a benchmark lesson (which is cognitively complex, aimed at eliciting
Load Reduction Instruction
29
cognitive conflict and curiosity; Hunt & Minstrell, 1994; Minstrell, 2001). Although this risks high
cognitive load on the novice learner, we point out that students’ ideas are elicited through teachers’
carefully developed pre-instruction questions and the benchmark lesson provides an anchor which
can be revisited periodically throughout a unit that we would suggest are advance and cognitive
organizers creating a mental scaffold that ultimately eases cognitive load (Mayer & Moreno, 2010).
The third is the “inventing with contrasting cases” approach that is a guided discovery strategy
introduced early in the learning process and supported by instructional material (that has plain
language explanations of concepts, examples, etc.) as well as clear parameters (constraints) to help
students stay on track. Proponents of this approach contend that it readies students for the expert
solutions and deep structures they need to develop expertise themselves (Schwartz, Chase,
Oppezzo, & Chin, 2011). While debates about transfer are ongoing, one point we might again draw
here is that the instructional material and pre-set constraints represent something of an advance
organizer creating a cognitive scaffold helpful for the novice learner’s development. Relatedly, CLT
and LRI recognize the value of “signaling” the key learning elements, which if provided in advance
can alert the student to what is essential to learn and establish cognitive scaffolding that can ease
cognitive burden (De Koning, Tabbers, Rikers & Paas, 2009; Martin, 2016; Mayer & Moreno,
2010). Taken together, although these three approaches may appear to depart from what LRI would
recommend for novice learners, we suggest that some ways of implementing them has the potential
to ease cognitive load. Thus, whether it is LRI or another approach that eases cognitive load, the
point is that whatever instructional approach is implemented, appropriately managing cognitive
burden, especially during initial instruction, is important for learning.
The role of students’ background attributes.
Although background attributes (gender, etc.) were not the focus of the study (their inclusion
was primarily to control for their variance to better identify unique variance attributable to LRI,
etc.), their role in L1 science engagement and achievement is worth noting. The most salient of
these factors was gender, with girls significantly lower than boys in science engagement and
Load Reduction Instruction
30
science achievement. This is consistent with longitudinal research finding that girls are lower in
science engagement and achievement, due in part to their lower ability beliefs and importance value
in STEM (Watt, 2004). SES was another significant factor, with higher SES associated with higher
science achievement. This is consistent with meta-analysis by Sirin (2005) and reflects a range of
academic benefits known to be implicated in SES advantage, such as access to educational
resources and educational support when needed. Thus, beyond the role of LRI in students’
engagement and achievement, gender and SES remain factors to address when seeking to optimize
students’ science outcomes.
Implications for Assessment
A major measurement yield of this study was empirical confirmation of the LRIS-Short
(LRIS-S). Factor analysis was conducted at the student- and classroom-level and shown to provide
good fit. Indeed, given we also found significant variance in LRI from class-to-class (by way of
ICC1), it seems the LRIS-S is sensitive to between-class variations in instructional approaches.
Also, given the acceptable class-level reliability (ICC2), it appears the LRIS-S is appropriate to
administer in multilevel research seeking to investigate classroom instructional processes.
Encouragingly, the findings for the short form mirror what was found with the long form in the
Martin and Evans (2018) study. Specifically, the descriptive, reliability, and factor-loading statistics
found for the LRIS-S in this study (Table 1) were much the same as those derived for the higher
order LRIS factor based on the long form in the Martin and Evans (2018) study (see Table 2 in that
study). Also, the L1 (student-level) bivariate correlations found in this study were highly similar to
the student-level findings in Martin and Evans (2018). We therefore conclude that when seeking to
model LRI as an overarching instructional approach (i.e., one factor), the LRIS-S provides an
acceptable alternative to the higher order LRI factor that is based on the long LRIS.
There are also numerous ways that the LRIS (and LRIS-S) can be used in a target domain. For
example, given the nature of its wording, it can be readily administered in different school subjects
(the Evans & Martin [2020] and Martin & Evans [2018] studies were in mathematics; the present
Load Reduction Instruction
31
investigation focused on science). The LRIS could also be used as an observation checklist for
classroom/teacher practice (Martin & Evans, 2019), to assist micro-teaching opportunities (Hattie,
2009). Teachers might also use the LRIS for self-reflection. Depending on what these applications
of the LRIS reveal, teachers might adjust their instructional practiceor, indeed, sustain practices
that align with LRI and its principles.
Implications for Practice
Numerous pedagogical strategies follow from the five principles of LRI (for detail, see
Martin, 2016; Martin & Evans, 2018, 2019). For example, to reduce difficulty in the initial stages of
learning (principle #1), segmenting (or, “chunking”) and pre-training are possible approaches (e.g.,
Mayer & Moreno, 2010; Pollock et al., 2001). For support and scaffolding (principle #2), worked
examples, structured templates, advance and graphic organizers, and prompting have been
suggested (e.g., Berg & Wehby, 2013; Hughes, Regan, & Evmenova, 2019; Renkl, 2014; Renkl &
Atkinson, 2010; Sweller, 2012). For ample practice (principle #3), mental rehearsal and deliberate
practice have been identified (e.g., Ginns, 2005; Nandagopal & Ericsson, 2012; Purdie & Ellis,
2005; Sweller, 2012). For feedback-feedforward (principle #4), corrective and improvement-
oriented information has been proposed, as has the possibility of personal best goal-setting that is
focused on self-improvement (e.g., Basso & Belardinelli, 2006; Burns et al., 2019; Hattie, 2009;
Martin & Liem, 2010). For independence (principle #5), guided discovery learning has been
identified (e.g., Mayer, 2004). In sum, there are numerous evidence-based instructional practices
relevant to the LRI framework.
However, when implementing LRI there are some conditions to note when applying its
principles. First, to most effectively reduce task difficulty in the initial stages of learning it is
important to know the level of students’ prior knowledge; this may require some pre-testing.
Second, if guided autonomy is introduced before the necessary skill and knowledge are sufficiently
fluent and automated, educators risk imposing too great a cognitive burden on learners (Sweller,
2012). Third, if LRI principles #1 to #4 continue to be applied after skill and knowledge have
Load Reduction Instruction
32
become fluent and automated, this can pose a barrier to learning—the “expertise reversal effect”
(Kalyuga, 2007; Kalyuga et al., 2001; 2003). Fourth, LRI argues for guided independent learning
rather than pure independent learning. Whereas guided independence has empirical support (see
Liem & Martin, 2013; Kirschner et al., 2006; Mayer, 2004), frustrations have been raised about
continued advocacy of pure independent learning: “Like some zombie that keeps returning from its
grave, pure discovery continues to have its advocates. However, anyone who takes an evidence-
based approach to educational practice must ask the same question: Where is the evidence that it
works? In spite of calls for free discovery in every decade, the supporting evidence is hard to find”
(Mayer, 2004, p17).
The significant role of engagement as a mediator suggests it might also be a focus for
educational intervention. Considering engagement from a tripartite perspective is helpful in
providing specific direction for educators. Thus, for example, enhancing students’ participation in
tasks and activities in science is a means of boosting behavioral engagement; having students
develop positive future plans and goals for science will assist their cognitive engagement; and,
looking for opportunities to develop tasks and activities that are interesting, fun (where
appropriate), and arouse curiosity may foster greater emotional engagement in science (Burns et al.,
2019; Martin & Liem, 2010; Nagro, Fraser, & Hooks, 2019). Indeed, science is well placed to
activate each of these elements of engagement (Office of the Chief Scientist, 2014).
