Using concept maps to reveal understanding: a two tier analysis

Using concept maps to reveal understanding: a two tier analysis

School Science Review, March, 2000, 81, (296) Pages, 41-46

Written by: Ian M. Kinchin

A combination of qualitative and quantitative analyses of concept maps can provide teachers with a powerful classroom tool

ABSTRACT Secondary pupils often exhibit very complex understanding of scientific concepts, frequently incorporating numerous misconceptions. Additionally, pupils fail to make links with related topic areas, though such links may be assumed within the teaching scheme. The construction of concept maps may help to make such ideas transparent so that they can be shared with others and provide a focus for reflection. Different methods of concept map analysis are described to illustrate different levels of conceptual development. Changes in concept maps may be helpful in illustrating conceptual change at both the gross and specific levels. The content of the map will also help to highlight the personal relevance of a topic to an individual pupil and so help the teacher to activate prior knowledge in preparation for further learning.

Concept maps have been said to provide a 'window into pupils' minds'. This might be rather overstating the case, but concept maps do seem to provide a valuable tool that can help in communicating the structure of complex ideas. They are effective aids for representing prepositional linkages within systems of related concepts. Teaching strategies using concept maps include those where maps are produced by the teacher for dissemination, and those based on pupil centered mapping activities. This dichotomy is significant and indicative of different teaching philosophies: lessons that use teacher-produced maps are generally expository and product- centered, while those that feature a pupil -concept- mapping approach tend to be process- orientated, focusing on the pupils' construction of meaning. Numerous studies reported in the literature indicate that the greatest benefit of concept mapping accrues to the author of the map. It is, therefore, pupil-produced maps that are the focus of this work.

Concept mapping can be introduced in a variety of ways to suit the level and ability of the pupils. Providing concept labels and/or prescribing the nature of the links to be used can assist pupils in producing maps that can be easily assessed for 'correctness'. However, Ghaye and Robinson (1989) interpreted maps that exhibited a close match with the teacher's knowledge structure as being indicative of a 'passive and reproductive' learning posture rather than a meaningful learning approach (as described by Novak and Gowin, 1984). Emphasis on correctness of maps would fit with a traditional philosophy of 'transmission teaching' and 'absorption learning'. The construction of a concept map is intended to reveal the perceptions of the map's author, reflecting his/her beliefs, experiences and biases, rather than the reproduction of memorized facts (Jonassen et al., 1997). If meaningful learning is to be encouraged through concept-mapping activities, then pupils require freedom in constructing their maps. The elements of a map need to come from the pupils if the map is to indicate the depth of their understanding. The more that is imposed by the teacher, the less intellectual ownership will be felt by the pupil. However, allowing pupils to devise their own descriptors means that they will display the depth of their understanding in an idiosyncratic manner. This will result in a greater diversity of maps which may be more problematic for teachers to interpret. Kinchin, Hay and Adams (2000) have described the value of such diversity and how it may be utilized as a teaching tool in the classroom through qualitative interpretation. This scheme is summarized in Box 1.

Box One

Structure	Spoke	Chain	Net
Hierarchy	single level	many levels, but often inappropriate	several justifiable levels
Additional	additions to centrial concept do not interfere with associated concepts	cannot cope with additions near the beginning of the sequence	additions/deletions may have varing influence as 'other routes' are often available through the map
Deletions	have no effect on overall structure	disrupt the sequence below the deletion	additions/deletions may have varing influence as 'other routes' are often available through the map
Represents	National Curriculum structure (England)	non-integrated Lesson sequence	Meaningful learning

Qualitative interpretation has received less attention in the literature than quantitative analysis. The tendency has been to grade maps so that the pupils could be given a final score that would summarize their performance. Most of the scoring protocols that have been reported are modifications of that suggested by Novak and Gowin (19 84). This considered the number of valid links presented, the degree of cross-linkage indicated, the amount of branching and the hierarchical structure. However, this aggregation of scoring elements creates a blurring of what the score actually reveals and can also be quite taxing on the scorer. Novak and Gowin were not convinced of the value of such scores themselves, stating:

Scoring was in many respects irrelevant, for we were looking for qualitative changes in the structure of children's concept maps. But because we live in a numbers-orientated society, most students and teachers want to score concept maps. (Novak and Gowin, 1984: 97)

The qualitative scheme described in Box I has the advantage of being quick and easy to use, providing teachers with a simple starting point for concept map analysis. It has been shown, however, that most writers visualize conceptual development (or knowledge restructuring) as occurring at two or more levels (Harrison, Grayson and Treagust, 1999). I suggest that recognition and consideration of these levels in the concept-mapping context will require separate methods of analysis. The qualitative 'spoke-chain-net' classification put forward by Kinchin et al. (2000) is able to describe gross changes in a concept map, indicative of radical restructuring, but is too coarse to pin-point details of weaker restructuring. Analysis of maps possessing similar gross structures (e.g. two spoke-style maps, as in Figure 1) shows that they differ in detail, particularly in the quality of the links used. In order to adequately describe these differences, a finer focus on the details of the map is required.

