Design and Procedure

A within subjects post-test design with no control group was used. All subjects were given background information about analogies prior to analogy generation. This information included a definition of an analogy, its components, and examples of other analogies. Subjects were then given different paragraphs of scientific content and asked to individually generate analogies for each. There was no collaboration in the generation of analogies.

Selection of Concepts for Analogy Generation.

Science content has been organized in a number of ways, most frequently by discipline. Because of the range of concepts within each discipline, the researchers felt that a more specific categorization was needed to operationalize the types of concepts for which analogies would be generated. Categorizing science content into whether it involved structure or process seemed to be one possible way to study concrete and abstract concepts, without the constraint of scientific discipline. This dichotomy was similar to Lawson’s (1993) grouping of “descriptive” and “theoretical” concepts (p. 1213) and to a previous categorization of analogies. Thiele (1994) categorized analogies in another way, as structural or functional. Structural analogies were identified as those where the analog and target were structurally similar whereas functional analogies were defined as those where the analog reflected the behavior of the target (Thiele, 1994). Although the researchers in the current study were examining the effect of content on the quality of generated analogies, it appeared as if process or structure concepts might also affect the type of analogy generated. Therefore, the researchers decided to use a structure-process dichotomy for the current study.

Structure was defined as content focusing on static, definable features such as the layers of the Earth, parts of a plant, cloud types, or the arrangement of atoms in a molecule. Even when these features were not observable, they appeared to have observable counterparts, similar to Lawson’s (1993) “perceptible exemplars” (p. 1213). It was hypothesized that the observable aspects would make these concepts more concrete to learners. Process content was defined as a dynamic condition or a series of actions ending in a specific result, such as convection in the mantle, photosynthesis, or the water cycle. Many unobservable elements seemed to make these concepts more abstract and comparable to Lawson’s (1993) “theoretical” (p. 1213) categorization. For example, when discussing the convection currents in the mantle, it is not possible to observe the slow heating of rock, the melting of rock, or the movement of rock like a fluid.

Multiple process and structure concepts were generated for possible analogy generation and then organized according to scientific discipline. Two concepts found in elementary science texts, one in the domain of physical science and the other in earth science, were initially selected for analogy generation. The Earth’s layered interior was selected for the structure concept while heat conduction in solids was chosen for the process concept. Heat conduction in solids was also selected because of the role of particulate matter; a content area where misconceptions have previously been identified (American Association of Physics Teachers, 1996; DeVos & Verdonk, 1987; Gabel, Samuel, & Hunn, 1987). It was hypothesized that generated analogies and accompanying information about the similarities and differences between analog and target would reveal possible misconceptions.

The researchers decided to begin with one structure and one process concept in order to determine whether this dichotomy was specific enough to capture quantitative and qualitative differences in analogies generated for concrete and abstract concepts across scientific disciplines. Starting with the generation of two analogies would also enable an examination of possible assessment tools to determine whether quantifiable differences between analogies could be captured.

Evaluating Generated Analogies. A Validity Scale (Zook & Myer, 1994) was used to obtain an analogy quality score. Key aspects of each concept were determined prior to scoring. For example, for the Earth’s layered interior, the following were determined to be main points: layered interior (at least 3 layers evident), layers different thickness’ (ideally, outermost thinnest), layers composed of different materials (ideally, outer layer brittle, rock-like), and each successive layer surrounded by others. To receive the maximum number of points (5), all key aspects had to be evident in the analogy. Table 2 illustrates the scoring of one analogy using the previously identified key aspects.

Table 2

Critical aspects used to evaluate heat conduction analogies included both general and specific components. There were key aspects related to heat conduction such as evidence of energy application and the initiation of particle movement. There were also key aspects related to the differences between a good and a poor conductor such as particles tightly packed in a rigid structure compared to particles structured in a random manner with more space between them.

Explanations about the analogies written by subjects were coded according to whether similarities and differences between source and target concepts were provided and whether the information was scientifically correct. Analogy limitations were additionally coded according to whether they were superficial, an obvious observation dealing with surface level similarities and differences. For example, one explanation listed as superficial involved the difference between a layered candy (“Gobstopper”) and the Earth’s layered interior. In this case, the subject wrote, “One is much smaller than the other, and one is food.”