Limitations and Future Directions
There are several study limitations important to consider when interpreting findings and that
provide some direction for future research. First, although we demonstrated that alternative
orderings of our factors yielded poor fit (or poorer fit than the hypothesized model) and we believe
there are strong conceptual grounds for our hypothesized factor ordering, we concede our data were
cross-sectional and our hypothesized ordering requires examination from a longitudinal perspective.
Evans and Martin (2020) have utilized longitudinal data in high school mathematics to analyze the
relationship between LRI and engagement and achievement; they found that beyond prior variance
Load Reduction Instruction
33
in engagement and achievement, LRI contributed to significant gains in these two outcomes. We
therefore infer that our study’s findings in science would be replicated after controlling for auto-
regression, but this needs to be tested. Simultaneity (i.e., reciprocal causation) is another
longitudinal element that needs to be tested; this, for example, would require us to have prior
achievement data to better understand the process and findings such as the non-significant direct
path between LRI and achievement. Data along these lines would also enable tests of “causal
ordering” (e.g., between LRI and achievement) such as through cross-lagged panel analyses. We
also cannot rule out the existence of a suppressor variable that was not included in the model but
which (if included) may suppress irrelevant variance and lead to lower paths than we found here.
Along similar lines, we cannot rule out the possibility of an omitted variable that is related to both
the dependent variable and one or more independent variable. For example, teacher-student
relationships feature in student learning (Martin & Dowson, 2009) and perhaps teachers who
connect better with students also engage in LRI and find other ways to engage students and help
them achieve. Or, perhaps teachers’ efficacy is related to LRI, engagement, and achievement and is
a factor worth including. Another omitted variable might be disengagement or alienation from one’s
academic studies (Mann, 2001). Moreover, engagement is a broad construct; capturing other
dimensions of it is important for future research. For example, our cognitive engagement construct
(positive intentions) was selected to accord with early (Corno & Mandinach, 1983) and later
(Fredricks et al., 2004; Fredricks & McColskey, 2012) ideas around psychological investment, but
there are other dimensions such as self-regulation and strategy use (Fredricks et al., 2004) that may
be investigated. On a further related note, exploring plausible moderators of students’ perceptions
of LRI is another important next step. For example, to what extent might “need for cognition”
(one’s inclination to engage in and enjoy cognitively effortful tasks; Cacioppo, Petty, Feinstein, &
Jarvis, 1996) amplify the association between LRI on engagement and achievement?
Second, there is a need for experimental work that manipulates LRI principles and explores
for any subsequent shifts in educational outcomes and thus establish (or not) the causal role of LRI.
Load Reduction Instruction
34
Also as noted earlier, the finding that LRI was not associated with achievement directly but was
indirectly linked via engagement suggests the need to build “non-cognitive” factors (such as
engagement and motivation) into experimental work. Third, this study relied on student reports of
LRI. Although we argued support for the validity of this methodology (see Materials) and the ICC2
values were acceptable (indicating class-level reliability), additional indicators such as
observational ratings and teacher self-reports might be used in future to triangulate findings with
students’ reports of LRI. On a related note, there may be principles of instruction that are not
represented under LRI and which are therefore not captured by the questions in the LRIS. We
recommend research to explore if there are instructional principles that appropriately manage the
cognitive burden on (novice) learners but are not represented in the LRI framework. We also
recommend research that can identify different combinations of LRI principles and their
relationships to academic outcomes. For example, latent profile analysis may identify distinct
groups of teachers who balance the five LRI principles in different waysways that may have
significant implications for students’ learning outcomes.
Fourth, although we did have many science classrooms involved in the study, we noted
potential power issues at L2 and so future research designs would do well to resolve these, such as
through larger-scale sampling. Fifth, there is also a need to understand LRI and its role in
engagement in real-time. Researchers have emphasized the in-situ dimensions of students’ science
engagement (Schneider, Krajcik, Lavonen, Salmela‐Aro, Broda, Spicer, ... & Viljaranta, 2016),
confirmed by recent real-time engagement research in other STEM subjects such as mathematics
(Martin, Mansour, & Malmberg, 2019). Sixth, we strived to develop an achievement test that would
be a reasonable in-survey means of capturing what students know about what they have been taught
in science. Given constraints of time, we could not administer an extensive test and so our findings
are bounded by the extent to which items were a representative snapshot of what students have
learned in class. Seventh, the study was conducted in single-sex schools; there is a need for research
in co-educational contexts. Research has suggested gender effects for science engagement may
Load Reduction Instruction
35
differ according to whether a class is co-educational or single-sex in composition (Rowe, 1988;
Shumow & Schmidt, 2013). Our study could not assess this effect in relation to LRI and
engagement. Finally, the study was focused on just one subjectscience. Although other research
has identified the positive role of LRI in mathematics, there is a need for research in non-STEM
areas. It may be that STEM subjects are more (or less) amenable to LRI principles. Relatedly,
exploring LRI and the LRIS in other educational settings is important to test for generalitye.g.,
what is the role of LRI in engagement and achievement among university/college students?
Conclusion
Load reduction instruction (LRI) seeks to reduce cognitive burden on students by
appropriately balancing explicit instruction and guided independence. The present study explored
the link between student-reported LRI and science achievement via student-reported science
engagement as a mediating factor. Findings showed that LRI was associated with engagement,
engagement was associated with achievement, the role of LRI in achievement was significantly
mediated by engagement, and these findings were significant at L2 (classrooms), beyond the
significant effects at L1 (students). Findings hold implications for how teachers deliver instruction
aimed at reducing the cognitive load on students as they learn and for optimizing these students’
academic engagement and achievement.
Load Reduction Instruction
36
References
Alfieri, L., Brooks, P. J., Aldrich, N. J., & Tenenbaum, H. R. (2011). Does discovery-based
instruction enhance learning? Journal of Educational Psychology, 103, 1-18.
https://doi.org/10.1037/a0021017
Anderman, E.M. (2013). Middle school transitions. In J. Hattie & E.M. Anderman (Eds).
International guide to student achievement. New York: Routledge.
Anderman, E.M., & Mueller, C.E. (2010). Middle school transitions and adolescent
development. In J.L. Meece & J.S. Eccles (Eds). Handbook of research on schools,
schooling, and human development. New York: Routledge.
Anderson, R. C., & Pearson, P. D. (1984). A schema-theoretic view of basic processes in reading
comprehension. In P. D. Pearson, R. Barr, M. L. Kamil, & P. Mosenthal (Eds), Handbook
of reading research (pp. 255-291). New York: Longman, Inc.
Arbuckle, J. L. (1996). Full information estimation in the presence of incomplete data. In G. A.
Marcoulides & R. E. Schumacker (Eds.), Advanced structural equation modeling: Issues
and techniques (pp. 243-278). Mahwah, NJ: Lawrence Erlbaum Associates.
Archer, A. L., & Hughes, C. A. (2011). Explicit instruction: Effective and efficient teaching.
New York: Guilford Press.
Ashcraft, M. H., & Kirk, E. P. (2001). The relationships among working memory, math anxiety,
and performance. Journal of Experimental Psychology: General, 130, 224-237.
https://doi.org/10.1037/0096-3445.130.2.224
Australian Bureau of Statistics (2019). Schools Australia. Canberra: Author.
Load Reduction Instruction
37
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social
psychological research: Conceptual, strategic and statistical considerations. Journal of
Personality and Social Psychology, 51, 1173-1182.