The quality of the links between concepts was analyzed by Ghaye and Robinson (1989). Those authors devised a classification for such links and arrived at seven kinds: structural, functional, locational, procedural, logical, composite and erroneous. These categories have been applied to the maps in Figure 1. Consideration of links in this way certainly adds to the richness of the description of a concept map and would provide the teacher with a tremendous insight into the depth of understanding displayed by his/her pupils. However, devising classifications of this kind (which will vary from topic to topic) makes substantial demands on a teacher's time and powers of analysis. In the absence of suitable published support materials, it seems unlikely that the adoption of such a strategy would be widespread. In the mean time, what is needed is a scheme that can be applied to any domain so that teachers may quickly gain familiarity with its use. It is most likely that a quantitative approach would satisfy this demand. In their comparison of various scoring protocols, McClure, Sonak and Suen (1999) have shown that the most reliable are probably those that place least strain on the working memory of the scorers. In particular, they found that analysis of the propositions within a map is least likely to tax the scorer. This focuses on the links in the map, the element over which pupils seem to have most difficulty, but which reveals so much about their depth of understanding. The scoring protocol they suggest is summarized in Figure 2 and has been applied to the maps in Figure 1.

Figure 1 Comparison of two concept maps with similar overall 'spoke' structures, but displaying differences in link quality. These maps are simplified composites drawn by the author based on elements taken from maps drawn by year 8 pupils. The numbers of propositions conforming to each of the seven types described by Ghaye and Robinson (1989) are given in the tables below each map, as are the scores achieved by each map using the prepositional scoring protocol (see Figure 2) of McClure et al. (1999).

Figure 2 Summary of a relational scoring protocol to assess proposition quality within a concept map. After McClure et al. (1999).

In combination, these two methods of analysis (qualitative analysis of gross structure and quantitative analysis of links) would seem to provide a valuable tool to highlight key characteristics of maps. This allows teachers to monitor both radical and weak restructuring of pupils' knowledge over time by observing changes in successive maps.

It has been emphasized by Black and Harlen (1993) that, in order to generate enthusiasm for learning, the content of teaching should have relevance to the learner. 'Relevant' should not be confused with I mundane', but should refer to ideas that may be related to components within pupils' existing conceptual frameworks, whether or not they represent part of their everyday experience. When viewed in this way, certain abstract concepts (such as Big Bang theory or the lifestyles of dinosaurs) may have relevance for particular pupils (Kinchin, 1993), and in such instances the abstract nature of the concept does not create a barrier to learning. Black and Harlen (1993) talk about relevance as seen by the pupil, and acknowledge that this will vary from one pupil to the next, according to past experiences and interests. The idea of 'relevant science', therefore, has to be viewed in a personal context and the nature of this 'personal relevance' can be located through concept mapping.

Personal relevance is not an 'all - or- nothing' phenomenon. A particular concept may be held by a number of pupils, but each will have it embedded in a unique array of associated concepts. For example, Figures 3 and 4 show two perspectives on the biology of flowers. Both maps suggest a detailed understanding of the central idea, but are presented in very different frameworks. The map in Figure 3 places strong emphasis on the anatomical structures found in a typical flower and the part each plays in reproduction. However, apart from the inclusion of 'hay fever', there is little connection to a wider view of flowers in biology. In contrast, the map in Figure 4 skims over the reproductive function and rather emphasizes a zoocentric view of biology typically exhibited by school children (e.g. Kinchin, 1999), fuelled by what Wandersee and Schussler (1999) have described as 'plant blindness'. Such a map suggests that adopting an 'Insect's view' of flowers may be a way of maintaining enthusiasm for the topic among pupils, by phrasing questions in the style, 'if you were a bee, which flowers would you visit most often, and why?' or 'which Parts of a flower would you be most interested in ?'This would increase the overlap between the topic area and the pupil's existing conceptual framework, thereby making the topic more relevant to the pupil.

Comparison of the two maps also highlights the point that pupils' selection of superordinate concepts within their maps is a major determining factor of the final structure of the map. In the map in Figure 4, the selection of 'Insects' as a key idea effectively limits the number of routes that can be taken from that point. This selection can also be hampered by a lack of appropriate vocabulary. For example, I observed a class of year 8 pupils constructing maps on 'breathing'. A group of pupils was struggling to explain 'breathing in' and 'breathing out'. They wanted to put this idea in a single box from which their other ideas could flow, but they did not know how. The introduction of the term 'ventilation' suddenly allowed them to organize their ideas lower down their map in the way that they wanted. This illustrates two important points about concept mapping:

Figure 3 A concept map of flower biology in which functions of reproductive organs are stressed.

Figure 4 A concept map of flower biology in which the role of insects is stressed.

Whilst concept maps are excellent vehicles for eliciting pupils' ideas, the teacher still has a vital role in guiding and supporting the activity. This is particularly important in the early stages of a map's development when the overall structure is being established.

Language is a vital tool the construction and organisation of ideas. Without appropriate vocabulary, pupils find it difficult to express their ideas in a logical manner. A concept map can highlight gaps in pupils' vocabulary.

There are numerous claims in the literature that concept mapping is a powerful tool that can have a significant, positive effect on learning in the sciences. However, my own observations suggest that science departments are not currently exploring the possible benefits. Part of the explanation for this under-use may be that teachers are not sure how to evaluate pupils' maps and record their key characteristics for later comparison. The mechanism described here is an attempt to address this problem. I am currently investigating possible patterns of development of pupils' map structures. In particular, I am interested in the relationships between development of gross structure and of link quality. It is not clear whether one triggers the other or whether they occur simultaneously. Greater insight into such issues may influence teaching strategies to promote meaningful learning and so inform curriculum development.