View PDF VersionPrevious ArticleNext Article
 
DOI: 10.1039/C5RP00140D (Paper) Chem. Educ. Res. Pract., 2016, 17, 58-71

Features of representations in general chemistry textbooks: a peek through the lens of the cognitive load theory

James M. Nyachwaya *a and Merry Gillaspie b
aDepartment of Chemistry and Biochemistry, and School of Education, North Dakota State University, P.O Box 6050, Fargo, ND 58108, USA. E-mail: james.nyachwaya@ndsu.edu
bDepartment of Chemistry, Wartburg College, 100 Wartburg Blvd, Waverly, IA 50677

Received 23rd July 2015 , Accepted 29th September 2015

First published on 29th September 2015

Abstract

The goals of this study were (1) determine the prevalence of various features of representations in five general chemistry textbooks used in the United States, and (2) use cognitive load theory to draw implications of the various features of analyzed representations. We adapted the Graphical Analysis Protocol (GAP) (Slough et al., 2010) to look at the type of representations used, the function of each representation, the physical integration of representations with associated text, the presence and nature of captions and labels, the indexing of representations, and the number of representations requiring conceptual integration on a given page. Results indicate that on average, in all five textbooks each page had at least four representations. Most representations served a ‘representational’ function, but a number functioned as decorative representations. Most representations were directly integrated with text, but some of the remaining representations were separated by a whole page from associated text. While many pages had an average of two representations that required conceptual integration with text or other representations, some pages had as many as six representations requiring integration. While using textbooks, learners can experience intrinsic, germane or extraneous cognitive load (Sweller, 1994). Our findings indicate that there are various features of representations that could help reduce intrinsic or extraneous cognitive load. However, we also found prevalent features of representations that imply high intrinsic cognitive load or are likely to lead to extraneous cognitive load. Implications for textbook authors and editors, textbook selection, instruction, and science teacher preparation are discussed.

Introduction

Alongside text, the discipline of chemistry commonly employs multiple representations to communicate content (Kozma, 2003; Ainsworth, 2008). Kozma and Russel (1997) defined multiple representations as involving models, illustrations of processes, diagrams, graphs, equations, formulae and charts among others. Johnstone (1991), defined three levels or types of representation: macroscopic, symbolic, sub-microscopic or particulate levels. According to Johnstone (1991), the macroscopic level involves observable phenomena that could be experienced through touch, smell and sight; the symbolic level involves symbols, chemical formulas, and graphs. The submicroscopic, or particulate level, involves particles such as atoms, molecules and ions. Johnstone's ideas have been very influential in driving research and curriculum in chemistry education for over two decades (Talanquer, 2011; Taber, 2013). Over the years, as the field of chemistry education has continued to grow, more levels or types of representation have been suggested. For example, Dori and Hameiri (2003) suggested a process level, Dangur et al. (2014) suggested a quantum level which they defined as encompassing understanding of the relationship between quantum mechanics and the electronic structure of atoms, molecules and the solid state and Gkitzia et al. (2011) proposed the existence of multiple, mixed and hybrid representations which are a combination of two or more of Johnstone's (1991) three levels of representation. Multiple representations in chemistry encompass the levels or types of representation identified above. Regardless of which terms or levels one chooses to use, chemistry is a multi-representational discipline. Text is more general, presenting information in a more explicit manner by using sentences. On a page with representations, the (written) text acts as the main source of information with representations providing a supporting function in communicating ideas (Corradi et al., 2012).

Learners pay the most attention to text because it is the mode that more directly presents information, and it is therefore more accessible to learners (Rapport and Ashkenazi, 2008). However, research by Mayer (2001) showed that when information is presented via text only students have a hard time remembering what they read; transferring knowledge from text only sources, especially into problem solving situations, was also difficult. On the other hand, when multimedia presentations are used, learners performed better on problem solving (Mayer, 2001). Other research has found that when learners were presented with only text it was more difficult for them to comprehend material compared to when the text was paired with a representation (Carney and Levin, 2002; Gyselinck et al., 2008). Indeed, communicating scientific information to learners through multiple representations is known to enhance conceptual understanding (Kozma, 2003; Seufert, 2003; Ainsworth, 2006). Multiple representations help organize information, improve comprehension, and support the remembering of information (Carney and Levin, 2002; Butcher, 2006; Homer and Plass, 2010). Combining multiple representations from different formats has been found to be beneficial to learners (Mayer, 1997; Kintsch, 2004; Sadoski and Paivio, 2007; Schnotz, 2008).

Thus, the process of integrating representations with text or other representations is part of an active learning process (Mayer, 1999, 2001). Text is initially processed in the visual subsystem of working memory, the same place where visual representations are processed. Text and representations therefore compete for the learner's attention, decreasing the number of representations and text elements that one can pay attention to at a given time (Mayer et al., 2001). In addition, the mere presence of a representation in a textbook does not mean that it will necessarily communicate and facilitate the intended conceptual understanding of content (Gkitzia et al., 2011). In light of the research that highlights both the benefits and potential pitfalls of representations, this study has two aims: (1) analyzing features of representations in general chemistry textbooks (their nature and characteristics), and (2) interpreting the features through the lens of the cognitive load theory. The study is guided by the following research questions:

1. What is the average number of representations per page?

2. What function do representations serve in General Chemistry textbooks?

3. To what extent are representations physically integrated with the running text?

4. To what extent are representations and mathematical equations indexed within the running text?

5. (a) What proportion of representations have captions?

(b) What proportion of representations have labels?

6. (a) To what extent do representations require conceptual integration with the text and each other?

(b) On average, how many ‘interacting groups’ of representations are on a page?

These questions are significant because, besides text, representations are the most prominent feature in general chemistry textbooks. It is therefore beneficial to examine features of representations in order to get a sense of their utility as part of chemistry curriculum. Multiple features characterize representations in textbooks, such as instructional guidance, function, and spatial contiguity (Slough et al., 2010; Gkitzia et al., 2011). It is important to understand these various features, and how they aid in understanding of associated content, since these characteristics should be significant criteria in choosing and assigning textbooks as resources for students. While studies have been conducted looking at representations in general chemistry textbooks at the high school level (e.g.Gkitzia et al., 2011), this study examines representations in textbooks at the collegiate level. We use the cognitive load theory to draw implications of the features of representations in the five textbooks.

Theoretical framework

Representations in chemistry textbooks

Textbooks are an important resource in exposing students to representations and are also central to the teaching and learning of chemistry (Chiappetta and Fillman, 2007; Gkitzia et al., 2011). For instructors, textbooks guide both the organization of curriculum materials (Koppal and Caldwell, 2004) and assessment. Since representations play a central role in the teaching and learning of chemistry, representations used in the books should facilitate students' understanding of chemistry.

Representations are a prominent feature of modern day science textbooks. As Lee (2010) notes, “representations have become one of the most pervasive and visible elements of the modern-day science textbook” (p. 1099). When properly designed, representations can aid understanding of scientific ideas. However, their prevalence in textbooks has been a source of criticism, with some believing that they are in excess (Woodward, 1992). Representations can also cause confusion to students; to a point of being counterproductive to the very learning they are supposed to enhance (Linn and Hsi, 2000; Stern and Roseman, 2004). For this very reason, it is important that representations used in textbooks be reviewed to ensure they are relevant and will contribute to student learning.

