Effectiveness of field-based approach in science.

Abstract

A meta-analytic investigation was conducted to quantitatively synthesize and examine research findings about the effectiveness of field-based instruction on students' achievement in science. Effect sizes were calculated for 34 qualified studies. The overall effect was found to be meaningfully large. It was also revealed that the effect sizes were homogeneous, indicating a uniform positive effect of field-based instruction across the studies.

Introduction

A field trip is "a trip arranged by the school undertaken for educational purposes, in which students go to places where the materials of instruction may be observed and studied directly in their functional setting; for example, a trip to a factory, a city waterworks, a library, a museum" (Krepel & DuVall, 1981). The use of field trips in education has a rich history. Beidleman (1952) traced roots of field-based teaching back to Aristotle and Socrates. Many professionals in science education claim that the firsthand experience from a field trip enhance what is taught in classroom. The National Science Education Standards (NSES) also recommend the use of field-based instruction in science education (National Research Council [NRC], 1996). Actually, the latest program standards proposed a change in emphasis from "textbook- and lecture-driven curriculum to the curriculum that supports the Standards and includes a variety of components, such as laboratories emphasizing inquiry and field trips" (NRC, 1996, p.224). It is highlighted that experiences gained beyond the halls and walls of the school "can contribute greatly to the understanding of science" (NRC, 1996, p.45).

Although numerous studies have been conducted to explore cognitive or affective aspects of field-based instruction, as Orion and Hofstein (1991) points out, most of them utilize qualitative methods in evaluations and presently, no certain agreement exists among empirical findings whether field trip enhances students' achievement. Some of the studies emphasize that students' achievement was affected in a positive direction using field trips (Abad, 2003; Folkomer, 1981; Mackenzie & White, 1982) whereas other studies report no significant influence on students' achievement (Lloyd, 1965; McNamara, 1971). On the basis of the previous research findings, further research is obviously needed. This study uses a meta-analytic approach that assists revealing invaluable information regarding these discussions. Such empirical evidence is vital for science educators to illustrate whether there are potential effects of field-based instruction on student learning. This understanding motivates them to design better learning environments for the development of scientific individuals in the future.

Method

An effort was made to include all published and unpublished research studies since 1930 that examined the effectiveness of field-based instruction on science achievement. To identify studies, a search of on-line databases (Dissertation Abstracts International and ERIC) was conducted, using keywords "field trips in science education" and "achievement". Also, reference lists in the documents from the initial search were reviewed in an attempt to identify additional studies. As a result, 109 documents were identified. Studies included in this meta-analysis were limited to those written in English and reported in the United States. A study also had to provide quantitative information to compare a field trip group and a non-field-trip group. Studies without a non-field-trip group were excluded. Also, a study had to use some science achievement measure as a dependent variable. After considering these conditions, 34 studies were found to be qualified for this meta-analysis. For the 34 studies, various study characteristics were coded, including publication data, study conditions, instructional setting, types of field-based instruction, learner characteristics, outcome variables and quantification of outcomes. To establish reliability of the coding procedures, two individuals coded a sample of 10 studies. The agreement rate between the two coders was 90%.

When means and standard deviations were available for field-trip and non-field-trip groups, effect size (d-index) was calculated using a modified Glass unbiased estimator (Hedges & Olkin, 1985). If a study did not report the means and standard deviations, the d-index were computed from other statistics such as t and F statistics. In order to combine effect sizes, the weighted linear combination of estimated effect sizes (Hedges & Olkin, 1985) was used for this study. Since the experiments from 34 studies had different sample sizes, a weighting factor was also employed in order to improve the precision of the combined estimates. In this study, a fixed-effect model was employed, where it assumed that each experiment in the series was a replication of the others, differing only in response scale and sample sizes.

In order to determine whether the studies can reasonably be described as sharing a common effect size, the statistical test for homogeneity of effect sizes was conducted. If the null hypothesis for the test is not rejected, the studies are considered that they are random samples from a single effect size. Thus, a common effect size would be obtained using the information from all studies. On the other hand, if the null hypothesis for the test is rejected, we consider that effect sizes do not share a common effect size. The statistic, Q[T], referring to total class homogeneity, was used (Hedges and Olkin, 1985). To test for homogeneity of effect size for a study characteristic, Q[T] was partitioned into independent parts corresponding to within-class fit, Q[W], and between-class fit, Q[B], for each class of the study characteristic. The alpha level of significance was to be set at .05. Computations were performed by the "Meta-Analysis" computer program (Schwarzer, 1989) when it was possible.

