C 2004) Journal of Science Education and Technology, Vol. 13, No. 3, September 2004 (


Instructional Technology and Molecular Visualization Jeffrey R. Appling1,2 and Lisa C. Peake1

The effect of intervening use of molecular visualization software was tested on 73 first-year general chemistry students. Pretests and posttests included both traditional multiple-choice questions and model-building activities. Overall students improved after working with the software, although students performed less well on the model-building portion of the evaluations. First semester and second semester students exhibit differences in abilities consistent with their different exposure to molecular geometry. KEY WORDS: multimedia; chemistry; model-building.

INTRODUCTION

Interactive software has the potential to increase the ability of chemistry students to visualize molecules in three dimensions. De Jong and Van Joolingen (1998) examined the use of computer simulations to teach science and proposed that there was potential advantage using a simulation-based curriculum to promote student understanding. However, just because a student uses interactive, multimedia science software does not mean that this software is effective at teaching students science. Frear and Hirschbuhl (1999) regard the need for research on student use of asynchronous computer-based learning as critical to progress in predicting student motivation and performance with computerized materials. It is presumed that students need to develop spatial reasoning ability in order to connect models to reality and perform successfully in chemistry (Brownlow et al., 2003; Haidar and Abraham, 1991; Pribyl and Bodner, 1987; Williamson and Abraham, 1995). Unfortunately, many software programs developed for student learning of chemistry topics have not included student use and feedback in their design cycles. The present study was initiated to attempt a connection between design goals and true student behaviors using computer-based materials created to teach the concept of molecular visualization. Wu et al. (2000) showed that the use of computer-generated molecular representations could improve performance for chemistry students learning at the high school level. It is presumed that the use of on-screen models helps students create

Molecular structure is considered a core concept in chemistry, and visualization of molecules and description of their shapes are critical skills for beginning chemistry students. Students are known to have difficulty translating between chemical formulas, molecular representations, and physical models (Keig and Rubba, 1993). The abstract nature of these tasks creates a barrier to connecting symbolic representations of molecules to a consistent visualization of reality at the molecular level (Ben-Zvi et al., 1987; Gabel et al., 1987; Hoffman and Laszlo, 1991; Wu et al., 2000). To bridge this gap, instructional strategies including stylized textbook art, model kits, and computerized three-dimensional graphics are used to various extents in most introductory courses. It is perceived that exposure to activities using computerbased instructional materials will help students attain the needed three-dimensional awareness necessary to interpret and describe molecular shapes (Ealy, 1999; Wu et al., 2000). This assumption was tested with a population of 73 college students enrolled in a first-year general chemistry course for science majors.

1 Department

of Chemistry, Clemson University, 223 Hunter Hall, Clemson, South Carolina 29634. 2 To whom correspondence should be addressed; e-mail: japplin@ clemson.edu

361 C 2004 Plenum Publishing Corporation 1059-0145/04/0900-0361/0


362 more effective mental images of chemical structures. Student conceptual understanding of chemical representations requires a great deal more research (Ben-Zvi et al., 1988), however there is significant evidence that three-dimensional representations of molecules using computer displays do promote three-dimensional skills (Ealy, 1999; Williamson and Abraham, 1995; Wu et al., 2000). A combination of computer images and physical models (ball-and-stick models, for example) has been shown to yield higher exam performance and better retention of microscopic visualization skills (Barnea and Dori, 1996; Copolo and Hounshell, 1995). The instructional technology instrument used in this study was the Discover Chemistry CD-ROM (J. Appling and D. Frank, Brooks-Cole Publishing, 1999). This software was designed as an interactive, learner-centered tutorial for general chemistry students. Many of the 75 activities involve elements of exploration and application, often with sophisticated feedbacks delivered to advance student self-study. The molecular visualization interaction activities allow students to match models of bonding to molecular shapes, demonstrating understanding of structural parameters such as bond angles, hybridizations, and dipole moments.