Representations in textbooks serve a number of functions. Representations can be decorative, ‘representational’, organizational or interpretational (Slough et al., 2010). A decorative representation is not explicitly conceptually related to the associated text, and therefore does not support the text; an example could be a colorful molecule on the side of the page with no explanation as to how that molecule is relevant to the text or concept. A representational illustration (representation) complements text in communicating a concept – makes the information in the text concrete (Slough et al., 2010). For example, a three dimensional molecule where bond angles were highlighted could help a student visualize molecular shapes. Organizational representations summarize information into a form such as a table, map, or a graph. An interpretational representation goes beyond an organizational representation (Slough et al., 2010) by including additional information that may not be in associated text. Interpretational representations may include contextual details or useful information that would otherwise hinder the “flow” of the text, such a table of thermodynamic constants or experimental data.

Cognitive load theory

The main ‘idea’ of the cognitive load theory is that working memory has limited capacity, and is only able to process a few items at a time (Sweller, 1994; Mayer, 1999; Cook, 2006). Cognitive Load Theory provides a foundation and rationale for designing instructional materials in ways that reduce the load associated with content or information, and therefore enhance learning (Paas et al., 2004; Cook, 2006). One implication of the cognitive load theory for instructional design is that learning will be hindered if instructional materials overwhelm a learner's cognitive resources, since new information is received and processed in working memory, which has limited capacity (Kirschner, 2002). Instruction with multiple representations could burden a learner's working memory, depending on the nature of the material being presented, and how it is presented (Kirschner, 2002).

Three ‘types’ of cognitive load are associated with instructional material in textbooks: intrinsic cognitive load, germane cognitive load and extraneous cognitive load (Sweller, 1994; Cook, 2006). Intrinsic cognitive load is associated with the nature of the subject matter. For example, chemistry is multirepresentational, requiring learners to translate between and within representations that portray conceptual elements at the macroscopic and submicroscopic levels, sometimes simultaneously. This fact makes chemistry challenging to students (Johnstone, 2000). Representations and text require simultaneous processing and integration in order to facilitate understanding. If the number of interacting elements (elements that require integration to be understood) is high, the intrinsic cognitive load likely to be associated with them will be high, increasing the possibility of exceeding working memory capacity (Cook, 2006). For any given content area, intrinsic cognitive load is expected/constant since it is a function of the subject matter of that discipline (Sweller, 1994).

Germane cognitive load is experienced during the process of schema formation and automation (Cook, 2006). Schemas are ‘units of knowledge’ stored in long-term memory for later retrieval (Sweller, 1994). Facilitated by instructional design, germane cognitive load is beneficial to learning (Cierniak et al., 2009). According to Kirschner (2002), extraneous cognitive load is associated with poorly designed instructional materials, which require more cognitive resources to process. Resources devoted to process poorly designed instructional materials take away from meaningful learning (Kalyuga et al., 1999). Thus, extraneous cognitive load hinders learning (Cierniak et al., 2009). According to Cook (2006), both germane and extraneous cognitive load can be mitigated through instructional design. Ideally, the design of instructional materials should involve reducing extraneous cognitive load while increasing germane cognitive load (Cierniak et al. 2009). Of particular importance to this study is extraneous cognitive load, which results from poor instructional design (Kirschner, 2002).

Extraneous cognitive load and instructional design

Mayer (2003) noted that students learn best from multiple representations when extraneous material is eliminated from representations, which he called ‘the coherence effect’. The extraneous material is not relevant to target information. In a study on lightning, Mayer et al. (2001) found that students who were presented with interesting, but irrelevant, information performed poorly compared to those who were presented with concise information only. In essence, adding extraneous material hurt student understanding. Our classification of decorative representations in this study is parallel to what Mayer (2003) refers to as ‘seductive details’. Processing decorative and/or redundant material imposes extraneous cognitive load (Sweller, 1994). Redundant material that imposes additional cognitive load should be avoided in addition to decorative, irrelevant material (Cook, 2006).

The utility of multiple representations depends on the extent to which learners understand information contained in a given representation (Van der Meij, 2007). However, learners also have to translate between and within the different representations, a factor believed to enhance conceptual understanding and formation of a coherent mental model (Seufert, 2003; Schnotz, 2008; Berthold and Renkl, 2009). An important factor in integrating information in or from representations depends on the number of representations, and ultimately, the amount of information students need to integrate. In a recent study, Corradi et al. (2014) found that students with low prior knowledge struggled when presented with more than two representations alongside text that they were to integrate. If the number of representations that require simultaneous integration increases beyond what can be processed in working memory, learning will be hindered (Kalyuga et al., 2003; Cook, 2006). It is worth noting that this is more the case with low prior knowledge learners. For experts, the integration does not impose as high a cognitive load. To reduce this kind of cognitive load, it may be necessary to reduce the number of representations required to communicate a concept. This obviously raises the question of whether all representations used in textbooks (alongside text) are necessary. Corradi et al. (2012) found that when presented with multiple representations, students tended to focus on only one representation. We contend then, based on their results, that it may not be useful to present more/many representations to students which they may not pay attention to.

Despite the possibility of inducing a high cognitive load through the integration of multiple representations, representations that together communicate a given concept should be integrated instead of being presented separately. Multiple representations used in general chemistry textbooks require integration either between representations or with text in order to understand associated concepts. For this reason, the material is likely to impose intrinsic load (Sweller and Chandler, 1994). However, the representations and text cannot stand alone and need to be ‘used’ together. The process of integrating representations with text has the potential to impose high extraneous cognitive load if the integration is hindered by instructional design, especially for learners with low prior knowledge and limited working memory (Cook, 2006). This problem is exacerbated when a representation is distal or proximal to the referencing text. A learner's attention is split in the process of going back-and forth between text and representations, which may be on different pages. As Wu and Shah (2004) note, one way to reduce such cognitive load is to present text and accompanying representations close together, to help learners easily form associations between them. When designing instructional material, such as textbooks, efforts should be made to avoid ‘splitting learner's attention’, which could lead to extraneous cognitive load (Cook, 2006).

Instructional guidance for representations

One important factor in helping learners integrate representations with text is the instructional guidance they receive. This is especially important when using textbooks. Instructional guidance is necessary to help learners integrate text with representations in textbooks (Mayer, 2003). Research has shown that students with low prior knowledge are likely to benefit most from instructional guidance (Kalyuga et al., 2003; Mayer, 2003; Sweller, 2004). While guidance for novice learners will help them build schemas in long term memory, guidance for ‘experts’ will involve using already existing schemas (Cook, 2006). Different designs and levels of guidance will benefit learners differently. In the absence of instructional guidance, learners are not able to use multiple representations correctly (Van der Meij, 2007; Berthold and Renkl, 2009; Corradi et al., 2012), a fact which may be attributed to learner's low prior knowledge. Also, learners with low prior knowledge may not have built schema on which this knowledge on representations can build (Cook, 2006; de Vries et al., 2009). When presented with multiple representations, learners with low prior knowledge focus on surface features of these representations, at the expense of underlying concepts (Kozma, 2003; Rapport and Ashkenazi, 2008; Taber, 2009).