Basso, D., & Belardinelli, M. O. (2006). The role of the feedforward paradigm in cognitive
psychology. Cognitive Processing, 7, 73-88. https://doi.org/10.1007/s10339-006-0034-1
Berg, J. L., & Wehby, J. (2013). Preteaching strategies to improve student learning in content
area classes. Intervention in School and Clinic, 49, 14-20.
https://doi.org/10.1177/1053451213480029
Board of Studies NSW (2012). NSW Science Syllabus K-10. Sydney: Author.
Burns, E. C., Martin, A. J., & Collie, R. J. (2018). Adaptability, personal best (PB) goals setting,
and gains in students' academic outcomes: A longitudinal examination from a social
cognitive perspective. Contemporary Educational Psychology, 53, 57-72.
https://doi.org/10.1016/j.cedpsych.2018.02.001
Burns, E.C., Martin, A.J., & Collie, R.J., (2019). A future time perspective of high school
students' academic engagement and disengagement: A longitudinal investigation.
Submitted for publication.
Byrne, B. M. (2012). Structural equation modeling with Mplus: Basic concepts, applications,
and programming. New York, NY: Routledge. https://doi.org/10.4324/9780203807644
Cacioppo, J. T., Petty, R. E., Feinstein, J. A., & Jarvis, W. B. G. (1996). Dispositional
differences in cognitive motivation: The life and times of individuals varying in need for
cognition. Psychological Bulletin, 119, 197-253.
Load Reduction Instruction
38
Chang, C. C., Liang, C., Chou, P. N., & Lin, G. Y. (2017). Is game-based learning better in flow
experience and various types of cognitive load than non-game-based learning? Perspective
from multimedia and media richness. Computers in Human Behavior, 71, 218-227.
Chandler, P., & Sweller, J. (1991). Cognitive load theory and the format of instruction.
Cognition and Instruction, 8(4), 293-332.
Coe, R., Searle, J., Barmby, P., Jones, K., & Higgins, S. (2008). Relative difficulty of
examinations in different subjects. Report for SCORE (Science Community Supporting
Education). Durham, UK: Centre for Evaluation and Modelling, Durham University.
Corno, L., & Mandinach, E. (1983). The role of cognitive engagement in classroom learning and
motivation. Educational Psychologist, 18, 88-108.
Credé, M., Roch, S. G., & Kieszczynka, U. M. (2010). Class attendance in college: A meta-
analytic review of the relationship of class attendance with grades and student
characteristics. Review of Educational Research, 80, 272-295.
https://doi.org/10.3102/0034654310362998
Cromley, J. G., & Byrnes, J. P. (2012). Instruction and cognition. Wiley Interdisciplinary
Reviews: Cognitive Science, 3, 545-553. https://doi.org/10.1002/wcs.1192
Csikszentmihalyi, M. (1990). Flow: The psychology of optimal experience. New York, NY:
Harper & Row.
de Bilde, J., Vansteenkiste, M., & Lens, W. (2011). Understanding the association between
future time perspective and self-regulated learning through the lens of self-determination
theory. Learning and Instruction, 21, 332-344.
https://doi.org/10.1016/j.learninstruc.2010.03.002
Load Reduction Instruction
39
De Koning, B.B., Tabbers, H.K., Rikers, R.M. & Paas, F. (2009). Towards a framework for
attention cueing in instructional animations: Guidelines for research and design.
Educational Psychology Review, 21, 113140.
Eccles, J.S. & Roeser, R.W. (2009). Schools, academic motivation, and stage-environment fit. In
R.M. Lerner & L. Steinber (Eds.) Handbook of adolescent psychology (3rd ed.)(pp. 404-
434). Hoboken, N.J.: John Wiley & Sons.
https://doi.org/10.1002/9780470479193.adlpsy001013
Evans, P., & Martin, A.J. (2020). Load reduction instruction: Multilevel effects on motivation,
engagement, and achievement in mathematics. Submitted for publication.
Feldon, D. F., Franco, J., Chao, J., Peugh, J., & Maahs-Fladung, C. (2018). Self-efficacy change
associated with a cognitive load-based intervention in an undergraduate biology course.
Learning and Instruction, 56, 64-72.
Finn, J. D., & Zimmer, K. S. (2012). Student engagement: What is it? Why does it matter? In S.
L. Christenson, A. L. Reschly, & C. Wylie (Eds.), Handbook of research on student
engagement (p. 97131). New York, NY: Springer. https://doi.org/10.1007/978-1-4614-
2018-7_5
Fredricks, J. A., Blumenfeld, P. C., & Paris, A. H. (2004). School engagement: Potential of the
concept, state of the evidence. Review of Educational Research, 74, 59-109.
https://doi.org/10.3102/00346543074001059
Fredricks, J. A., & McColskey, W. (2012). The measurement of student engagement: A
comparative analysis of various methods and student self-report instruments. In S. L.
Christenson, A. L. Reschly, & C. Wylie (Eds.), Handbook of research on student
engagement (pp. 763-782). New York, NY: Springer. https://doi.org/10.1007/978-1-4614-
2018-7_37
Load Reduction Instruction
40
Ginns, P. (2005). Meta-analysis of the modality effect. Learning and Instruction, 15, 313-331.
https://doi.org/10.1016/j.learninstruc.2005.07.001
Gogol, K., Brunner, M., Goetz, T., Martin, R., Ugen, S., Keller, U., ... & Preckel, F. (2014). "My
questionnaire is too long!" The assessments of motivational-affective constructs with three-
item and single-item measures. Contemporary Educational Psychology, 39, 188-205.
https://doi.org/10.1016/j.cedpsych.2014.04.002
Gourlay, L. (2015). ‘Student engagement’ and the tyranny of participation. Teaching in Higher
Education, 20, 402-411.
Graham, C., & Hill, M. (2003). Negotiating the transition to secondary school. Scottish Council
for Research in Education. Edinburgh: Author.
Green, J., Liem, G. A. D., Martin, A. J., Colmar, S., Marsh, H. W., & McInerney, D. (2012).
Academic motivation, self-concept, engagement, and performance in high school: Key
processes from a longitudinal perspective. Journal of Adolescence, 35, 1111-1122.
https://doi.org/10.1016/j.adolescence.2012.02.016
Haas, M. (2005). Teaching methods for secondary algebra: A meta-analysis of findings. NASSP
Bulletin, 89, 24-46. https://doi.org/10.1177/019263650508964204
Hattie, J. (2009). Visible learning: A synthesis of over 800 meta-analyses relating to
achievement. London and New York: Routledge. https://doi.org/10.4324/9780203887332
Hsu, H. Y., Kwok, O. M., Lin, J. H., & Acosta, S. (2015). Detecting misspecified multilevel
structural equation models with common fit indices: A Monte Carlo study. Multivariate
Behavioral Research, 50, 197-215. https://doi.org/10.1080/00273171.2014.977429
Load Reduction Instruction
41
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis:
Conventional criteria versus new alternatives. Structural Equation Modeling: A
Multidisciplinary Journal, 6, 1-55. https://doi.org/10.1080/10705519909540118
Hughes, M. D., Regan, K. S., & Evmenova, A. (2019). A computer-based graphic organizer with
embedded self-regulated learning strategies to support student writing. Intervention in
School and Clinic. https://doi.org/10.1177/1053451219833026
Hunt, E., & Minstrell, J. (1994). A cognitive approach to the teaching of physics. In K. McGilly
(Ed.), Classroom lessons: Integrating cognitive theory and classroom practice (pp. 51-74).
Cambridge: MIT Press.
Inkinen, J., Klager, C., Schneider, B., Juuti, K., Krajcik, J., Lavonen, J., & Salmela-Aro, K.
(2019). Science classroom activities and student situational engagement. International
Journal of Science Education, 41, 316-329.