EXPERIMENTAL DESIGN Experiments took place over two semesters at Clemson University. A first phase of experiments was conducted during the spring semester, with a following set of replicating experiments in the following fall semester. Students in the study were recruited by their instructors and offered a small bonus for participation (dropping one additional class quiz). The instructor gave each student a ticket with a number that identified them as a participant and allowed the experimenter to verify their credit. Student identities were not known to the experimenter, conforming to approved IRB methods for the study. Students met the experimenter (L. P.) in a computer laboratory for approximately 75 min. Each student took a set of pretests to establish an entry point for their understanding of molecular shapes. Two tests were given, a traditional multiple-choice test and a test of model-building ability. These pretests were followed by 25 min of interaction with the Molecular Geometry module of Discover Chemistry. Students were asked to do as many examples as they could in the 25-min interaction period. Afterwards,

Appling and Peake students took a set of posttests, again including a traditional paper test and a test of model-building ability. The posttests had similar questions to those on the pretests, but none were identical. These tests were judged to be of equivalent difficulty. Typical molecular models have “atoms” with predrilled holes for attachment of sticks that represent bonds. A model kit would thus give each student an advantage by providing only a few templates from which choices could be made. To avoid this bias, model pieces were made from Styrofoam balls covered in clay. After each attempt the clay could be smoothed so that subsequent trials could be made without the student seeing attempts made by others. Large toothpicks were used as bonds in these models. In Phase 1, 41 students were recruited from both the first semester course (CH 101) and the second semester course (CH 102). These two populations should differ in their molecular visualization abilities, since the second semester students have already taken the first semester course where the topic is covered. First semester students participated in the study after the topic had been introduced in class, but before they had been formally examined on it. Study of the two populations was intended to reveal differences in how students might benefit from the instructional technology. For second semester students the interaction should be a review, whereas first semester students will experience the material as part of their initial exposure to the topic. A second phase of experiments was performed on 32 first semester students in the fall semester. This class is the “in sequence” group of incoming freshman students and has a different demographic than the “off sequence” class sampled in the spring semester experiments. It was felt that the students in Phase 2 were more representative of a typical CH 101 class, and the experimenter wanted to replicate the study.

RESULTS AND DISCUSSION Results from Phase 1 of the study are presented in Table I, where all population averages were analyzed using a matched sample t test (two-tailed). For the entire population, the increase in performance on the traditional multiple-choice problems was 13.9% (p < 0.001). Gains in performance are not reported as percentage of pretest value, rather they are absolute increases in the percent correct. For example, a

Instructional Technology and Molecular Visualization

363

Table I. Phase 1 Results Multiple-choice (out of 8)

Model-building (out of 7)

Population

n

Pretest

Posttest

Performance increase (%)

p (2-tailed)

Pretest

Posttest

Performance increase (%)

p (2-tailed)

All First semester Second semester Female Male

41 21 20 14 27

4.94 4.02 5.90 4.86 4.98

6.05 5.55 6.58 6.11 6.02

13.9 19.1 8.4 15.6 13.0

<0.001 0.003 0.009 0.03 0.002

3.63 3.43 3.85 3.96 3.46

4.30 3.86 4.78 4.25 4.33

9.6 6.1 13.3 4.1 12.4

0.003 0.3 <0.001 0.5 0.001

pretest of four correct out of eight (50%) followed by a posttest of six correct (75%) is a gain of 25%. Performance on the model-building portion of the tests showed a more modest increase of 9.6% (p = 0.003). This difference is to be expected, as building models requires higher order thinking skills. In both cases the null hypothesis is rejected, and the interaction with the instructional technology is shown to result in improved learning of the molecular geometry objectives. Multivariant ANOVA analysis of Phase 1 data did not reveal additional higher order interactions. Analysis of data for first (n = 21) and second (n = 20) semester students reveals differences between the two populations. For both multiple-choice and model-building tests the second semester group scored higher than first semester students on the pretests, which is consistent with a group that is reviewing known material. First semester students had a 19.1% performance gain (p = 0.003) on the multiple-choice section, whereas second semester students managed only an 8.4% gain (p = 0.009). This limited gain may be due to these students achieving a natural maximum in performance (82% correct) on the eight-question test. The intervention of the computer activity helped second semester students regain their visualization abilities. For the model-building test, second semester students improved by 13.3% (p < 0.001). This is in contrast to the first semester students, whose smaller gain of 6.1% was determined not to be significant at the 95% confidence level (p = 0.3). This indicates that the task of building physical models of molecules was too difficult for students that had just recently been exposed to the ideas for the first time. Their understanding of the actual geometric relationships of the atoms with molecules had not yet matured. Second semester students used the computer examples to reactivate their previously learned visualization skills. The significant gains by first semester students on the multiple-choice tests supports the use of the technology for learning by