Captions and labels offer instructional guidance to support learning with representations by pointing out important features or information about a representation, thereby helping learners to focus on more than just surface features (Bodemer and Faust, 2006; Seufert and Brunken, 2006; Van der Meij, 2007; Gkitzia et al., 2011). Extended captions help students learn with graphics (Gkitzia et al., 2011). Research has shown that directives such as captions are more effective when they require active processing from a reader (the learner). With respect to captions, this means that they ‘engage’ learners by asking them to pay attention to specific aspects of a representation (Peeck, 1993). Also important are labels that accompany representations. Labels point to the different elements of a representation, thereby facilitating learning from the representations (Mayer and Gallini, 1990). While there is research indicating that students may not necessarily use captions as aids (Schnotz, 2008), we believe that the intention of having them accompany representations is so that they can help learners better make sense of those representations.

Indexing is another form of instructional guidance. Through indexing, text explicitly references a representation, using conventions such as “see figure x.x” (Slough et al., 2010). Readers are, as a result, able to relate representations to associated text. This is what directs a reader to text and representations that require integration. It is also common in general chemistry for mathematical equations to be indexed, usually by assigning the given equation a number and chapter. In most cases, this numbering system helps link the equation with associated text. In other cases, when authors use an equation in more than one page or chapter, this indexing helps locate the equation, or make associations between related concepts explicit.

Methodology

Textbook selection and sampling

Five textbooks currently in use at one of the researcher's institution were selected for analysis. Although this selection of textbooks was readily available and thus, a convenient sample (Cohen et al., 2011), four of the five textbooks that were used in this study are among the most commonly used General chemistry textbooks in the United States (Pub Track, 2012, as cited in Pyburn and Pazicni, 2014). This publication data substantiates that our chosen sample of texts represents a large proportion of the curricular materials that General Chemistry students use in the United States. The fifth book is the primary general chemistry text at one of researcher's institutions and was chosen because the researchers were curious as to its suitability (in the context of representations) as a resource for students. The number of textbooks chosen for analysis is consistent with those used in recent related analyses (e.g.Slough et al., 2010; Gkitzia et al., 2011; Pyburn and Pazicni, 2014). The five books chosen were: Chang and Goldsby (2013), Gilbert et al. (2015), Brown et al. (2015); Tro (2015), and Silberberg and Amateis (2015).

A random sample of textbook pages was obtained for each textbook through the use of a random number generator (http://www.random.org). A sufficient total number of pages were selected and analyzed from each text to provide a sample with a 95% confidence of being representative. (Cohen et al., 2011). Since each book was of a different length, this method allowed us to select an appropriate sample size for each book (i.e. larger books had larger sample sizes). The number of pages analyzed per book ranged from 216 to 235 pages. If one of the pages selected by the random number generator was a duplicate page, a new page that fell between the previous and following selected pages was randomly chosen. Because the focus of the study was not how representations were used in questions, pages that contained only review/end of chapter questions were not analyzed. Nearly all chapters in each book had at least one page that was analyzed, so representations were studied in a variety of chemistry topics in all of the texts. Due to the consistent nature in the layout of each textbook, there is a high likelihood that our randomly selected pages do constitute a representational sample that gives a comprehensive picture of representation characteristics for each text.

Coding and analysis procedures

We modified the Graphical Analysis Protocol (GAP) (Slough, et al., 2010) to analyze representations in the general chemistry textbooks. The GAP was used to analyze representations according to their form, level of systematicity, physical contiguity, indexical references, caption presence and type, and semantic integration (graphic function) (Slough, et al., 2010). In modifying the GAP for our study, we left out the ‘form’ category in the GAP which looked at type of representation (symbolic, sub-microscopic and text) since it is already known that chemistry is multirepresentational. We added a category of ‘labels’, as well as a category to measure the required conceptual integration between a single or multiple representations with text. We also re-wrote descriptions to make them more user friendly while still adhering to the intent of the original GAP protocol. Each representation was coded into each category as applicable. Two researchers coded all five textbooks together.

During the coding process, each page was analyzed separately; the number of representations on that page that fit in a given category were tallied and recorded. To establish inter-rater agreement, we coded common pages randomly picked from each of the five textbooks and compared our results. Our individual codes were tracked in Excel. The number of coding boxes which exhibited agreement were counted and converted into a percentage in order to establish inter-rater agreement which was 91% across the 5 textbooks. Differences were resolved through discussion. We then divided the remaining sampled pages into groups of 20 pages each. Each coder analyzed alternating groups of 20 pages (i.e. coder 1 analyzed the first set of 20 pages, the third set, fifth set, etc.). In this way, no textbook was coded solely by one researcher, and any differences in coding bias between the two researchers were distributed throughout the entire textbook. These procedures were repeated for every textbook used. With each textbook, the two researchers made sure that there was agreement of all codes in all categories through discussion. In the following section, categories used in the study are described, and illustrations for some of the categories where we felt it was necessary, are provided.

Categories

a. Number of representations. The total number of representations per page was counted and recorded in the rubric. The average number of representations per page was calculated by dividing the total number of representations by total number of pages. It is important to note that some chapters by their nature will have more representations than others. Also, while it is common to see pages as in Fig. 2 (below), it is common for two sub-topics to be presented on a page. Because of the variation, it is helpful to use the average number of representations per page.
b. Physical integration. This category analyzes the proximity of a representation to its associated text. A representation could fall into one of four categories:

Distal: placement requires you to turn a page to see the index/most relevant content text.

Facing: representation and index/most relevant content text are on facing pages.

Proximal: representation and the index/most relevant content text are on the same page, but separated by one half page or more of text.

Direct: the index/most relevant content text and representation are directly adjacent to each-other or are closer than half a page.

c. Figure indexing. This category looked at how representations were referenced in the main text, paying attention to where a representation is ‘first’ indexed. Representations were coded into three indexing classifications: no reference, a same page reference, or a reference on a different page from the representation. Usually, indexing is done using a “Figure x.x” or “Table x” designation in the text.
d. Extended captions. We coded “extended captions” (in sentence or long format) only in this category. One-word descriptors were counted as labels. Equations of reactions, or derivations that occur within text are not considered in this category. Representations either had or did not have captions.
e. Labeling (original GAP did not have this category). Labels are short identifiers of parts, processes, features, etc. These include graph axes, species identifiers, names of compounds or labels of formulae and column headers and titles in tables, among others. Labels were coded as either present or absent.
f. Representation function. This category analyzed the purpose of a representation. Each representation was coded into only one category, and could fall into one of four categories:

Representational: the representation is presenting content that is already in the text in a new way – adds concreteness. For example, an in-line symbolic representation adds concreteness to the process that is being described in words by presenting it in another format.

Decorative: the representation does not serve to meaningfully increase student understanding of the conceptual topic at hand. Deleting the representation would not negatively impact conceptual understanding.

Organizational: the representation adds coherence. Such a representation summarizes or organizes the content that was presented in the text. For example, such representations summarize concepts, link concepts together, or provide chapter layouts.

Interpretational: adds information that was not presented in the text thereby helping support more difficult or unfamiliar concepts. Usually adds an element of organization as well in that it is giving more context to the content being presented. Tables presenting data or constants for substances are considered to be interpretational.