It was more difficult for subjects to generate analogies for the process than the structure concept. Over one-fourth (27%) did not generate an analogy for heat conduction while only 5% failed to do so for the Earth’s layered interior. When the absence of analogies was further examined, it revealed that 43% of the “no analogy” responses for heat conduction were examples of good or poor conductors rather than analogies while 14% were ideas for activities or demonstrations. One respondent noted, “You could demonstrate this very easily by heating the metal and showing how hot it gets and same with wood – just keep it from burning. I don’t think you need an analogy.” None of the “no analogy” responses for the Earth’s interior were examples or activities.

The majority of written explanations about the similarities and differences between source and target concepts were coded as superficial. Examples of explanations coded in this way included, “The Earth’s interior is not made up of food,” “You can eat a sandwich, you can’t eat the Earth,” and “Particles in metals are smaller than cars.”

Analog domains for analogies generated for the Earth’s interior structure were primarily food. Source domains for heat conduction analogies included both indoor and outdoor games, game equipment, experiences on highways, and comparisons between cities and towns. Tables 3 and 4 provide a representative sample of analogies generated for each concept.

Table 3

Table 4

Table 5 illustrates how one of the representative analogies for heat conduction was scored.

Table 5

Scored Analogy: A good conductor is like a ball pit with a lot of balls while a poor conductor is like an empty ball pit.

Analogy Validity

Analogies generated for the Earth’s layered interior (structure) and heat conduction (process) were evaluated using the adapted Validity Scale (Zook & Myer, 1994). Subjects generated significantly more valid analogies for structure than process content, Z = -6.15, p = .00. The mean validity score for generated analogies was higher for the Earth’s interior than heat conduction, 3.53 and 1.26, respectively. Only one pair of conduction analogies received the maximum score. Table 6 provides descriptive statistics for each general content area.

Table 6

Previous research found a dissociation between structure and process in the understanding of tectonic processes (Nottis, 1996). In this study, there was some dissociation found between structure and process in the analogies generated for heat conduction. Although the topic of heat conduction was labeled a “process” concept, key aspects identified for scoring the validity of these analogies included both structural and process components. Two structural aspects dealt with solids being composed of a group of small particles and the compactness of particles varying, depending upon whether something was designated a good or a poor conductor. Generated analogies were most likely to include these structural aspects, even though they were unobservable, and least likely to include process aspects, such as the application of energy (heat) and its transference from particle to particle. They could be categorized more as structural rather than functional analogies (Thiele, 1994). Finally, there was a significant, weak positive correlation between the structure and process scores: r_s = .26, p < .05.

Inter-Rater Reliability

In addition to the researchers, an individual with a background in geoscience also rated each of the generated analogies using the Validity Scale (Zook & Myer, 1994). Inter-rater reliability was then computed. There was a significant, strong positive correlation between the researchers and the outside rater’s scores for structure content, r = .87, p = .00. However, this was not maintained for scoring of process content. In this case, scores were only moderately, positively correlated, r = .49, p < .05. An examination of scoring differences revealed that the outside rater tended to give lower scores than the researchers. After a discussion of more explicit “possible violation” and “violation” criteria, the outside rater blindly rated the generated analogies after a month time lapse. Scoring of process content yielded an increased positive correlation, r = .57, p < .001.

Detection of Possible Misconceptions

Generated analogies and an analysis of qualitative data focusing on similarities and differences between analog and target concepts revealed a number of misconceptions. These misconceptions were primarily about the particulate nature of matter during heat conduction. Four general misconceptions were found in multiple conduction analogies. Table 7 summarizes the frequency and percentage of occurrence of each.

Table 7

Conduction Analogies With Varying Misconceptions, n = 44*

* Some analogies showed more than one misconception.

The first misconception indicated that the difference between a good and a poor conductor was due to the presence or absence of particles in solids. A good conductor was composed of a lot of particles (e.g., a ball pit full of balls) while a poor conductor had no particles (e.g., empty ball pit). One related misconception, found in one analogy, dealt with particle size. In this analogy, a good conductor was composed of smaller particles than a poor conductor was (tennis balls versus basketballs). This particular analogy also had the tennis balls in a hot room while the basketballs were in a cold room.