Results

The 34 studies included in this meta-analysis provided a data set based on 6,837 students; where 3,716 students received field-based instruction and 3,121 did not receive field-based instruction. The 34 studies produced a total of 90 effect sizes because many studies included more than one effect size. If a study had more than one effect size, they were combined within each study by the weighted linear combination of effect sizes (e.g., Hedges & Olkin, 1985; Raudenbush, 1994). Some study characteristics, effect sizes, standard deviations of the distribution of the effects, and references for the 34 studies are summarized in the Appendix A and B available at http://gamet.acns, fsu.edu/~akamata/papers/AEQ.pdf.

The overall effect size based on the 34 studies was .25 with the standard error of .02. Assuming a standard normal curve, the result can be interpreted that the average student's academic gain would improve from the 50th percentile to approximately the 60th percentile by implementing field trips. Homogeneity statistics, Q[T], was equal to 32.17 (degrees of freedom was 33; p value was greater than .05). Therefore, it was concluded that the effect sizes were homogeneous.

Although it was concluded that the effect sizes share the same population effect size, some study characteristics were examined if they were related to the variations of the effect sizes. Three study characteristic variables were considered; academic areas (subject matter), students' grade levels (age groups), and the number of participants. The test of homogeneity was conducted for between-classes, Q[B], and within-classes, Q[W], for each of the three study characteristics. Also, unbiased estimate of effect size and homogeneity statistics were computed for each one of the classes for each study characteristic variable.

Three academic areas (sample sizes of life, earth, and general science classes were 15, 13, and 6, respectively) were examined in terms of homogeneity of effect sizes. The value of the within-class fit statistic, Q[W], was found as 28.32 (the degrees of freedom was 30; p value was greater than .05), indicating that there is no significant overall difference between effect sizes within any of the three academic areas. Similarly, the value of the between-class fit statistic, Q[B], was obtained as 3.85 (the degrees of freedom was 2; p value was greater than .05), indicating that no statistically significant difference was observed between the 3 academic areas in their effect sizes. This is an indication that students in all academic areas, general science (the effect size was .38), earth science (the effect size was .26), and life science (the effect size was .23), benefited equally from field-based instruction.

Five classes of age groups (sample sizes of elementary, middle, high, college, and adult classes were 9, 7, 11, 5, and 2, respectively) were examined. The value of the within-class fit statistic, Q[W], equaled 30.26 (the degrees of freedom was 28; p value was greater than .05) , indicating that there is no significant overall difference between effect sizes within any age groups. In the same vein, the value of the between-class fit statistic, Q[B], was found to be 1.91 (the degrees of freedom was 4; p value was greater than .05), indicating that there is also no significant difference in effect sizes between age-group classes. However, when unbiased estimates of the effect sizes for the five age-group classes were compared, some difference in magnitude of effect sizes were observed. Elementary school (the effect size was .25), middle school students (the effect size was .22), high school (the effect size was .28) and college (the effect size was .27) students benefited a lot from field-based instruction in science education. However, adults (the effect size was .04) did not benefit much in their achievement levels.

Four classes of the number of participants (sample sizes of "20-89", "90-159", "160-229", and "230 or more" classes were 8, 11, 7, and 8, respectively) were examined. The within-class fit statistic, Q[W], was obtained as 24.81 (the degrees of freedom was 29; p value was greater than .05), indicating that there is no significant difference in effect sizes within any sizes of participants. Similarly, the between-class fit statistic, Q[B], was found to be 7.36 (the degrees of freedom was 3, p was greater than .05), indicating that effect sizes were not different between classes. The effect sizes for the first three classes were similarly high (the effect sizes equaled .32 for 20-89, .27 for 90-159, and d = .37 for 160-229), while the effect size value (.20) was somewhat lower for the last class (230 or more).