Fig. 1 below shows a page taken from Brown et al. (2015) illustrating some of the categories described above:


image file: c5rp00140d-f1.tif
Fig. 1 Sample page showing representations. Brown T., Lemay E. H., Bursten B. E., Murphy C., Woodward P. and Stoltzfus M. E., Chemistry: The Central Science, 13th edn, ©2015. Electronically reproduced by permission of Pearson Education, Inc., New York, New York.
g. Conceptual integration (original GAP did not have this category). Representations are considered to be conceptually integrated with text if they require the student to look at information in a representation as they read text in order to understand a given concept. In this category, the number of representations requiring or not requiring conceptual integration on the page were counted. The group sizes of those which did require conceptual integration were recorded. For example, if a page had two representations that required conceptual integration with one another and text, the page would be recorded as having an integration group of three. We also found that on a given page, different representations may require to be integrated with each other and text. Here, text and an associated representation make one group. If a series of representations and associated text were related on a given page, we called them ‘interacting groups’ of representations. Fig. 2 below shows a page with ‘interacting groups’ of representations. On the page, 7 different but related representations are integrated with each other and text to bring about understanding of target content. Table 1 below shows sample coding using the page above.
image file: c5rp00140d-f2.tif
Fig. 2 Illustration of ‘interacting groups’ of representations. Brown T., Lemay E. H., Bursten B. E., Murphy C., Woodward P. and Stoltzfus M. E., Chemistry: The Central Science, 13th edn, ©2015. Electronically reproduced by permission of Pearson Education, Inc., New York, New York.
Table 1 Sample coding for Fig. 1
Category Code/representation
Number of representations 6 (Numbered 1–6).
Function Representational (eqn (3)–(6))
Interpretational (Table 15.1 and Figure 15.5)
Physical integration Representation 1 (Table 15.1 is ‘facing’).
All others are directly integrated
Labels Table 15.1 and Figure 15.5 (both require labels)
Indexing Table 15.1 is indexed on a different page
Figure 15.5 is indexed on the same page
Captions Figure 15.5 has a caption (Figure 15.5 is the only one needing a caption on the page. The label is sufficient for table 15.1)
Integration groups a. Representation 1 and associated text
b. Representation 2 and associated text
c. Representation 3 and 4 and associated text
d. Representation 5 and 6 and associated text
Interacting groups Representation 1 and text, with representation 2 and text make two interacting groups of representations as the two need to be integrated.
On the page, no cross-referencing is needed for the other groups-therefore the groups do not interact.

In Fig. 2 above representation/eqn (1) and text is one group, text and representations/eqn (2)–(5) is another group, representation/eqn (2) is integrated with both representation/eqn (3) and (4) (making two groups), representations/eqn (2)–(4) interact with eqn (5), and eqn (5) interacts with eqn (6). Representation 7 interacts with all of the six previous representations, and text, making an interacting group of 8 in total.

Results

Our goals of this study were two-fold: analyze features of representations in general chemistry textbooks and use cognitive load theory as a lens to make sense of the features in terms of how they help minimize or increase cognitive load, and therefore draw implications of these features on student learning from textbooks as instructional materials. In the following section, we will present and discuss results from our analyses, by research question. We should note that in Tables 3–10, the total number of representations (‘out of’) in each category varies. In some categories, only certain types of representations apply. For example, an equation will not be expected to have a caption, therefore equations will not be counted in the category.

1. What is the average number of representations per page?

Table 2 below gives a summary of the average number of representations per page. The average number of representations is obtained by dividing total number of symbolic representations in a given textbook by the total number of pages sampled.
Table 2 Average number of different types of representations per page
Book Average total number of representations per page
Chang and Goldsby 3.63
Gilbert et al. 3.80
Brown et al. 3.79
Tro 4.21
Silberberg and Amateis 4.31

All 5 textbooks had on average, about four representations per page. This means that in addition to text, there were four other representations that were associated with given text on a given page. While not always true, there is a very high likelihood that the representations are related to the same concept. As such, a student will need to integrate on average four representations with text while reading a page of each of the textbooks. While representations on a page pertain to one topic, it is possible that a page has two sub-topics. As an example, in Fig. 1 above, there are two related sub-topics on the page. In a recent study, Corradi et al. (2014) found that students with low prior knowledge struggled more when they were presented with two representations alongside text than when presented with text plus one representation. Students with low prior knowledge are therefore likely to struggle reading the textbooks we analyzed, given that the average total number of representations per page, alongside text is more than two. The high average number of representations per page could imply high intrinsic cognitive load (Sweller, 1994; Cook, 2006).

2. What function do representations serve in general chemistry texts?

Table 3 below summarizes the percentage of representations by function, for each of the five textbooks. The percentages for each category were obtained by the count of representations in each category, as a percentage of the total number of representations in a given book.
Table 3 Summary of the proportion of each representation function in the five textbooks
Book/function (%) Representational Decorative Interpretational Organizational
Chang and Goldsby 79 (592/753) 14 (108/753) 5 (35/753) 2 (18/753)
Gilbert et al. 86 (737/857) 9 (80/857) 2 (13/857) 3 (27/857)
Brown et al. 90 (772/857) 5 (45/857) 2 (20/857) 2 (20/857)
Tro 89 (842/948) 7 (68/948) 1 (11/948) 3 (27/948)
Silberberg and Amateis 85 (802/947) 8 (76/947) 4 (36/947) 3 (33/947)

Of the total number of representations analyzed, a majority served a ‘representational’ function (i.e. the representation was related to information that was discussed or described in the text), and some served organizational and interpretational functions. All of these help students understand associated concepts with explicitly provided links to the content and could be considered ‘instructionally useful’ representations. However, an interesting finding in this analysis is that across the five textbooks decorative representations are the second most prevalent class of representations. These representations were classified as ‘decorative’ since they did not have a conceptual connection to content on a given page. Decorative representations are extraneous material (Mayer, 2001, 2003), which add extraneous cognitive load, and are therefore likely to take away from cognitive resources devoted to learning (Cook, 2006). We should note here that any speculation on the role of representations we considered decorative is beyond the scope of this work. We are not saying that decorative representations are necessarily bad: we hope to highlight the implications of those representations on cognitive load and how they are therefore likely to impact learning from textbooks.

3. To what extent are representations physically integrated with the running text?

Table 4 below gives a summary of the physical integration (how close a representation is to associated text) of representations to associated text across the five textbooks.
Table 4 Proportion of representations' physical integration to text across the 5 textbooks
Book/physical integration % (occurrences) Distal Facing Proximal Direct
Chang and Goldsby 5 (35/747) 3 (31/747) 3 (19/747) 89 (662/747)
Gilbert et al. 7 (56/850) 9 (83/850) 4 (34/850) 80 (677/850)
Brown et al. 7 (60/856) 8 (67/856) 5 (44/856) 80 (685/856)
Tro 2 (22/934) 8 (70/934) 5 (44/934) 85 (798/934)
Silberberg and Amateis 3 (28/938) 9 (88/938) 3 (31/938) 84 (791/938)

Physical integration refers to the proximity of a representation to its related text and/or referencing index. Of the representations analyzed, a majority, at least 80% of representations in all textbooks were within one half a page of their associated text, i.e. directly integrated. However, the fact that there are representations that are distal (requires a turn of page between text and representation), or facing (text and representation are on facing pages), or proximal (text and representation are on the same page, but further than a half a page apart) is likely to pose challenges to students as they go between text and representations, especially as is the case with representations that are distal to text. As McTigue and Slough (2010) note, “students should not be expected to flip pages to find referent graphics, and most readers will simply not put forth such effort” (p. 223). This will of course mean that the desired conceptual integration between text and representations will not occur. We believe that a number of factors, such as the size of a given representation determine placement of representations. For example, a large table may not fit on the same page as associated text, necessitating a facing or distal placement. While the placement is justified, flipping a page as one goes between text and representation has implications for extraneous cognitive load. A closer examination of representations that were either distal or facing to associated text across the five textbooks indicated that only one-third were large tables. Representations which are spatially closer to their associated text present less (extraneous) cognitive load than if they were more spatially separated (Plass et al., 2009).