Heat generated from within particles, rather than heat transference, was another misconception found in multiple analogies. For example, one analogy comparing people to particles noted how people get angry if they are bumped, thereby generating internal, “emotional heat.”

The third misconception related to the position of the particles in solids. Rather than having the particles of the solid vibrating in a fixed position, these analogies had particles moving from place to place. Both good and poor conductors had individuals (usually less in a poor conductor) running around a room (e.g., gym, classroom) or playing a game (e.g., football, tag).

The final misconception was that volume determined conductivity. In these analogies, good conductors had smaller volumes than poor conductors (e.g., a classroom versus a gymnasium). One subject even noted in the accompanying explanation that the way the source and target were similar was, “[V]olume varies in both.”

It could be hypothesized that elaboration about the similarities and differences between the analog and target could mediate possible misconceptions. However, an examination of subjects’ explanations showed a tendency to target obvious, superficial differences rather than substantive differences between the analog and target. For example, one generated analogy likened a good conductor to a full elevator while its companion compared a poor conductor to an empty elevator. When asked to explain how the source and target concepts were different, the subject only noted, “Elevators and conductors are different.” Another individual who generated analogies comparing good and poor conductors to a full and an empty ball pit respectively, noted that the difference between the source and target was, “You can actually touch the balls and demonstrate.” With one pair of analogies where volume-change was seen, the subject wrote that the difference between the analog and target was that, “Students are much larger than particles.” There was also one explanation where it was questionable whether the subject understood that humans could be classified as matter. When asked to indicate the difference between source and target, this subject wrote, “Particles are all pieces of matter, kids are not.”

It was easier for subjects to generate analogies for the layers of the Earth than heat conduction. These analogies primarily involved food, a trend previously found in analogies commonly used to convey the Earth’s layered interior (Nottis, 1999). Analogies that were generated for the Earth’s layered interior were also more valid than those for heat conduction were. There are a number of possible explanations for these findings. The first relates to content; relevant content knowledge and aspects of the content itself.

Previous research has found that the ability to generate a personal analogy to help understand an unfamiliar concept is mediated by the quality of the learner’s prior knowledge (Zook & Roller, 1996). Although subjects were given background content, the depth and quality of their prior knowledge may have varied for content areas. This, in turn, may have compromised the subjects’ ability to effectively use the information that was provided.

Misconceptions about the particulate level of matter have been detected in children (DeVos & Verdonk, 1987), pre-college students (American Association of Physics Teachers, 1996) and pre-service teachers (Gabel, Samuel, & Hunn, 1987). For example, children have been found to ascribe macroscopic qualities such as color and taste to particles (DeVos & Verdonk, 1987). The classification of matter can also be difficult to understand and misconceptions have been detected in children in that area as well (Stavy, 1991). It is reasonable to assume that ideas labeled as misconceptions may have also been a part of the prior knowledge of the current sample, especially since analogies generated for heat conduction appeared to contain misconceptions previously identified in the literature. These included particulate level matter having macroscopic properties and anthropomorphic qualities (e.g., particles moving “[B]ecause they are ‘uncomfortable’ from the heat,” or getting emotionally “hot” because they have been bumped). It may be significant that both institutions required, at most, two science courses, which may have resulted in subjects drawing upon middle and high school science knowledge.

The inability of subjects to articulate definitive differences between the source and target concepts suggests that in presenting an analogy to future students, the subjects might not discuss their limitations. This could lead to additional misconceptions. This lack of elaboration has important implications for using analogy teaching models which require the articulation of the limitations as part of the instructional process (e.g., Glynn, 1994).

In addition to prior knowledge issues, there is also the visualizability of the concepts. Certain content areas seem to be more easily visualized than others are. Although the layers of the Earth may not be directly visible, examples of layering can be found in topography, food, and everyday materials. It was because of commonplace examples such as these that two outside reviewers concurred with the researchers that “layers of the Earth,” although not observable, could be classified as “visualizable.” This visualizability seemed to make the Earth’s interior structure more concrete to pre-service teachers, possibly contributing to the ease with which analogies were generated.