Jang, H., Reeve, J., & Deci, E. L. (2010). Engaging students in learning activities: It is not
autonomy support or structure, but autonomy support and structure. Journal of Educational
Psychology, 102, 588-600. https://doi.org/10.1037/a0019682
Kalyuga, S. (2007). Expertise reversal effect and its implications for learner-tailored instruction.
Educational Psychology Review, 19, 509-539. https://doi.org/10.1007/s10648-007-9054-3
Kalyuga, S., Ayres, P., Chandler, P., & Sweller, J. (2003). The expertise reversal effect.
Educational Psychologist, 38, 23-31. https://doi.org/10.1207/S15326985EP3801_4
Kalyuga, S., Chandler, P., Tuovinen, J., & Sweller, J. (2001). When problem solving is superior
to studying worked examples. Journal of Educational Psychology, 93, 579-588.
https://doi.org/10.1037/0022-0663.93.3.579
Kapur, M. (2008). Productive failure. Cognition and Instruction, 26, 379-424
Load Reduction Instruction
42
Karabenick, S. A., Woolley, M. E., Friedel, J. M., Ammon, B. V., Blazevski, J., Bonney, C. R.,
... & Kelly, K. L. (2007). Cognitive processing of self-report items in educational research:
Do they think what we mean? Educational Psychologist, 42, 139-151.
Kauffman, D. F., & Husman, J. (2004). Effects of time perspective on student motivation:
Introduction to a special issue. Educational Psychology Review, 16, 1-7.
https://doi.org/10.1023/B:EDPR.0000012342.37854.58
Khattab, N. (2015). Students’ aspirations, expectations and school achievement: What really
matters? British Educational Research Journal, 41, 731-748.
Kirschner, P.A., Sweller, J., & Clark, R.E. (2006). Why minimal guidance during instruction
does not work: An analysis of the failure of constructivist, discovery, problem-based,
experiential, and inquiry-based teaching. Educational Psychologist, 41, 75-86.
https://doi.org/10.1207/s15326985ep4102_1
Kline, R.B. (2016). Principles and practice of structural equation modeling (4th ed.). New York:
Guilford Press.
Kreft, I. G., & De Leeuw, J. (1998). Introducing multilevel modeling. London: Sage.
Lambert, J., Kalyuga, S., & Capan, L. A. (2009). Student perceptions and cognitive load: What
can they tell us about e-learning Web 2.0 course design? E-Learning and Digital Media, 6,
150-163. https://doi.org/10.2304/elea.2009.6.2.150
Lee, H.S., & Anderson, J.R. (2013). Student learning: What has instruction got to do with it?
Annual Review of Psychology, 64, 445-469. https://doi.org/10.1146/annurev-psych-
113011-143833
Leppink, J. (2017). Cognitive load theory: Practical implications and an important challenge.
Journal of Taibah University Medical Sciences, 12, 385-391.
Load Reduction Instruction
43
Leppink, J., & van den Heuvel, A. (2015). The evolution of cognitive load theory and its
application to medical education. Perspectives on Medical Education, 4, 119-127.
Liem, G.A.D., & Martin, A.J. (2013). Direct instruction and academic achievement. In J. Hattie
& E. Anderman (Eds.). International guide to student achievement. Oxford: Routledge.
Lüdtke, O., Marsh, H. W., Robitzsch, A., & Trautwein, U. (2011). A 2 × 2 taxonomy of
multilevel latent contextual models: Accuracy-bias trade-offs in full and partial error
correction models. Psychological Methods, 16, 444-467. https://doi.org/10.1037/a0024376
Lysakowski, R. S., & Walberg, H. J. (1982). Instructional effects of cues, participation, and
corrective feedback: A quantitative synthesis. American Educational Research Journal, 19,
559-572. https://doi.org/10.3102/00028312019004559
Maas, C. J. M., & Hox, J. J. (2005). Sufficient sample sizes for multilevel modeling.
Methodology, 1, 86-92. https://doi.org/10.1027/1614-2241.1.3.86
Mann, S. J. (2001). Alternative perspectives on the student experience: Alienation and
engagement. Studies in Higher Education, 26, 7-19.
Marsh, H. W., Dicke, T., & Pfeiffer, M. (2019). A tale of two quests: The (almost) non-
overlapping research literatures on students' evaluations of secondary-school and university
teachers. Contemporary Educational Psychology, 58, 1-18.
Marsh, H. W., Hau, K.-T., & Wen, Z. (2004). In search of golden rules: Comment on hypothesis-
testing approaches to setting cutoff values for fit indexes and dangers in overgeneralizing
Hu and Bentler's (1999) findings. Structural Equation Modeling, 11, 320-341.
https://doi.org/10.1207/s15328007sem1103_2
Marsh, H. W., Lüdtke, O., Nagengast, B., Trautwein, U., Morin, A. J., Abduljabbar, A. S., &
Köller, O. (2012). Classroom climate and contextual effects: Conceptual and
Load Reduction Instruction
44
methodological issues in the evaluation of group-level effects. Educational Psychologist,
47, 106-124.
Marsh, H. W., Lüdtke, O., Robitzsch, A., Trautwein, U., Asparouhov, T., Muthén, B., &
Nagengast, B. (2009). Doubly-latent models of school contextual effects: Integrating
multilevel and structural equation approaches to control measurement and sampling error.
Multivariate Behavioral Research, 44, 764-802.
https://doi.org/10.1080/00273170903333665
Martin, A.J. (2016). Using Load Reduction Instruction (LRI) to boost motivation and engagement.
Leicester, UK: British Psychological Society.
Martin, A. J., & Dowson, M. (2009). Interpersonal relationships, motivation, engagement, and
achievement: Yields for theory, current issues, and educational practice. Review of
Educational Research, 79(1), 327-365.
Martin, A. J., & Evans, P. (2018). Load reduction instruction: Exploring a framework that
assesses explicit instruction through to independent learning. Teaching and Teacher
Education, 73, 203-214. https://doi.org/10.1016/j.tate.2018.03.018
Martin, A.J., & Evans, P. (2019). Load reduction instruction: Sequencing explicit instruction and
guided discovery to enhance students' motivation, engagement, learning, and achievement.
In S. Tindall-Ford, S. Agostinho, & J. Sweller (Eds). Advances in cognitive load theory:
Rethinking teaching. New York: Routledge.
Martin, A.J., & Jackson, S.A. (2008). Brief approaches to assessing task absorption and
enhanced subjective experience: Examining 'Short' and 'Core' flow in diverse performance
domains. Motivation and Emotion, 32, 141-157. https://doi.org/10.1007/s11031-008-9094-
0
Load Reduction Instruction
45
Martin, A. J., & Liem, G. A. D. (2010). Academic personal bests (PBs), engagement, and
achievement: A cross-lagged panel analysis. Learning and Individual Differences, 20, 265-
270. https://doi.org/10.1016/j.lindif.2010.01.001
Martin, A.J., Mansour, M., & Malmberg, L-E. (2019). What factors influence students’ real-time
motivation and engagement? An experience sampling study of high school students using
mobile technology. Educational Psychology.
Martin, A.J., Way, J., Bobis, J., & Anderson, J. (2015). Exploring the ups and downs of
mathematics engagement in the middle years of school. Journal of Early Adolescence, 35,
199-244. https://doi.org/10.1177/0272431614529365
Mayer, R.E. (2004). Should there be a three-strikes rule against pure discovery learning? The
case for guided methods of instruction. American Psychologist, 59, 14-19.
https://doi.org/10.1037/0003-066X.59.1.14
Mayer, R.E., & Moreno, R. (2010). Techniques that reduce extraneous cognitive load and
manage intrinsic cognitive load during multimedia learning. (pp. 131-152). In J.L. Plass.,
R. Moreno., & R. Brunken (Eds). Cognitive load theory. Cambridge: Cambridge
University Press. https://doi.org/10.1017/CBO9780511844744.009
McNeish, D. M. (2014). Modeling sparsely clustered data: Design-based, model-based, and
single-level methods. Psychological Methods, 19, 552-563.