4. To what extent are representations indexed within the running text?

Table 5 below gives a summary of figure indexing for the five textbooks: whether they were indexed on the same page, different page or not indexed at all.
Table 5 Proportion of representations indexed on same page, different page or not at all
Book/indexing % (occurrences) Same page Different page Unindexed
Chang and Goldsby 44 (129/296) 17 (51/296) 39 (116/296)
Gilbert et al. 68 (223/328) 27 (89/328) 5 (16/328)
Brown et al. 67 (185/278) 29 (80/278) 5 (13/278)
Tro 66 (148/225) 8 (19/225) 26 (58/225)
Silberberg and Amateis 77 (252/329) 19 (61/329) 5 (16/329)

The percentage of representations indexed (using conventions such as Figure x.x or Table x) varied among different books, as shown in Table 5 below. All five textbooks had most representations indexed within the same page. This is a positive feature since indexing is a form of instructional guidance which directs students to view a representation associated with text. In each textbook however, there is a significant percentage of representations either indexed on a different page, or not indexed at all. Ideally, representations should be indexed, and be on the same page as the associated text. When indexed representations are on a different page, a student would have to flip to the particular page, and then back to text, a process that will likely lead to loss of concentration. Of concern is the high proportion of representations that are either indexed on a different page or not indexed at all in all five the textbooks (25–56%), as this assumes that students will figure out text and representation associations, which may not always be possible. This is an indication of poor design of instructional materials. Indexing can decrease cognitive load by cueing students when to look at a representation (Plass et al., 2009) and helping students to conceptually integrate material (Slough et al., 2010). Thus, unindexed representations add to extraneous cognitive load.

5(a). To what extent do representations have captions?

Table 6 below shows the proportions of representations that had a caption in the five textbooks.
Table 6 Proportion of representations with captions
Book/caption % (occurrences) Exists Does not exist
Chang and Goldsby 89 (263/296) 11 (33/296)
Gilbert et al. 96 (313/326) 4 (13/326)
Brown et al. 97 (269/278) 3 (9/278)
Tro 92 (206/225) 8 (19/225)
Silberberg and Amateis 93 (306/329) 7 (23/329)

As can be seen in Table 6, across the five textbooks, most of the analyzed representations had captions, ranging in proportion from 89% to 96%, which is an encouraging trend. On the other hand the absence of captions (between 4% and 11%) is likely to leave a student struggling to make sense of a representation. Captions provide instructional guidance (Pozzer and Roth, 2003; Gkitzia et al., 2011) which often helps to decrease extraneous cognitive load (Cook, 2006). Specifically, extended captions describe a given representation, thereby helping students understand information in that representation. An absence of a caption in a representation leads to extraneous cognitive load, as learners would have to figure out on their own what a representation shows, and they may not always succeed in this.

5(b). What proportion of representations had labels on them?

Table 7 below summarizes the percentage representations that had or were missing labels. We counted titles of tables, graphs, labels of axes of graphs, and labels of individual elements of representations in this category.
Table 7 Proportion of representations with labels on them
Book/labels % (occurrences) With labels No labels
Chang and Goldsby 68 (201/296) 32 (95/296)
Gilbert et al. 83 (272/326) 17 (54/326)
Brown et al. 90 (249/278) 10 (29/278)
Tro 84 (189/225) 16 (36/225)
Silberberg and Amateis 93 (305/329) 7 (24/329)

Labels, as a form of instructional guidance, play an important role in defining a representation, informing readers of what is in a representation. For example, a data table without a label leaves a reader wondering what information the table is portraying. A high percentage of representations analyzed across the five textbooks contained labels. However, as can be seen in Table 7, the proportions of representations without labels ranges from 7–32%, painting a less than ideal picture, especially in the case of the Chang and Goldsby textbook. Not every representation may require a label, but ideally, nothing should be left to the second-guessing of a reader as they process a representation. Across the five textbooks, lack of labels was most common in pictures, symbolic representations, and space filling models, where components were not labelled. Less common were missing labels in tables and graphs. For graphs, the most common missing label was the title, although to a small extent there were graphs missing labels of axes. For tables, the title was the most common missing label. As part of instructional guidance, necessary labels, such as titles should be included in representations. This level of guidance would certainly help decrease extraneous cognitive load (Cook, 2006).

An important observation we should make is that the same representations were coded for the caption and labels categories – a representation that had a caption likely also had a label. To a less extent, in each textbook, there were representations that had captions only or labels only. Table 8 below shows the relative percent occurrence of representations which had a caption and label(s), a caption only and a label only in each of the five textbooks.

Table 8 Proportion of representations that had both a caption and label, a caption only or a label only
Book/caption and label % (occurrences) Caption and label Caption only Label only
Chang and Goldsby 57 (168/296) 32 (95/296) 11 (33/296)
Gilbert et al. 79 (259/326) 17 (54/326) 4 (13/326)
Brown et al. 86 (240/278) 11 (29/278) 3 (9/278)
Tro 76 (170/225) 16 (36/225) 8 (19/225)
Silberberg and Amateis 86 (282/329) 7 (24/329) 7 (23/329)

6(a). To what extent do representations require conceptual integration with the text and/or each other?

Table 9 below details the relative percent occurrence of different sizes of groups of representations which required conceptual integration with each other. Text was included as a representation in these groups, so in the table, a group of two means one representation with/and text were needed to be conceptually integrated in order to reach understanding of a particular idea.
Table 9 Summary of percentage of different group sizes requiring conceptual integration
Book/group size occurrences % (occurrences) 2 3 4 ≥5
Chang and Goldsby 81 (390/481) 14 (67/481) 3 (16/481) 2 (8/481)
Gilbert et al. 65 (302/466) 19 (90/466) 10 (45/466) 6 (29/466)
Brown et al. 67 (339/507) 17 (86/507) 10 (53/507) 6 (29/507)
Tro 68 (363/534) 18 (97/534) 8 (42/534) 6 (32/534)
Silberberg and Amateis 76 (442/580) 13 (73/580) 7 (39/580) 4 (26/580)

A majority of the representations required the reader to conceptually integrate one or two representations with the text and/or another representation to achieve understanding. Across the five textbooks, there were instances where students were required to conceptually integrate more than four representations with text. In a study by Corradi et al. (2014), students struggled when they were required to integrate text with more than one representation simultaneously. Working memory can be overburdened if many elements of novel information require simultaneous processing, as is the case with many representations requiring integration with text (Sombatteera and Kalyuga, 2012). If material is new to learners (especially learners with low or no prior knowledge), the more representations one has to integrate with text, the more overload they are likely to experience due to intrinsic cognitive load.