Although the result of heat conduction can be felt, the process is not as easily visualized. The continuous vibration of unseen particles in matter, as well as their movement and subsequent bumping from an energy source appear to make conduction a more theoretical concept. These multiple, unobservable aspects, as well as the lack of everyday analogs, may have contributed to the difficulty pre-service teachers had understanding heat conduction, ultimately compromising their ability to generate effective analogies which showed the difference between good and poor conductors.

The second possible explanation for the disparity in scores between analogies generated for the layered structure of the Earth (structure concept) and those generated for heat conduction (process concept) relates to subjects’ level of thinking. More than content knowledge is needed to generate effective analogies. Bloom (1956) categorized cognitive objectives into six major levels of increasing complexity -- knowledge, comprehension, application, analysis, synthesis, and evaluation. In order to generate valid analogies, students must have the ability to generalize information learned for use in other situations (application), break down course content into component parts so that relationships can be determined and understood (analysis), see the parts and construct the whole (synthesis), and judge the value of the material to determine its usefulness for future applications (evaluation) (Jacobs & Chase, 1992). In the current study, it appeared as if subjects tended to use lower order thinking skills such as basic knowledge and comprehension to generate analogies. This may explain why analogies generated for more concrete structure content were completed more successfully than those generated for process content.

Another related explanation for the score disparity, and worthy of further examination, is the level of reflective judgement needed to generate valid analogies for abstract and/or complex concepts. King and Kitchener (1994) have identified three levels (seven stages) of reflective judgment. Using the researchers’ criteria, it would seem that a pre-reflective person could generate valid analogies for what has been designated as structure content whereas an individual would need to be at least quasi-reflective, with an understanding beyond concrete instances, to generate valid analogies for what has been labeled process content.

Problems with scoring the analogies could also account for differences found in scores. Inter-rater reliability was lower for scoring process than structure analogies. The current scale may need to be refined for use with more theoretical and/or complex concepts. It may be that the complexity of some of the concepts should be addressed in another kind of scale. Development of a group of scored sample analogies might also help. A related scoring issue ties to the body of scientific knowledge and whether all previously identified misconceptions would clearly be identified as such at all levels of science instruction. This may have been a factor in some of the misconceptions about particulate matter found in the conduction analogies.

The third identified misconception was that particles in solids move around a given space. An elementary level understanding of particles in solids notes, “[T]he particles are arranged in regular patterns, with each particle being able only to vibrate around a fixed position” (DeVos & Verdonk, 1996, p. 659). However, conduction in solids comes from two mechanisms, the vibration of a particle in a comparatively fixed position, taught at the elementary level, and the movement of free electrons. Metals are known to be better conductors due to having free (valence) electrons that move throughout the substance (Long, 1988). It is the free electrons that contribute the majority of their conductivity. Therefore, those who generated analogies where particles moved from place to place were not completely wrong at all levels of understanding but needed to properly identify the aspects analogous to electrons and atoms. However, it should be noted that in no analogy or accompanying explanation was the term “electron” used.

The issue related to particle movement suggests the need for an assessment of subjects’ prior knowledge, providing clearer background information where critical concepts are highlighted, and an alternative scoring rubric where analogies having particles vibrating in place and particles moving around receive maximum scores if electrons and atoms are correctly identified in accompanying explanations. A highly valid analogy might be redefined to provide for both mechanisms with recognition that, depending upon the conductor, one mechanism would be dominant over the other. An accompanying scoring manual with exemplars of highly valid heat conduction analogies might also eliminate some of the interpretation difficulties.

The use of analogies has been promoted in the professional literature (Thiele & Treagust, 1991). However, not all analogies are equally effective. Preliminary results from this study indicated that the type of concept appeared to affect analogy generation. Although analogies can help students understand nonvisible and theoretical concepts (Lawson, 1993), it was for heat conduction, categorized as a more abstract, process concept, that pre-service teachers had the most difficulty generating valid analogies.

Teaching abstract, intricate information about the particulate level of matter is difficult, even at the secondary level (DeVos & Verdonk, 1996). Therefore, it is important to further investigate the source of pre-service teachers’ difficulties. Future studies should determine whether their difficulties stemmed from a lack of prior knowledge, the persistence of alternative conceptions, the absence of accessible analogs, the evaluation tool used, or a need for higher-order thinking skills and higher levels of reflective judgement.