Minstrell, J. (2001). The role of the teacher in making sense of classroom experiences and
effecting better learning. In Carver, S. M., & Klahr, D. (eds.), Cognition and instruction:
Twenty-five years of progress (pp. 121149). Mahwah, NJ: Laurence Erlbaum Associates.
Moreno, R. (2010). Cognitive load theory: More food for thought. Instructional Science, 38,
135-141. https://doi.org/10.1007/s11251-009-9122-9
Load Reduction Instruction
46
Moreno, R., & Mayer, R. E. (2007). Interactive multimodal learning environments. Educational
Psychology Review, 19, 309-326. https://doi.org/10.1007/s10648-007-9047-2
Morin, A. J., Marsh, H. W., Nagengast, B., & Scalas, L. F. (2014). Doubly-latent multilevel
analyses of classroom climate: An illustration. The Journal of Experimental Education, 82,
143-167. https://doi.org/10.1080/00220973.2013.769412
Muthén, L. K., & Muthén, B. O. (2017). Mplus [computer software]. Los Angeles: Muthén &
Muthén.
Nagro, S. A., Fraser, D. W., & Hooks, S. D. (2019). Lesson planning with engagement in mind:
Proactive classroom management strategies for curriculum instruction. Intervention in
School and Clinic, 54, 131-140. https://doi.org/10.1177/1053451218767905
Nakamura, J., & Csikszentmihalyi, M. (2009). Flow theory and research. In C. R. Snyder, & S. J.
Lopez (Eds.), Oxford handbook of positive psychology (pp. 195-206). Oxford, MS: Oxford
University Press. https://doi.org/10.1093/oxfordhb/9780195187243.013.0018
Nandagopal, K., & Ericsson, K.A. (2012). Enhancing students' performance in traditional
education: Implications from the expert performance approach and deliberate practice (pp.
257-293). In K.R. Harris., S. Graham., & T. Urdan (Eds). APA educational psychology
handbook. Washington: American Psychological Association.
https://doi.org/10.1037/13273-010
Office of the Chief Scientist (2014). Benchmarking Australian science. Canberra: Australian
Government.
Paas, F., Renkl, A., & Sweller, J. (2003). Cognitive load theory and instructional design: Recent
developments. Educational Psychologist, 38, 1-4.
https://doi.org/10.1207/S15326985EP3801_1
Load Reduction Instruction
47
Pekrun, R., & Linnenbrink-Garcia, L. (2012). Academic emotions and student engagement. In S.
L. Christenson, A. L. Reschly, & C. Wylie (Eds.). Handbook of research on student
engagement (pp. 259-282). New York, NY: Springer. https://doi.org/10.1007/978-1-4614-
2018-7_12
Pollock, E., Chandler, P., & Sweller, J. (2002). Assimilating complex information. Learning and
Instruction, 12, 61-86. https://doi.org/10.1016/S0959-4752(01)00016-0
Purdie, N., & Ellis, L. (2005). A review of the empirical evidence identifying effective
interventions and teaching practices for students with learning difficulties in Years 4, 5,
and 6. Melbourne: Australian Council for Educational Research.
Reeve, J. (2012). A self-determination theory perspective on student engagement. In S.L.
Christenson, A. L. Reschly, & C. Wylie (Eds.). Handbook of research on student
engagement (pp. 149-172). New York. NY: Springer. https://doi.org/10.1007/978-1-4614-
2018-7_7
Renkl, A. (2014). Toward an instructionally oriented theory of example‐based learning.
Cognitive Science, 38, 1-37. https://doi.org/10.1111/cogs.12086
Renkl, A., & Atkinson, R.K. (2010). Learning from worked-out examples and problem solving.
(pp. 91-108). In J.L. Plass., R. Moreno., & R. Brunken (Eds). Cognitive load theory.
Cambridge: Cambridge University Press. https://doi.org/10.1017/CBO9780511844744.007
Rosenshine, B.V. (2009). The empirical support for direct instruction. In S. Tobias & T.M.
Duffy (Eds). Constructivist instruction: Success or failure? New York: Routledge.
Rowe, K. J. (1988). Single-sex and mixed-sex classes: The effects of class type on student
achievement, confidence and participation in mathematics. Australian Journal of
Education, 32, 180-202.
Load Reduction Instruction
48
Schneider, B., Krajcik, J., Lavonen, J., Salmela‐Aro, K., Broda, M., Spicer, J., ... & Viljaranta, J.
(2016). Investigating optimal learning moments in US and Finnish science classes. Journal
of Research in Science Teaching, 53, 400-421.
Schunk, D. H., & Mullen, C. A. (2012). Self-efficacy as an engaged learner. In S. Christenson,
A. Reschly, & C. Wylie (Eds.). Handbook of research on student engagement. New York:
Springer. https://doi.org/10.1007/978-1-4614-2018-7_10
Schute, V.J. (2008). Focus on formative feedback. Review of Educational Research, 78, 153-
189. https://doi.org/10.3102/0034654307313795
Schwartz, D. L., Chase, C. C., Oppezzo, M. A., & Chin, D. B. (2011). Practicing versus
inventing with contrasting cases: The effects of telling first on learning and transfer.
Journal of Educational Psychology, 103, 759-775.
Shumow, L., & Schmidt, J. A. (2013). Academic grades and motivation in high school science
classrooms among male and female students: Associations with teachers’ characteristics,
beliefs and practices. In R. Haumann & G. Zimmer (Eds). Handbook of academic
performance (pp. 53-72). New York: Nova.
Sierens, E., Vansteenkiste, M., Goossens, L., Soenens, B., & Dochy, F. (2009). The synergistic
relationship of perceived autonomy support and structure in the prediction of self-regulated
learning. British Journal of Educational Psychology, 79, 57-68.
https://doi.org/10.1348/000709908X304398
Sirin, S. R. (2005). Socioeconomic status and academic achievement: A meta-analytic review of
research. Review of Educational Research, 75, 417-453.
https://doi.org/10.3102/00346543075003417
Load Reduction Instruction
49
Sobel, M. E. (1982). Asymptotic confidence intervals for indirect effects in structural equation
models. Sociological Methodology, 13, 290-312. https://doi.org/10.2307/270723
Strati, A. D., Schmidt, J. A., & Maier, K. S. (2017). Perceived challenge, teacher support, and
teacher obstruction as predictors of student engagement. Journal of Educational
Psychology, 109, 131-147.
Swann, W. (2013). The impact of applied cognitive learning theory on engagement with e-
learning courseware. Journal of Learning Design, 6, 61-74.
https://doi.org/10.5204/jld.v6i1.119
Sweller, J. (1993). Some cognitive processes and their consequences for the organisation and
presentation of information. Australian Journal of Psychology, 45, 1-8.
https://doi.org/10.1080/00049539308259112
Sweller, J. (2004). Instructional design consequences of an analogy between evolution by natural
selection and human cognitive architecture. Instructional Science, 32, 9-31.
https://doi.org/10.1023/B:TRUC.0000021808.72598.4d
Sweller, J. (2012). Human cognitive architecture: Why some instructional procedures work and
others do not (pp. 295-325). In K.R. Harris., S. Graham., & T. Urdan (Eds). APA
educational psychology handbook. Washington: American Psychological Association.
https://doi.org/10.1037/13273-011
Sweller, J., Ayres, P., & Kalyuga, S. (2011). Cognitive load theory. New York: Springer.
https://doi.org/10.1007/978-1-4419-8126-4
Sweller, J., van Merriënboer, J. J., & Paas, F. (2019). Cognitive architecture and instructional
design: 20 years later. Educational Psychology Review, 1-32.