6(b). On average, how many ‘interacting groups’ of representations were on a page?

On a given page, not all representations pertain to only one idea: out of all representations on a page, there may be a number of ‘chunks’ of representations and associated text requiring integration (which we refer to by interacting group here). As an example, on a page with three interacting groups, there would be three different ‘chunks’ of text and representations that go together. These groups may or may not be related. Table 10 below shows the average proportion of different numbers of ‘groups of integration’ per page across the five textbooks.
Table 10 Proportion of pages with different numbers of ‘interacting groups’ per page across the five textbooks
Text/groups per page % (occurrences) 1 2 3 4 5 ≥6
Chang and Goldsby 22 (43/193) 35 (68/193) 25 (48/193) 10 (19/193) 6 (11/193) 2 (4/193)
Gilbert et al. 23 (46/203) 44 (90/203) 20 (40/203) 8 (17/203) 4 (8/203) 1 (2/203)
Brown et al. 24 (53/219) 36 (79/219) 27 (59/219) 10 (22/219) 2 (5/219) 0 (1/219)
Tro 20 (42/205) 33 (67/205) 26 (53/205) 14 (29/205) 3 (7/205) 3 (7/205)
Silberberg and Amateis 20 (40/197) 26 (52/197) 20 (40/197) 18 (36/197) 6 (12/197) 9 (17/197)

Table 10 below shows that in most cases across the five textbooks between one and three integrated groups of representations are present on a given page. There is, however, a (albeit small) percentage of cases where there are up six or more groups of integrated representations on a given page. A close examination of cases where such high numbers of interacting groups were found showed that most were related to the same topic content. We feel that such a high number of groups makes for a ‘noisy’ page, which in itself could overwhelm a student. If the representations are all necessary for understanding associated concepts, four, five and six groups of representations would certainly make for high intrinsic cognitive load. It would be especially important in these cases where necessary to use aspects of instructional guidance (captions, labels, indexing, etc.) to assist students.

Discussion and conclusion

It could be enough to just research features of representations in chemistry textbooks (e.g.Gkitzia et al., 2011). Additionally, while there may be other avenues of looking at the implications of features of representations in general chemistry textbooks, we believe that cognitive load theory is a potential and useful lens of interpreting features of representations in textbooks. Our study adds to current knowledge on representations in general chemistry, specifically by looking at features of representations and the implications of the features on learning as seen through the lens of the cognitive load theory. In the chemistry education community, the role of representations has been accepted for decades (Talanquer, 2011; Taber, 2013). As part of chemistry curricula, it is important to consider factors that increase or decrease the utility of representations, which are intended to enhance learning and understanding of chemistry. In any chemistry textbook today, aside from text, representations are the most prominent feature. The representations therefore are important in determining the utility of textbooks in communicating chemistry content.

In the five textbooks used for this study, there were an average of four representations per page. This fact both confirms the multi-representational nature of chemistry, and the role that representations play in enhancing understanding of chemistry content by playing a supporting and complementary role to text in communicating content (Pozzer and Roth, 2003; Taber, 2013). Indeed, our analysis of the function of representations shows that most representations support conceptual understanding, since most fall into one of three categories (‘representational’, organizational, and interpretational). While these findings paint a positive picture of the use of representations in the textbooks, research has also shown that when students, especially those with low prior knowledge are presented with more than two representations that require integration with text, they get overwhelmed (Corradi et al., 2014). In the context of this study, such students are likely to experience high intrinsic cognitive load.

In this study, we found that textbook authors employed various design features and instructional guidance that we felt will help students as they read the textbooks, especially in the context of decreasing cognitive load associated with using representations in textbooks. Specifically, the use of labels, extended captions, indexing of representations within running text (on the same page), and physically integrating representations with text (direct or proximal) are features that will not only help enhance the integration of text and representations, but also decrease extraneous cognitive load associated with instructional design (Cook, 2006).

It is worth noting however that in each of the categories related to instructional design, there are features of representations in the five textbooks that indicate poor instructional design, which is likely to lead to extraneous cognitive load. Specifically, representations without labels, representations that were either not indexed or indexed on a different page, decorative representations, representations that did not have an extended caption, and representations that were facing or distal to text are all elements of poor instructional design, which is likely to add to extraneous cognitive load (Sweller, 1994; Cook, 2006).

In using representations in chemistry textbooks, the expectation is that they be integrated with text to bring about conceptual understanding. Our results show that in most cases, across the five textbooks, students need to simultaneously integrate between one and two representations with text to attain conceptual understanding. In the study by Corradi et al. (2014), students with low prior knowledge were successful at integrating one representation with text, but struggled when asked to integrate two representations with text. In each of the textbooks, though to a small extent, students need to integrate more than three representations with text. Even though students may not necessarily look at all representations, we believe the representations are necessary for understanding of target concepts. Especially for students with low or no prior knowledge, a high number of representations requiring conceptual integration with text will be challenging (Sombatteera and Kalyuga, 2012) and likely impose high intrinsic cognitive load.

Alongside integrating representations with text, students are sometimes required to integrate ‘groups’ of text and representations together for conceptual understanding. An analysis of cases where this was evident in the textbooks showed that this involved related content, where an idea was being developed through a combination of text and equations. As an example, in Fig. 2 above, taken from a chapter on thermochemistry, is an example on using enthalpies of formation to calculate enthalpies of reaction. This certainly is an example of a context where students will likely experience intrinsic cognitive load due to the nature of the content. Given this reality, it is encouraging that on that particular page, there is no extraneous material, and that the associated representations are directly integrated with text.

In conclusion, an analysis of the five textbooks shows that representations are used ‘extensively’ in each textbook given the average number of representations in each textbook. Our findings present a ‘mixed message’ with respect to features of representations that enhance or are likely to hinder understanding and learning when using textbooks as seen through the lens of the cognitive load theory. A general encouraging trend across the five textbooks is notable: a number of design features as well as instructional guidance are used across the five textbooks which we believe help decrease cognitive load experienced while using textbooks. However, in each of these categories, there are features of representations that indicate poor instructional design, which will add extraneous cognitive load and likely hinder learning (Cook, 2006). Chemistry is a multirepresentational discipline, which is also abstract in nature (Johnstone, 2000; Talanquer, 2011; Taber, 2013). By its very nature, the subject possesses intrinsic cognitive load (Mayer, 2001; Cook, 2006). The extent to which one experiences the intrinsic cognitive load depends on factors such as prior knowledge and instructional design (Cook, 2006; Slough et al., 2010). While intrinsic cognitive load will therefore be ‘expected’ in chemistry, extraneous cognitive load ‘amplifies’ its effect (Sweller, 1994; Cook, 2006). Since the extraneous cognitive load we anticipate from the analyses above in each textbook is mostly related to instructional design, there is a chance to reduce the extraneous cognitive load and consequently its impact on learning from the textbooks.