Results of this analysis have important implications for pre-service teacher education programs. First, it would seem that there is a need to provide pre-service teachers with adequate science content knowledge. Pre-service teachers need enough knowledge to generate valid analogies. In addition, science-related courses should consider examining some of the prior knowledge of pre-service teachers so that recurrent misconceptions can be addressed in those courses. Analogies should be presented as pedagogical tools requiring a structured, systematic presentation. When teaching about and modeling recommended instructional models, science content known to be susceptible to misconceptions should also be incorporated.

Second, it appears as if pre-service teachers may need instruction in higher-order and reflective thinking skills. Teacher preparation programs should consider the inclusion of such activities in the area of science. Prospective teachers need to know how to get more information from what they are learning and reading rather than memorizing facts. Both increased science knowledge and instruction in higher-order and reflective thinking skills would presumably prepare pre-service teachers for increased standards and expectations in school systems. This two pronged approach should also facilitate and improve the use of science analogies throughout a teacher’s career.

Ameh, C., & Gunstone, R. (1985). Teachers’ concepts in science. Research in Science Education, 15, 151-157.

American Association of Physics Teachers. (1996). Powerful ideas in physical science: Nature of matter instructor materials (Vol. 4). College Park, MD: American Association of Physics Teachers.

Anderson, R. D., & Mitchener, C. P. (1994). Research on science teacher education. In D.L. Gabel (Ed.), Handbook of research on science teaching and learning (pp. 3-44). NY: Macmillan.

Atwood, R. K., & Atwood, V. A. (1996). Preservice elementary teachers conceptions of the courses of seasons. Journal of Research in Science Teaching, 33, 553-563.

Beiswenger, R. E., Stepans, J. I., & McClurg, P. A. (1998). Developing science courses for prospective elementary teachers. Journal of College Science Teaching, XXVII(4), 253-257.

Bloom, B. S. (1956). Taxonomy of educational objectives: The classification of educational goals (Handbook I: Cognitive domain). NY: McKay.

Callister, J. Coplestone, L., Consuegra, G., Stroud, S., & Yasso, W., (1992). Earthquakes. Washington, DC: FEMA-159.

Curtis, R. V., & Reigeluth, C. M. (1984). The use of analogies in written text. Instructional Science, 13, 99-117.

Dagher, Z. R. (1995). Analysis of analogies used by science teachers. Journal of Research in Science Teaching, 32, 259-270.

DeVos, W., & Verdonk, A. H. (1987). A new road to reactions: 4. The substance and its molecules. Journal of Chemical Education, 64, 692-694.

DeVos, W., & Verdonk, A. H. (1996). The particulate nature of matter in science education and in science. Journal of Research in Science Teaching, 33, 657-664.

Gabel, D. L., Samuel, K. V., & Hunn, D. (1987). Understanding the particulate nature of matter. Journal of Chemical Education, 64(8), 695-697.

Gentner, D. (1988). Metaphor as structure-mapping: The relational shift. Child Development, 59, 47-59.

Goswami, U. (1992). Analogical reasoning in children. Hillsdale, NJ : Lawrence Erlbaum.

Glynn, S. M. (1994). Teaching science with analogies: A strategy for teachers and textbook authors (Reading Research Rep. No. 15). University of Georgia: National Reading Research Center.

Glynn, S. M., Britton, B. K., Semrud-Clikeman, M., & Muth, K. D. (1989). Analogical reasoning and problem solving in science textbooks. In J. A. Glover, R. R. Ronning, & C. R. Reynolds (Eds.), Handbook of creativity (pp. 383-398). NY: Plenum.

Harrison, A. G. (1995). Teaching analogies in science in a systematic way. Paper presented at the annual meeting of the National Science Teachers Association, Philadelphia, PA.

Harrison, A. G., & Treagust, D. F. (1994). The three states of matter are like students at school. Australian Science Teachers Journal, 40(2), 20-23.