Load Reduction Instruction
50
Tobias, S., & Duffy, T. M. (Eds.). (2009). Constructivist instruction: Success or failure? New
York, NY: Routledge. https://doi.org/10.4324/9780203878842
Voelkl, K. E. (1995). School warmth, student participation, and achievement. The Journal of
Experimental Education, 63, 127-138.
Wang, M. T., & Eccles, J. S. (2012). Adolescent behavioral, emotional, and cognitive
engagement trajectories in school and their differential relations to educational success.
Journal of Research on Adolescence, 22, 31-39. https://doi.org/10.1111/j.1532-
7795.2011.00753.x
Watt, H. M. G. (2004). Development of adolescents' self-perceptions, values and task
perceptions according to gender and domain in 7th- through 11th-grade Australian
students. Child Development, 75, 1556-1574. https://doi.org/10.1111/j.1467-
8624.2004.00757.x
Winne, P.H., & Nesbit, J.C. (2010). The psychology of academic achievement. Annual Review of
Psychology, 61, 653-678. https://doi.org/10.1146/annurev.psych.093008.100348
Zambrano, J., Kirschner, P. A., & Kirschner, F. (2019). How cognitive load theory can be
applied to collaborative learning: Collaborative cognitive load theory. In S. Tindall-Ford,
S. Agostinho, & J. Sweller (Eds). Advances in Cognitive Load Theory (pp. 30-39).
Routledge. https://doi.org/10.4324/9780429283895-3
Load Reduction Instruction
51
Table 1. Descriptive and measurement properties
Mean
SD
Skew
Kurtosis
ICC
1
ICC
2
L1
(Student)
Omega
L2
(Class)
Omega
L1
(Student)
Factor
Loading
Mean
L2
(Class)
Factor
Loading
Mean
LRI in science
5.29
1.12
-0.92
1.09
.16
.72
.85
.96
.73
.92
Science engagement
5.18
1.25
-0.69
-0.08
.18
.71
.81
.92
.74
.89
Science achievement
0.00
1.00
-0.17
-0.66
.37
.80
.79
.86
.74
.93
Notes: Achievement is a single score and represented by a standardized estimated loading using error-adjusted scoring that
adjusts for unreliability (see Materials); L1 = Level 1; L2 = Level 2; In multi-level modeling, Level 1 also comprises
residual variance; LRI = load reduction instruction; SD = standard deviation; ICC = intraclass correlation.
Load Reduction Instruction
52
Table 2. Multilevel CFA Correlations
1
2
3
4
5
6
7
8
9
10
LEVEL 1: SCIENCE STUDENTS
1. Age
1
2. SES
-.04*
1
3. Gender (M/FM)
.01
-.05
1
4. NESB
-.08**
-.02
.01
1
5. LRI in science
-.15***
.02
-.11***
.01
1
6. Science engagement
-.13***
.03†
-.21***
.04
.61***
1
7. Science achievement
-.09
.08**
-.39***
.06
.24***
.40***
1
LEVEL 2: SCIENCE CLASSROOMS
8. LRI in science
-
-
-
-
-
-
-
1
9. Science engagement
-
-
-
-
-
-
-
.73***
1
10. Science achievement
-
-
-
-
-
-
-
.27*
.74***
1
* p < .05; ** p < .01; *** p < .001; p < .10
Notes: In multi-level modeling, Level 1 also comprises residual variance; LRI = load reduction instruction; SES = socio-economic status; M = male; FM = female; NESB = non-
English speaking background
Load Reduction Instruction
53
Table 3. Doubly Latent Multilevel SEM Results
Science Engagement
Science Achievement
LEVEL 1: SCIENCE STUDENTS
Age
-.04
-.03
SES
.01
.06*
Gender (M/FM)
-.14**
-.33**
NESB
.03
.05
LRI in science
.59**
-
Science engagement
-
.33**
LEVEL 2: SCIENCE CLASSROOMS
LRI in science
.69**
-
Science engagement
-
.68**
Level 1 R2
.39**
.27**
Level 2 R2
.49**
.47**
* p < .05; ** p < .001
Notes: LRI = load reduction instruction; SES = socio-economic status; M = male; FM = female; NESB = non-English
speaking background
Load Reduction Instruction
54
Figure 1. Hypothesized Multi-level Model
Notes: Direct path from LRI to Achievement not hypothesized, but is tested in preliminary analyses for completeness;
LRI = load reduction instruction; SES = socio-economic status; NESB = non-English speaking background
LRI
Engagement
Achievement
LEVEL 2: SCIENCE CLASSROOMS
LRI
Engagement
Achievement
LEVEL 1: SCIENCE STUDENTS
Covariates
- Age
- SES
- Gender
- NESB
Load Reduction Instruction
55
Figure 2. Doubly Latent Multi-level SEM Results
* p < .05; ** p < .01; *** p < .001
Notes: Model controls for variance attributable to background attributes (gender, etc.); In multi-level modeling, Level 1
also comprises residual variance; Only significant substantive paths shown hereall other paths (significant and non-
significant) are presented in Table 3; Indirect paths: Level 1 LRI → Engagement → Achievement, = .19, p < .001;
Level 2 LRI → Engagement → Achievement, = .48, p < .001.
LRI
Engagement
Achievement
LEVEL 2: SCIENCE CLASSROOMS
LRI
Engagement
Achievement
LEVEL 1: SCIENCE STUDENTS
.68***
.69***
.59***
.33***
... This self-management is crucial as too much cognitive load can hinder the achievement of learning goals and impede cognitive development (Costley, 2020). Furthermore, by reducing unnecessary cognitive load, students' affective engagement is enhanced, leading to greater enjoyment and immersion in the learning activity, thus optimizing knowledge acquisition (Martin et al., 2021). ...
Article
Full-text available
This study delved into the self-directed learning strategies employed by university students to construct knowledge and manage cognitive load, aiming to achieve their learning goals. Addressing a gap in understanding how students align these strategies with cognitive and motivational principles, the research aimed to investigate specific strategies and their effectiveness. The novelty of this research lies in its qualitative exploration within diverse educational contexts, contributing practical frameworks for integrating self-directed learning (SDL) with knowledge construction. Employing a qualitative methodology, the study involved interviews with 10 university students from various disciplines to uncover their learning strategies and cognitive load management techniques. Results indicated that students commonly utilized task prioritization, content organization tools, and collaborative discussions for knowledge construction, while managing cognitive load through task breakdown, rest periods, and focus on essential material. Digital platforms and interactive media emerged as critical tools in SDL. In summary, this study selects appropriate SDL strategies for learners, enhances their learning goals, and promotes the implementation of SDG4.
... Load-reducing instructional strategies are specifically intended to minimize task difficulty and enhance learners' support through scaffolding, practice, feedback, and well-designed instructions 16,17 . Such strategies minimize extraneous cognitive load (i.e., load that consumes working memory without contributing to learning) by making a task environment more predictable, thereby optimizing attentional resources and motivation 18 . ...