Implications

For chemistry instructors, our findings point to the need for scaffolding for our students as we teach with representations. As we present content, we should model how content is integrated with representations and use only those representations which are critical for understanding given content. Equally important is explicitly instructing students on how to ‘read’ representations. Our findings also present a model for assessing the suitability of textbooks that we recommend for our students to purchase; considering that most undergraduate chemistry courses recommend reading outside of class, it is essential that instructors evaluate current textbooks and are aware of and know how to look for potential drawbacks that a book may have in the realm of representations.

For textbook authors and publishers, our findings have implications for how representations, as used in textbooks, may or may not support learning. The textbooks we analyzed in this study are generally good models based on what we sought to study. In each textbook however, there are a number of elements of representations that could be improved. While one can appreciate the fact that a number of considerations go into for example placement of a figure in a textbook, one consideration has to be how easy a representation will be to use, based on how close it is to the accompanying text. For example, as Sweller (1994) notes, extraneous cognitive load, which can be mitigated through instructional design, should be an important consideration for designing instructional material. Any elements that could contribute to extraneous cognitive load should be avoided.

Our research and findings have implications for teacher preparation. One criticism that has specifically been labelled against teacher education programs is that they do not prepare science teachers in the area of curriculum selection, specifically selection of textbooks (Seufert, 2003). As noted by Cook (2006), representations, especially those considered decorative, have unfortunately been used for purposes of marketing. As science teachers choose curriculum materials, such as textbooks, they need criteria to go by. Our study, as well as others, such as Slough et al. (2010) provides a model of such criteria. For science teacher educators, we recommend that we engage in such an exercise to help pre-service teachers learn how to apply such criteria and make decisions on suitability of textbooks.

Limitations and future work

One limitation we feel owes to the nature of our study. We analyzed textbooks based on existing protocols and research. We used existing research, from cognitive load theory, to interpret our findings. While we strongly believe in the relevance of our findings, and the validity of the conclusions we drew, we believe it is important to conduct empirical studies to explore the issues we raise. For example, we note that design aspects such as the number of representations requiring integration have implications for cognitive load experienced. It will be important to measure the levels of cognitive load imposed when students attempt to interpret large groups of integrated representations (Mayer, 2005).

A second limitation stems from the fact that we sampled pages from each of the five textbooks and did not examine every page in the texts. We know that the number of representations for example vary depending on the chapter – so that some chapters will by their nature contain more representations than others. However, as noted in the methodology section, we believe the number of pages sampled from each textbook is high enough to make it representative.

There are a few potential projects we would like to pursue. First, since chemistry students are expected to be diverse, based on aspects such as prior knowledge, it will be useful to research how students on the prior knowledge continuum interact with text and representations (Cook, 2006). A recent study by Corradi et al. (2014) involved students with low prior knowledge. We are interested in using a population of students with high levels of prior knowledge as a comparison to the Corradi et al. (2014) study.

Another important study we wish to explore involves eye tracking to see how students translate between text and representations. We are interested in tracking a learner's ‘path’ as they read a page and integrate text and representations. In cases where multiple representations need to be integrated with text to attain conceptual understanding, we would like to determine whether learners integrate all representations that are necessary for conceptual understanding. Such an eye tracking study could also yield implications for the design layout of textbooks.

Acknowledgements

We are grateful to the National Science Foundation (NSF DUE-1156974) for funding Merry Gillaspie during the summer of 2014.