Heywood, D., & Parker, J. (1997). Confronting the analogy: Primary teachers exploring the usefulness of analogies in the teaching and learning of electricity. International Journal of Science Education, 19, 869-885.

Jacobs, L. C., & Chase, C. I. (1992). Developing and using tests effectively: A guide for faculty. San Francisco: Jossey-Bass.

Jarman, R. (1996). Student teachers’ use of analogies in science instruction. International Journal of Science Education, 18, 869-880.

Lawson, A. E. (1993). The importance of analogy: A prelude to the special issue. Journal of Research in Science Teaching, 30, 1213-1214.

Lin, H., Shiau, B., & Lawrenz, F. (1996). The effectiveness of teaching science with pictorial analogies. Research in Science Education, 26, 495-511.

Long, D. D. (1988). The physics around you (2^nd ed.) Belmont, CA: Wadsworth.

McNamara, D. (1991). Subject knowledge and its application: Problems and possibilities for teacher educators. Journal of Education Teaching, 17(2), 113-128.

National Education Goals Panel. (1997). The national education goals report summary. Washington, DC: author.

Nottis, K. E. K. (1996). The effective use of analogies in earth science. Unpublished doctoral dissertation, University New York at Buffalo, Buffalo.

Nottis, K. E. K. (1999). Using analogies to teach plate-tectonics concepts. Journal of Geoscience Education, 47, 449-454.

Pittman, K. M. (1999). Student-generated analogies: Another way of knowing? Journal of Research in Science Teaching, 36, 1-22.

Schoon, K. J. (1995). The origin and extent of alternative conceptions in the earth and space sciences: A survey of pre-service elementary teachers. Journal of Elementary Science Education, 7(2), 27-46.

Shulman, L. S. (1986). Those who understand: Knowledge growth in teaching. Educational Researcher, 5(2), 4-14.

Silver, Burdett, & Ginn. (1991). Science Horizons (Teacher Ed. 4). USA: Silver, Burdett, & Ginn.

Stavy, R. (1991). Children’s ideas about matter. School Science and Mathematics, 91(6), 240-244.

Thiele, R. B. (1994). Teaching by analogy. Education in Chemistry, 31(1), 17-18.

Thiele, R. B., & Treagust, D. F. (1991). Using analogies in secondary chemistry education. The Australian Science Teachers Journal, 37(2), 10-14.

Thiele, R. B., & Treagust, D. F. (1994). An interpretive examination of high school chemistry teachers’ analogical explanations. Journal of Research in Science Teaching, 31, 227-242.

Thiele, R. B., & Treagust, D. F. (1995). Analogies in chemistry textbooks. International Journal of Science Education, 17, 783-795.

Venville, G. J., & Treagust, D. F. (1996). The role of analogies in promoting conceptual change in biology. Instructional Science, 24, 295-320.

Wandersee, J. H., Mintzes, J. J., & Novak, J. D. (1994). Research on alternative conceptions in science. In D. L. Gabel (Ed.), Handbook of research on science teaching and learning (pp. 177-210). NY: Macmillan.

Wong, E. D. (1993). Understanding the generative capacity of analogies as a tool for explanation. Journal of Research in Science Teaching, 30, 1259-1272.

Zook, K. B., & DiVesta, F. J. (1991). Instructional analogies and conceptual misrepresentations. Journal of Educational Psychology, 83, 246-252.

Zook, K. B., & Myer, G. A. (1994, April). Learner generation and interpretation of instructional analogies. Paper presented at the annual meeting of the American Educational Research Association, New Orleans, LA.

Zook, K. B., & Roller, T. (1996). Effects of age, expertise, and schema induction on the generation and interpretation of instructional analogies. Paper presented at annual meeting of the American Educational Research Association, New York, New York.

We would like to thank Dr. James W. Baish, Professor of Mechanical Engineering, Bucknell University, who provided feedback on the misconceptions found in the conduction analogies and Gary N. Nottis, GN Geophysical, Lewisburg, PA who provided additional feedback on the manuscript. Correspondence concerning this article should be addressed to Katharyn E. K. Nottis, Department of Education, Bucknell University, Lewisburg, PA 17837. Electronic mail may be sent via Internet to knottis@bucknell.edu.

Back to the EJSE