Article
Full-text available
According to the extant literature, inquiry-based learning instruction can be an effective pedagogical tool as it frames learning as problem-solving activities. However, considerable debate exists about how to implement inquiry-based instruction. The present study compared the academic performance and engagement (as measured by course learning outcomes) of freshmen exposed to one of two inquiry-based learning modes differing in the volume of load-reducing instruction. The context was a writing-intensive, research-oriented course that enrolled second-language speakers with a didactic past. Thus, the study also examined whether the two modes differentiated freshmen’s second-language writing anxiety, a key emotive negative reaction likely to emerge in a writing-intensive course. At the end of the semester, the condition that involved greater load-reducing instruction before problem-solving activities yielded superior attainment of learning outcomes encompassing knowledge acquisition and use of problem-solving skills. No differences in engagement, withdrawal rates, or writing anxiety were detected. Writing anxiety was only minimally related to students’ performance and engagement as measured by the course learning outcomes. These findings illustrate that ostensibly minimal variations in the implementation of inquiry-based learning can have a selective impact on cognitive rather than emotive processing.
... Science engagement has significance in multiple disciplines; thus, the extant literature has been vast and conceptualizations often disparate [13]. In the educational sciences, science engagement has typically been used to predict student interest and achievement in science classes [33][34][35]. Scholars in this area have defined science engagement as consisting of behavioral, emotional, and cognitive factors that may include or be synonymous with interest, enjoyment, motivation, and achievement in science activities in academic contexts [33,34,36,37]. In the communication discipline, science engagement has been defined as communication about scientific or technical subjects that involves scientists, stakeholders, and citizens as well as science outreach focused on promoting, among other attitudes, the public understanding of science by various communicators (e.g., scientists, museum professionals, media professionals, etc., [38]). ...
Article
Full-text available
As science media content creators strive for inclusivity in communication design and promotion, they must consider the influence of audiences’ identities on their engagement with science media. A gender gap in viewership or "missing audience" has been identified for women viewers for educational science content on digital media; one such example of this is Deep Look, a science video series from KQED public media and PBS Digital Studios distributed on YouTube. This study used a mixed method design (1) to examine women’s preferences for Deep Look YouTube video promotions (i.e., episode titles and thumbnail images—the images that act as a preview for the video) to best inform future design of promotional content for these videos to attract more women viewers, and (2) to explore how women’s preferences for science content are linked to their social identities, science identity, and science curiosity. Findings indicated that women’s preferences for promotions for Deep Look YouTube science and nature videos followed expected trends with most women preferring images perceived as visually attractive and colorful more than images perceived as disgusting or gross. However, these preferences were conditional on science curiosity and science identity. Findings indicated that to boost women’s engagement with YouTube science and nature videos, content creators may find it useful to consider how science curious various women audiences are, how strongly women viewers identify with being a science person, and how their most salient social identities motivate engagement.
... In addition, several KI tenets emerging in instructional research guided design in this work (Linn & Eylon, 2011). These include making science relevant with familiar examples (Lee et al., 2007;Walkington, 2013); making thinking visible using models and simulations (Collins et al., 1991;Liu et al., 2015); building on the ideas of others through collaboration (Matuk et al., 2019;Retnowati et al., 2017), and promoting autonomy through supporting students to generate their own questions and findings (Martin et al., 2021;Modrek & Sandoval, 2020). ...
Article
Full-text available
Advances in graphing technologies and learning sciences pedagogy have the potential to equitably support students when exploring complex systems depicting dynamic relationships across multiple disciplinary topics in Science, Technology, Engineering, and Mathematics (STEM). We report on the cumulative impact of science units designed in a Research Practice Partnership (RPP) that leveraged Knowledge Integration (KI) pedagogy to support middle school students to generalize insights to new graph representations and science topics. Teachers across 11 schools incorporated the graph-science units into their curriculum plans. We analyzed ~ 8000 responses to validated and reliable graph-science KI assessment items administered before the first year and after one, two, or three years of instruction aligned with KI pedagogy. With random coefficient, multi-level, mixed-effect regression modeling, we analyzed performance after one-, two-, and three-years of graph-science KI instruction. We also analyzed the growth trajectories of subgroups, i.e., multilingual learners. Data suggest two years of graph-science KI instruction is needed to make significant impacts on student learning and ameliorated the disparity between students with different native language fluencies. These results illustrate the value of technology-enhanced, pedagogically aligned K-12 science instruction that is designed to support connecting diverse graph data and science knowledge comprehensively and cumulatively.
... Many studies explore the use of wearable technology to support social skills development in children with autism spectrum disorder, demonstrating how real-time feedback and prompts can enhance social engagement [5]. Tools that enable teachers to set parameters for group discussions aimed at load reduction instruction, such as time limits and prompts, have been shown to increase student engagement and participation [13]. The Superpower Glass intervention, deployed via Google Glass, is a wearable technology worn by a child, equipped with a smartphone app, promotes facial engagement and emotion recognition by detecting facial expressions and providing reinforcing social cues for children within the Autistic Spectrum [22]. ...
Conference Paper
Full-text available
This paper presents a preliminary design and development of a novel wristband connected to a mobile application to enhance social interaction among children aged 7-12 years who struggle with expressing themselves in group settings. Targeting pedagogical environments, the wristband can facilitate structured group discussions by allowing teachers to set parameters such as time limits via the app. Each child involved in the group discussion would wear an individual wristband, receiving notifications according to the set parameters. The system not only aids teachers in managing inclusive classroom discussions but also benefits parents by providing valuable insights into their child's social development. Moreover, it helps children build communication skills and confidence during group interactions. This innovative technology has the potential to positively impact educational settings by fostering inclusive happiness and a supportive environment for children's social and cognitive growth.
Chapter
In increasingly sophisticated urban environments, smart cities emerged as a focal point for innovation and sustainability. This chapter investigates the role of education in fostering sustainable urban futures through curricula that respond to the dynamic needs. The convergence of technology, pedagogy, and sustainability within educational frameworks is pivotal in preparing learners to navigate and contribute to advanced urban landscapes. Sustainable curriculum design incorporates principles of environmental stewardship, economic viability, and social equity. This chapter examines how integrating these principles into educational programs cultivates a sustainability mindset among learners, preparing them for active participation in sustainable development. It explores the role of smart technologies, innovative pedagogies, and ecological literacy in enhancing educational outcomes. Through global case studies, challenges, and strategies, this chapter equips educators, policymakers, and curriculum developers with the tools to support sustainable development in smart cities.
Article
Full-text available
In the past decade, numerous literature reviews have delved into understanding the factors influencing STEM achievement. While prior studies have concentrated on specific associations, a holistic synthesis of diverse factors is crucial for comprehensive comprehension. This review, guided by Bronfenbrenner's Ecological Systems Theory, offers an overview of factors associated with secondary school students' STEM achievement by scrutinising literature from 2019 to 2023 through content analysis. Within the myriad associations and subfactors, 15 factors across four ecological levels (i.e., individual characteristics, psychological factors, learning abilities, learning approaches and educational involvement at the self‐level; classroom characteristics and culture, teacher characteristics and instructional practices, family support and influence, and access to resources and technology at the microsystem level; teacher‐parent interaction at the mesosystem level; school characteristics and culture, school leadership and practices, school resources and technology, educational policies, and local environment at the exosystem level) emerged. Methodologically, the reviewed studies predominantly employed quantitative analyses, often utilising statistical and variable‐centred approaches, concentrating on science and mathematics domains of STEM, and employing achievement tests. This review sheds light on the current landscape and provides valuable insights for future policies, practices and research directions. Context and implications Rationale for this study Numerous reviews have explored specific factors influencing STEM achievement, emphasising the need for a holistic synthesis of available factors to achieve a comprehensive understanding of the complex interaction between these influences and how they collectively impact student outcomes in STEM education. Why the new findings matter Students' achievement in STEM is influenced by various factors, suggesting that attributing success solely to minimal number of factors may oversimplify the complex nature of the factors that influence achievement. Implications for policy makers, practitioners, the public and researchers A thorough understanding of the factors influencing success in STEM achievement can inform the development of educational policies and investments across different levels, from local schools to national initiatives. Educators can use this knowledge to implement tailored early interventions that enhance student achievement and foster better prospects in STEM careers. Additionally, students and parents can benefit from valuable insights to make informed decisions about educational and career pathways. This review also sheds light on future research directions.