References

  1. Ainsworth S., (2006), DeFT: a conceptual framework for considering learning with multiple representations, Learn. Instr., 16, 183–198.
  2. Ainsworth S., (2008), The educational value of multiple representations when learning complex scientific concepts, in Gilbert J. K., Reiner ve M. and Nakhleh M. (ed.) Visualization: Theory and Practice in Science Education, Springer, pp. 191–208.
  3. Berthold K. and Renkl A., (2009), Instructional aids to support a conceptual understanding of multiple representations, J. Educ. Psychol., 101(1), 70–87.
  4. Bodemer D. and Faust U., (2006), External and mental referencing of multiple representations, Comput. Hum. Behav., 22(1), 27–42.
  5. Brown T. L., LeMay H. E., Bursten B. E., Murphy C. J., Woodward P. M. and Stoltzfus M. W., (2015), Chemistry: The Central Science, 13th edn, New York, NY: Pearson.
  6. Butcher K. R., (2006), Learning from text with diagrams: promoting mental model development and inference generation, J. Educ. Psychol., 98(1), 182–197.
  7. Carney R. N. and Levin J. R., (2002), Pictorial illustrations still improve students' learning from text, Educ. Psychol. Rev., 14(1), 5–26.
  8. Chang R. and Goldsby K. A., (2013), Chemistry, 11th edn, New York, NY: McGrawHill.
  9. Chiappetta E. L. and Fillman D. A., (2007), Analysis of five high school biology textbooks used in the United States for inclusion of the nature of science, Int. J. Sci. Educ., 29(15), 1847–1868.
  10. Cierniak G., Scheiter K. and Gerjets P., (2009), Explaining the split-attention effect: is the reduction of extraneous cognitive load accompanied by an increase in germane cognitive load? Comput. Hum. Behav., 25, 315–324.
  11. Cohen L., Manion L. and Morrison L., (2011), Research methods in education, 7th edn, London: Taylor & Francis, Inc.
  12. Cook M. P. (2006), Visual representations in science education: the influence of prior knowledge and cognitive load theory on instructional design principles, Sci. Educ., 90(6), 1073–1091.
  13. Corradi D., Elen J. and Clarebout G. (2012), Understanding and enhancing the use of multiple external representations in chemistry education, J. Sci. Educ. Technol., 21, 780–795.
  14. Corradi D. M. J., Elen J., Schraepen B. and Clarebout G., (2014), Understanding possibilities and limitations of abstract chemical representations for achieving conceptual understanding, Int. J. Sci. Educ., 36(5), 715–734.
  15. Dangur V., Avargil S., Peskin U. and Dori J. Y., (2014), Learning quantum chemistry via visual-conceptual approach: students' bidirectional textual and visual understanding, Chem. Educ. Res. Pract., 15, 297–310.
  16. Dori Y. J. and Hameiri M., (2003), Multidimensional analysis system for quantitative chemistry problems: symbol, macro, micro and process aspects, J. Res. Sci. Teach., 40(3), 278–302.
  17. Gilbert T., R., Kirss R. V., Foster N. and Davies G., (2015), Chemistry, 4th edn, New York, NY: W.W. Norton & Company.
  18. Gkitzia V., Salta K. and Tzougraki C., (2011), Development and application of suitable criteria for the evaluation of chemical representations in school textbooks, Chem. Educ. Res. Pract., 12(1), 5–14.
  19. Gyselinck V., Jamet E. and Dubois V., (2008), The role of working memory components in multimedia comprehension, Appl. Cognitive Psychol., 22, 353–374.
  20. Homer B. D. and Plass J. L., (2010), Expertise reversal for iconic representations in science visualizations, Inst. Sci., 38, 259–276.
  21. Johnstone A. H., (1991), Why is science difficult to learn? Things are seldom what they seem, J. Comput. Assist. Lear., 7, 75–83.
  22. Johnstone A. H., (2000), Teaching of chemistry: logical or psychological? Chem. Educ. Res. Pract., 1(1), 9–15.
  23. Kalyuga S., Chandler P. and Sweller J., (1999), Managing split-attention and redundancy in multimedia instruction, Appl. Cognitive Psych., 13(4), 351–371.
  24. Kalyuga S., Ayres P. L., Chandler P. A. and Sweller J., (2003), The expertise reversal effect, Educ. Psychol., 38(5), 23–31.
  25. Kintsch W., (2004), The construction–integration model of text comprehension and its implications for instruction, in Ruddell R. B. and Unrau N. J. (ed.), Theoretical models and processes of reading, Newark, NJ: IRA, pp. 1270–1328.
  26. Kirschner P. A., (2002), Cognitive load theory: implications of cognitive load theory on the design of learning, Learn. Instr., 12(1), 1–10.
  27. Koppal M. and Caldwell A., (2004), Meeting the challenge of science literacy: project 2061 efforts to improve science education, Cell Biol. Educ., 3, 28–30.
  28. Kozma R., (2003), The material features of multiple representations and their cognitive and social affordances for science understanding, Learn. Instr., 13(2), 205–226.
  29. Kozma R. B. and Russell J., (1997), Multimedia and understanding: expert and novice responses to different representations of chemical phenomena. J. Res. Sci. Teach., 34(9), 949–968.
  30. Lee R. V., (2010), Adaptations and continuities in the use and design of visual representations in US middle school science textbooks, Int. J. Sci. Educ., 32(8), 1099–1126.
  31. Linn M. C. and Hsi S., (2000), Computers, teachers, peers: science learning partners, Mahwah, NJ: Lawrence Erlbaum.
  32. Mayer R. E., (1997), Multimedia learning: are we asking the right questions? Educ. Psychol., 32(1), 1–19.
  33. Mayer R. E., (1999), Multimedia aids to problem-solving transfer, Int. J. Educ. Res., 31, 611–623.
  34. Mayer R. E., (2001), Multimedia learning, Cambridge/New York: Cambridge University Press.
  35. Mayer R. E., (2003), Elements of a science of e-learning, J. Educ. Comput. Res., 29(3), 297–313.
  36. Mayer R. E., (2005), Principles for Reducing Extraneous Processing in Multimedia Learning: Coherence, Signaling, Redundancy, Spatial Contiguity and Temporal Contiguity Principles, in Mayer R. E. (ed.) The Cambridge Handbook of Multimedia Learning, Cambridge: Cambridge University Press, pp. 182–200.
  37. Mayer R. E. and Gallini J. K., (1990), When is an illustration worth ten thousand words? J. Educ. Psychol., 82, 715–726.
  38. Mayer R. E., Heiser J. and Lonn S., (2001), Cognitive constraints on multimedia learning: when presenting more material results in less understanding, J. Educ. Psychol., 93(1), 187–198.
  39. McTigue E. M. and Slough S., (2010), Student-accessible science texts: elements of design, Read. Psychol., 31, 213–227.
  40. Paas F., Renkl A. and Sweller J., (2004), Cognitive load theory: instructional implications of the interaction between information structures and cognitive architecture, Inst. Sci., 32, 1–8.
  41. Peeck J., (1993), Increasing picture effects in learning from illustrated text, Learn. Instr., 3, 227–238.
  42. Plass J. L., Homer B. D. and Hayward E. O. (2009), Design factors for educationally effective animations and simulations, J. Comput. High. Educ., 21(1), 31–61.
  43. Pozzer L. L. and Roth W., (2003), Prevalence, function and structure of photographs in high school biology textbooks, J. Res. Sci. Teach., 40(10), 1089–1114.
  44. Pyburn D. T. and Pazicni S., (2014), Evaluation of the linguistic characteristics of general chemistry textbooks, Abstracts of Papers of the American Chemical Society, Amer. Chem. Soc., Vol. 247.
  45. Rapport L. T. and Ashkenazi G., (2008), Connecting levels of representation: emergent versus submergent perspective, Int. J. Sci. Educ., 30(12), 1585–1603.
  46. Sadoski M. and Paivio A., (2007), Toward a unified theory of reading, Sci. Stud. Read., 1, 337–356.
  47. Schnotz W., (2008), Why multimedia learning is not always helpful, in Rouet J., Lowe R. and Schnotz W. (ed.) Understanding multimedia documents, New York, NY: Springer, pp. 17–42.
  48. Seufert T., (2003), Supporting coherence formation in learning from multiple representations, Learn. Instr., 13, 227–237.
  49. Seufert T. and Brünken R., (2006), Cognitive load and the format of instructional aids for coherence formation, Appl. Cognitive Psych., 20, 321–331.
  50. Silberberg M. and Amateis M., (2015), Chemistry: The Molecular Nature of Matter and Change, 7th edn, New Jersey: McGrawHill.
  51. Slough S. W., McTigue E. M., Kim S. and Jennings S. K., (2010), Science textbooks' use of graphical representation: a descriptive analysis of four sixth grade science texts, Read. Psychol., 31, 301–325.
  52. Sombatteera, S. and Kalyuga, S., (2012), When Dual Sensory Mode with Limited Text Presentation Enhance Learning, Procedia - Social and Behavioral Sciences, 69, 2022–2026.
  53. Stern L. and Roseman J. E., (2004), Can middle-school science textbooks help students learn important ideas? Findings from Project 2061's curriculum evaluation study: life science, J. Res. Sci. Teach., 41(6), 538–568.
  54. Sweller J., (1994), Cognitive Load Theory, Learning Difficulty and Instructional Design, Learn. Instr., 4(4), 295–312.
  55. Sweller J., (2004), Instructional design consequences of an analogy between evolution by natural selection and human cognitive architecture. Instr. Sci., 32, 9–31.
  56. Sweller J. and Chandler P., (1994), Why some material is difficult to learn, Cognition Instruct., 12(3), 185–233.
  57. Taber K. S., (2009), Learning at the symbolic level, in Gilbert J. K. and Treagust D. (ed.), Multiple representations in chemistry education: models and modeling in science education, Dordrecht: Springer, pp. 75–105.
  58. Taber K. S., (2013), Revisiting the chemistry triplet: drawing upon the nature of chemical knowledge and the psychology of learning to inform chemistry education, Chem. Educ. Res. Pract., 14(2), 156–168.
  59. Talanquer V., (2011), Macro, submicro and symbolic: The many faces of the chemistry “triplet”, Int. J. Sci. Educ., 33(2), 179–195.
  60. Tro N., (2015), Chemistry: Structure and Properties, New Jersey: Pearson.
  61. Van der Meij J., (2007), Support for learning with multiple representations. Designing simulation-based learning environments (Doctoral dissertation). Retrieved from http://doc.utwente.nl on August 15th, 2014.
  62. de Vries E., Demetriadis S. and Ainsworth S., (2009), External representations for learning, in Balacheff N., Ludvigsen S., de Jong T., Lazonder A. and Barnes S. (ed.), Technology-enhanced learning, New York, NY: Springer, pp. 137–154.
  63. Woodward A., (1992), Do illustrations serve an instructional purpose in US textbooks? in Britton B. K., Woodward A. and Binkley M. (ed.), Learning from textbooks: theory and practice, Hillsdale, NJ: Lawrence Erlbaum, pp. 115–134.
  64. Wu H.-K. and Shah P., (2004), Exploring visuospatial thinking in chemistry learning, Sci. Educ., 88, 465–492.
This journal is © The Royal Society of Chemistry 2016