this group, but the comparison of the manipulative performance for the two groups of students clearly indicates the level of understanding that can be expected for students new to the ideas. Additional study of model-building activities in the course could reveal potential for augmenting conventional twodimensional materials to improve this understanding. Phase 1 results were also sorted by gender. For multiple-choice tests, both groups performed similarly. Female students exhibited a performance gain of 15.6% (p = 0.03) and male students showed a gain of 13.0% (p = 0.002). Results for the model-building portion of the study were more mixed. Males had a performance increase similar to their multiple-choice increase, with a 12.4% gain (p = 0.001). The female students scored slightly higher on the model-building pretest yet their final performance was less than that for males, and their small 4.1% gain was not significant at the 95% confidence level (p = 0.5). Separation of the gender data by student semester reveals that the lower performance by women on the modelbuilding portion may be due primarily to the first semester female students. On the multiple-choice tests, first semester females (n = 7) outperform first semester males (n = 14): 29.5% gain (p = 0.02) vs. 13.8% gain (p = 0.03). However, the first semester males have a 11.3% performance gain (p = 0.04) on the model-building tasks, whereas female students have a 4.1% decrease in performance that is not significant at the 95% confidence level (p = 0.4). In contrast, the second semester females (n = 7) had a 12.2% gain (p = 0.02) on the model-building portion. Second semester males (n = 13) had a similar gain of 13.7% (p < 0.001). The lower performance by the first semester female students is difficult to fully describe, particularly since the number of female participants was small. Students in the first semester class taught in spring semester are an interesting mix due to the “off sequence” timing of the course. A number of the students are taking the course for a second time after a

364

Appling and Peake Table II. Phase 2 Results Multiple-choice (out of 8)

Model-building (out of 8)

Population

n

Pretest

Posttest

Performance increase (%)

p (2-tailed)

Pretest

Posttest

Performance increase (%)

p (2-tailed)

All Female Male

32 18 14

3.83 4.08 3.50

5.48 5.58 5.36

20.7 18.8 23.2

<0.001 <0.001 0.008

3.20 3.17 3.25

3.64 3.72 3.54

5.5 6.9 3.6

0.04 0.07 0.3

poor performance. This is one reason that it was decided to replicate the experiment in the following fall semester with the “on sequence” CH 101 students. Results for Phase 2 are presented in Table II. As in Phase 1, students show an increase in performance in the multiple-choice section, 20.7% overall (p < 0.001). The performance increase for the manipulative section is much more modest, 5.5% (p = 0.04). These results compare reasonably well to those accumulated in Phase 1 for the multiple-choice section. However, in Phase 1 the 6.1% performance increase for first semester students in the model-building section was judged not significant, whereas the same section in Phase 2 is seen to be significant. It was hoped that analysis by gender would reveal the discrepancy between the two groups. Performance increases for the gender groups in Phase 2 are different in various ways. Female students scored higher than males on the multiplechoice pretest, although their gains were marginally lower. Females improved by 18.8% (p < 0.001) whereas males improved by 23.2% (p = 0.008). Both groups clearly gained ability through interaction with the computer-based molecular geometry activities. There is a distinct difference, however, in the results for Phase 2 students in the model-building portion of the study. As is shown in Table II, performance increase for female students was 6.9% (p = 0.07), and that for male students was 3.6% (p = 0.3), neither of which was significant at the 95% confidence level. This result is different from the result obtained for the first semester students studied in Phase 1, where the performance increase for female students (4.1%) was observed to be insignificant and that for the males was 12.4% and significant (p = 0.001). The discrepancy between the model-building results for first semester students reveals the complexity of the measurement. The two populations are different: Phase 1 first semester students are “off sequence” and Phase 2 first semester students are “in sequence.” It is tempting to accept the Phase 2 data more readily due to the potential problems associated with investigating “off sequence” students.

More realistically, the confused results point to the need for a more in-depth approach to the issue of model-building by first semester students. This population is noted for difficulties associated with spatial ability and visualization of the particulate nature of matter (Harrison and Treagust, 2002). A more sophisticated study is required to better delineate factors that affect model-building capability derived from computer interaction for students at the introductory level.

