Teaching the Control of Variables Strategy in Fourth Grade Classrooms Robert F. Lorch, Jr., William J. Calderhead, Emily E. Dunlap, Emily C. Hodell, Benjamin Dunham Freer, & Elizabeth P. Lorch University of Kentucky ABSTRACT We investigated an intervention (Chen & Klahr, 1999) that was designed to teach the “control of variables strategy” (CVS) to young students. Four questions were addressed by the study: (1) Can the teaching intervention be successfully translated to the classroom? (2) Is the intervention effective with students from lower-achieving schools, as well as higher-achieving schools? (3) Is direct instruction sufficient for teaching CVS in 4 th -grade classrooms? (4) Does hands-on experience conducting experiments contribute to learning CVS beyond the gains produced by direct instruction in CVS? Students from high- and low-achieving schools were taught CVS by direct instruction, or hands-on experimentation, or both. The results demonstrate that the basic intervention can be successfully translated to the classroom, but it is relatively less effective in lower-achieving classrooms. Further, direct instruction alone was effective in teaching CVS, but it was more effective when combined with hands-on manipulation. Hands-on manipulation unaccompanied by direct instruction was relatively ineffective. INTRODUCTION The experimental method is based on the logic that in order to demonstrate a causal relationship between two variables, the experimenter must manipulate the independent variable and find corresponding variation in the dependent variable while holding constant all other variables. Chen and Klahr (1999) developed a brief instructional intervention that they showed to be very effective at teaching CVS to 4 th -grade students. Klahr and his associates (Klahr & Nigram, 2004; Toth, Klahr & Chen, 2000) have successfully applied the teaching intervention in the classroom, albeit on a limited basis. The current study scales up the intervention by both applying the intervention in many 4 th -grade classrooms (i.e., 36) and by testing its effectiveness in two, very distinct environments (i.e., high-achieving classrooms vs. lower-achieving classrooms). In addition to examining whether Chen and Klahr’s teaching intervention can be scaled up, the study addressed basic theoretical questions about the conditions necessary for successful teaching of CVS. The teaching intervention is comprised of both direct instruction in the logic of CVS, and a “discovery” component involving hands-on manipulation of the experimental apparatus. Klahr and associates have already demonstrated that relatively little learning results if students conduct hands-on manipulation in the absence of any direct instruction. Based on their results, they have argued that direct instruction is a necessary component of teaching of CVS. However, they have not demonstrated that direct instruction is sufficient for learning of CVS, nor have they shown that the hands-on component of their intervention produces no benefits beyond the gains of direct instruction. Our study addresses four questions concerning the basic teaching intervention first developed by Chen and Klahr: 1.Can we successfully translate the basic teaching intervention into 4 th- grade classrooms on a relatively large scale (i.e., more than 1 or 2 classrooms)? 2.Will the intervention be successful in both high- and low-achieving classrooms? 3.Direct instruction may be necessary for teaching CVS, but is it sufficient? 4.Does hands-on experience in designing and conducting experiments contribute to learning CVS beyond the gains found with direct instruction alone? RESULTS Figure 1 displays mean performance on the paper + pencil tests of CVS understanding. The data are averaged across high- and low-achieving schools. Several patterns can be seen in Figure 1: Performance is at chance levels (50%) on the pretest Significant gains are found in all three instructional conditions on the posttest. Gains are ordered: Both > Instruct > Manipulate Gains at posttest are maintained 4-5 months later at the delayed test Figures 2 and 3 present the effects of instructional condition across the three tests broken down separately for higher-achieving and lower-achieving schools. METHOD Participants Participants included in analyses were 543 children from 36, 4 th grade classrooms. Students were from high achieving (n = 269) and low achieving schools (n = 274) in the Fayette County school district, Lexington, KY. Design The experimental design included three factors: 1.Students attended schools that were either high or low scoring on a statewide test of science achievement (Kentucky Core Content Test). 2.Classrooms were assigned at random to one of three instructional interventions: (a) CVS instruction (Instruct), (b) hands-on manipulation (Manipulate), or (c) CVS instruction + hands-on manipulation (Both). 3.Students were tested at three points: (a) the day before instruction (i.e., Pretest), (b) the day after instruction (Posttest), and (c) 4-5 months after instruction (Delayed Test). Materials Procedure Day 1  All students completed the paper + pencil test to obtain a baseline measure of their knowledge of CVS.  Students in the Manipulate and Both conditions worked in groups of 3 to design and run experiments to test the effects of individual variables on the distance the ball rolled. Day 2  Students in the Instruct and Both conditions received the teaching protocol, which provided confounded “bad” experiments and corrected “good” experiments, with explanations. Students in the Manipulate condition did not receive the teaching protocol.  Students in the Manipulate and Both conditions designed and ran experiments with the ramps to test the effects of each of the variables. Students in the Instruct condition did not do experiments with the ramps. Day 3 and Delayed  All students received the paper + pencil test on Day 3 and at a delay of 4- 5 months. The version of the paper + pencil test differed for each student across the three testing phases. Comparison of Figures 2 and 3 shows several similarities between the performances of students in the higher- vs. lower-achieving schools: Students in both environments improve from pretest to posttest Students in both environments maintain their gains at the delayed test. The pattern of relative performance as a function of instructional condition is the same for students in both environments. However, comparison of Figures 2 and 3 also shows some differences between the performances of students in the higher- vs. lower-achieving schools: The same instruction produced greater gains in the higher-achieving schools than in the lower-achieving schools. The Manipulate condition produced significant learning only in the higher-achieving schools. The addition of hands-on experience to the direct instruction produced significant gains relative to direct instruction alone for the higher-achieving schools, but not for the lower-achieving schools (compare Both vs. Instruct in each graph). CONCLUSIONS 1.The teaching intervention developed by Klahr and his associates was successfully translated into 4 th- grade classrooms on a relatively large scale. Instruction in the Both condition resulted in substantial improvement from the pretest to the posttest. Further, the gains were maintained several months later on the delayed test. 2.Students in both higher- and lower-achieving classrooms benefited from instruction in the two conditions involving direct instruction (i.e., Instruct and Both conditions). However, the gains were much larger in the higher-achieving classrooms than in the lower-achieving classrooms. There were indications that hands-on experience benefited students in higher-achieving classrooms, but had relatively little benefit for students in lower-achieving classrooms. 3.Direct instruction is sufficient for teaching CVS, as demonstrated by the fact that the gains in the Instruct condition were reliable and were maintained at the delayed test. 4.However, including hands-on experience in designing and constructing experiments results in better learning of CVS than direct instruction alone. This is demonstrated by the finding that the Both condition produced better learning than the Instruct condition. References Chen, Z., & Klahr, D. (1999). All other things being equal: Children’s acquisition of the control of variables strategy. Child Development, 70, Klahr, D., & Nigram, M. (2004). The equivalence of learning paths in early science instruction. Psychological Science, 15, Toth, E.E., Klahr, D., & Chen, Z. (2000). Bridging research and practice: A research-based classroom intervention for teaching experimentation skills to elementary school children. Cognition and Instruction, 18, Acknowledgment The research reported here was supported by the Institute of Education Sciences, U.S. Department of Education, through Grant 1 R305 H to the University of Kentucky. The opinions expressed are those of the authors and do not represent views of the Institute or the U.S. Department of Education.