Article
Full-text available
Load Reduction Instruction (LRI) is an instructional framework aimed at managing cognitive demands of learning. This study explored the applicability of LRI beyond English-language contexts in a Chinese-language sample of 1,166 secondary school students from 60 classrooms in Taiwan. Findings supported the viability of LRI and its component factors, correlations with engagement and achievement, and measurement invariance with an English-speaking sample. LRI thus appears to be an appropriate framework for considering instruction that seeks to ease cognitive burden on students in a Chinese-speaking educational context. Contributions and implications for future practice and research are discussed
Article
Full-text available
Breakfast is often cited as the most important meal of the day and vital for students’ academic functioning at school. Although much research has linked students’ breakfast consumption to better achievement, there has been debate about why and how breakfast has academic benefits. The present study of 648 Australian high school students investigated (a) the role of breakfast consumption and breakfast quality in students’ self-reported motivation and their achievement in a science test, (b) the role of motivation in mediating the link between breakfast consumption and quality and students’ achievement, and (c) the extent to which breakfast consumption effects are moderated by the quality of breakfast (e.g., more vegetables, fruit, dairy/protein, wholegrains, cereals, water; less sugary drinks, processed meat, fast take-away and unhealthy snack foods). Findings indicated that beyond the effects of personal, home, and classroom factors, breakfast consumption predicted higher adaptive motivation (p < .05), breakfast quality predicted lower maladaptive motivation (p < .05), and in turn, students’ adaptive (positively, p < .01) and maladaptive (negatively, p < .01) motivation predicted their achievement. Moreover, adaptive motivation significantly mediated the relationship between breakfast consumption and achievement (p < .05). The effect of breakfast consumption was moderated by the quality of breakfast such that consuming a high-quality breakfast in the morning was associated with the highest levels of adaptive motivation (p < .01) and achievement (p < .05) later in the day. Findings have implications for educational practice and policy seeking to promote a healthy start to the school day to optimize students’ motivation and achievement.
Article
Full-text available
This study aimed to understand the role of load reduction instruction (LRI), a framework of instructional strategies designed to reduce cognitive load, and its association with motivation, engagement, and learning outcomes in mathematics. We examined data from 221 students from 39 classrooms and tested a hypothesized longitudinal multilevel model. The results indicated positive associations of LRI with gains in motivation, engagement, and achievement. These effects were evident at both student- and classroom-levels, after accounting for prior variance, and after accounting for age, gender, and socioeconomic status at the student level. The findings are discussed in light of the importance of considering motivation and engagement as vital educational outcomes alongside achievement, as well as the role of cognitive load reduction strategies—including those associated with explicit instruction and independent enquiry—in optimizing these outcomes.
Article
Full-text available
Students' conceptions of their academic futures, such as completing secondary school, have been found to play a significant role in their current behavior. Indeed, research regarding future time perspectives (FTP) indicates that students with extended FTPs are likely to be more engaged and less disengaged over time. Extended FTPs comprise two critical motivating elements: the cognitive (i.e., importance value) and the dynamic (i.e., school completion aspirations). Although these elements are hypothetically reciprocally related and without temporal limitation to their motivational effects, these claims have largely gone untested. These claims were examined via longitudinal structural equation modelling with cross-lagged panel analysis and invariance testing in a sample of 1327 Australian secondary school students. Findings indicated that importance value is directionally salient over school completion aspirations (such that it may precede school completion aspirations), both are associated with higher engagement and lower disengagement over time, and evidence of temporal limitations on the motivational benefits of the elements of extended FTPs was not found. School-based interventions that focus on improving importance value and school completion aspirations are discussed.
Article
Full-text available
Cognitive load theory was introduced in the 1980s as an instructional design theory based on several uncontroversial aspects of human cognitive architecture. Our knowledge of many of the characteristics of working memory, long-term memory and the relations between them had been well-established for many decades prior to the introduction of the theory. Curiously, this knowledge had had a limited impact on the field of instructional design with most instructional design recommendations proceeding as though working memory and long-term memory did not exist. In contrast, cognitive load theory emphasised that all novel information first is processed by a capacity and duration limited working memory and then stored in an unlimited long-term memory for later use. Once information is stored in long-term memory, the capacity and duration limits of working memory disappear transforming our ability to function. By the late 1990s, sufficient data had been collected using the theory to warrant an extended analysis resulting in the publication of Sweller et al. (Educational Psychology Review, 10, 251–296, 1998). Extensive further theoretical and empirical work have been carried out since that time and this paper is an attempt to summarise the last 20 years of cognitive load theory and to sketch directions for future research.
Chapter
Full-text available
John Sweller: "Zambrano, R. Kirschner and F. Kirschner in Chapter 3 correctly point out that cognitive load theory has been primarily concerned with individual learning rather than collaborative learning. They also point out that the field of collaborative learning currently does not have a unifying theory and that cognitive load theory is ideally placed to fulfil that role. They provide a compelling case by using the theory to explain when collaborative learning should and should not be effective." (p. 8).
Article
Full-text available
Using mobile technology and experience sampling in junior high school, real-time motivation and engagement were explored at four-levels: between lessons (up to 2 lessons per day; Level 1), between days (5 days per week; L2), between weeks (4 weeks; L3), and between students (113 students; L4). Findings for a ‘random effects’ model revealed substantial variance between students (M = 67%), followed by variance between lessons (M = 29%), with little variance between days and between weeks (M = 2%). In a ‘fixed and random effects’ model, we explored the roles of prior general academic motivation and engagement as well as ability and socio-demographics. Findings revealed that prior general academic motivation and engagement played a significant role in students’ real-time motivation and engagement in mathematics and English and reduced between-student variance by about half. Gender (females), ability (high), and socioeconomic status (high) were also associated with higher real-time motivation and engagement.
Article
Written expression is often difficult for students with and without learning disabilities. Research-based strategies to support this essential skill include self-regulated learning strategies, technology, and tools for planning and writing. This article describes a multifaceted technology-based writing intervention with embedded self-regulated learning strategies known as a computer-based graphic organizer or CBGO. The CBGO is described to assist teachers in their efforts to support upper elementary and middle school students to plan and write a high-quality persuasive paragraph. Recommendations for instructional use are discussed.
Article
Many 1000s of studies have been conducted on the validity and diagnostic usefulness of students' evaluations of university teaching (SET), but there is a surprising lack of research on ratings by secondary students. Integrating these two disparate research areas, we evaluate the appropriateness of university SET instruments to secondary settings. Secondary students evaluated an effective and less effective teacher using items adapted from two university instruments, supplemented by items for secondary settings, and rated the appropriateness and importance of each item (N = 761 sets of ratings of more than 400 teachers, Years 7–11, 10 schools). All items were seen as appropriate and important. Factor analyses of responses to both instruments supported their a priori factor structure, and multitrait-multimethod analyses supported their convergent and discriminant validity. We discuss directions for further research at the secondary level based on the extensive body of research on the reliability, validity, and usefulness of SETs at the university level.