CONCLUSIONS Students have varied responses to computerbased instruction in molecular geometry. Overall, it is seen that students respond positively to interaction with the molecular geometry activities of the Discover Chemistry CD-ROM. A short period of interaction is enough to raise performance levels of both first semester and second semester students. First semester students gain significant ability to answer typical multiple-choice questions aimed at determining understanding of geometric concerns including bond angles, hybridization of central atoms, and molecular dipole moments. These same students are less likely to extend this knowledge to the application of building accurate molecular models. Second semester students are able to reactivate their understanding of molecular geometry features after interaction with the computer-based materials. This result matches the assumption that these students should be only reviewing material that they have already conquered in a previous semester of study. Second semester students score higher than first semester students on the pretests, which is reassuring. Their performance gains for traditional multiple-choice assessments may be limited by their potential maximum performance, but their modelbuilding abilities seem to be superior to those of first semester students. Analysis of results with regard to gender is somewhat mixed. For both phases of the study,

Instructional Technology and Molecular Visualization both gender groups perform equally well on the multiple-choice measurements. In Phase 1, the study revealed that insignificant performance increases on the model-building portion of the study for female students may be due to the contribution of the first semester students in the study. Results of Phase 2 were not consistent, since both first semester female students and male students performed poorly on the model-building portion of the study. A more extensive study of model-building ability and computerbased instruction is required to better describe student behaviors. ACKNOWLEDGMENTS The authors thank Dr Craig Bowen for assistance with the ANOVA analysis. Partial support for this work was provided by the Clemson University Research Grant program. REFERENCES Barnea, N., and Dori, Y. J. (1996). Computerized molecular modeling as a tool to improve chemistry teaching. Journal of Chemical Information and Computer Science 36: 629–636. Ben-Zvi, R., Eylon, B., and Silberstein, J. (1987). Students’ visualization of a chemical reaction. Education in Chemistry July: 117–120. Ben-Zvi, R., Eylon, B., and Silberstein, J. (1988). Theories, principles, and laws. Education in Chemistry May: 89–92. Brownlow, S., McPherson, T. K., and Acks, C. N. (2003). Science background and spatial abilities in men and women. Journal of Science Education and Technology 12: 371–380. Copolo, C. F., and Hounshell, P. B. (1995). Using threedimensional models to teach molecular structures in high

365 school chemistry. Journal of Science Education and Technology 4: 295–305. De Jong, T., and Van Joolingen, W. (1998). Scientific discovery learning with computer simulations of conceptual domains. Review of Educational Research 68: 179–201. Ealy, J. B. (1999). A student evaluation of molecular modeling in first year college chemistry. Journal of Science Education and Technology 8: 309–321. Frear, V., and Hirschbuhl, J. (1999). Does interactive multimedia promote achievement and higher level thinking skills for today’s science students? Journal of Educational Technology 30: 323–329. Gabel, D. L., Samuel, K. V., and Hunn, D. (1987). Understanding the particulate nature of matter. Journal of Chemical Education 64: 695–697. Haidar, A. H., and Abraham, M. R. (1991). A comparison of applied and theoretical knowledge of concepts based on the particulate nature of matter. National Association for Research in Science Teaching 28: 919–938. Harrison, A. G., and Treagust, D. F. (2002). The particulate nature of matter: Challenges in understanding the submicroscopic world. In Gilbert, J. K., et al. (Eds.), Chemical Education: Towards Research-Based Practice, Kluwer Academic, Dordrecht, The Netherlands, pp. 189–212. Hoffman, R., and Laszlo, R. (1991). Representation in chemistry. Angewandte Chemie 30: 1–16. Keig, P. F., and Rubba, P. A. (1993). Translation of representations of the structure of matter and its relationship to reasoning, gender, spatial reasoning, and specific prior knowledge. Journal of Research in Science Teaching 30: 883– 903. Pribyl, J. R., and Bodner, G. M. (1987). Spatial ability and its role in organic chemistry: A study of four organic courses. Journal of Research in Science Education 24: 229–240. Williamson, V. M., and Abraham, M. R. (1995). The effects of computer animation on the particulate mental models of college chemistry students. Journal of Research in Science Teaching 32: 521–534. Wu, H.-K., Krajcik, J. S., and Soloway, E. (2000). Promoting conceptual understanding of chemical representations: Students’ use of a visualization tool in the classroom. Paper presented at the Annual Meeting of the National Association for Research in Science Teaching, New Orleans, LA, April 28 – May 1.