Summary and Discussion

This study attempted a systematic synthesis of the study results for the effectiveness of field trip on science achievement. Overall, it was revealed that science achievement was enhanced when field-based instruction was implemented. Furthermore, three study characteristics were not related to the variation of the effect sizes. According to the findings, all students in all academic areas equally benefited from the field-based instructions. It is quite evident that first hand experiences provided from field trips improve students' achievement in any domain of science. As previously motioned, the NSES recommend that science teachers should effectively use a variety of hands-on/minds-on instructional activities for the discipline(s) and the level(s) they teach, both in the classroom and in the field (NRC, 1996). Standards also indicate that schools must rearrange schedules so that teachers can use field experiences to give students many opportunities to engage them in serious scientific investigation as an integral part of their science learning.

Results also revealed that the field-based instruction had the largest effect on the science achievement of the elementary, middle, high school, and college students, whereas the effects were smaller for adults. Actually, the effect of field trips corresponding to the improvement of adults' performance is not straightforward to interpret because only two studies were included in this category. More empirical data are needed to make realistic decisions on the effect of field-based instruction on science learning for advanced age groups. It was also found that the field-based instruction was equally effective for all classes of the number of participated students.

Based on the overall result of this study, we recommend inservice science teachers to implement field-based activities in their teaching practice regardless of the content area (e.g., biology, earth, or general science) and the grade (e.g., elementary, middle or high school), because field-based activities potentially improve students' science achievement. A recent study (Cox-Petersen, Marsh, Kisiel, & Melber, 2003) addressed the key factors of a good field-based instruction that effectively contribute students' science learning. They recommend that a field trip should encourage more active student participation, meaningful links between knowledge and real-world contexts, and sharing learning with peers and experts. In order for teachers to be skilled in the use of field-based instruction as a part of their repertoire of teaching techniques, science teacher education programs should emphasize the important role of authentic field-based activities for students' learning. If pre-service teachers have field-based experiences, they would be likely to design more interactive lessons and teach science concepts by utilizing field-based activities. It is hoped that field-based instruction would provide students with opportunities to reflect their understanding about science concepts in a practical way and provide them first hand experience in science.

References

Abad, A. A. (2003). Field trip preparation. Science Teacher, 70, 44-47.

Beidleman, R.G. (1952). The field trip: A technique in natural science teaching. School Science and Mathematics, 52, 105-118.

Cox-Petersen, A. M., Marsh, D. D., Kisiel, J., & Melber, L. M. (2003). Investigation of guided school tours, student learning, and science reform recommendations at a museum of natural history. Journal of Research in Science Teaching, 40, 200-218.

Folkomer, T. H. (1981). Comparison of three methods of teaching geology in junior high school. Journal of Geological Education, 29, 74-75.

Hedges, L. V., & Olkin, I. (1985). Statistical Methods for Meta-Analysis. San Diego, CA: Academic.

Krepel, W. J. & DuVall, C.R. (1981). Field Trips: A guide for planning and conducting educational experiences. Washington, DC: National Education Association.

Llyod, M. B. (1965). A study of comparison of two instructional methods: the field method and the classroom method involving science content in ecology for the seventh grade.

Science Education, 49, 453-468.

Mackenzie, A., & White, R. (1982). Fieldwork in geography and long-term memory structures. American Educational Research Journal, 19, 623-632.

McNamara, E. S. (1971). A comparison of the learning behaviors of eighth and ninth grade ESCP earth science students: one half experiencing laboratory investigations in the indoor environment, the other half-experiencing laboratory investigations in the outdoor environment. Unpublished doctoral dissertation, Pennsylvania State University.

National Research Council. (1996). National Science Education Standards. Washington, DC: National Academy Press.

Orion, N. & Hofstein, A. (1991). The measurement of students' attitudes towards scientific field trips. Science Education, 75, 513-523.

Raudenbush, S. W. (1994). Random effects models. In H. Cooper & L. V. Hedges (Eds.), The handbook of research synthesis (pp. 301-322). Russell Sage Foundation; New York.

Schwarzer, R. (1989). Statistics software for meta-analysis. [On-line available at http://userpage, fu-berlin.de/~health/meta_e.htm].

Genc, M.Sc is a Ph.D. candidate in Science Education and Kamata, Ph.D., is Associate Professor of Educational Measurement at the College of Education.