Conceptests and Peer Instruction
RESEARCH
ConcepTests and Grading. ConcepTests are principally used to challenge students to confront the critical concepts that they will need to understand to be successful learners. This assessment process works best in a non-threatening, low stakes learning environment where the students are not penalized for getting the questions wrong. Consequently, it is recommended that ConcepTests are not used as the equivalent of quizzes or tests. For example, ConcepTests are often not graded but may be counted toward a participation grade. Students are aware that similar questions, sometimes the same questions, could be present on the exams. Students can earn up to 10% of the class grade on the basis of the number of days they answered ConcepTests (maximum points for 90+% participation), regardless of the accuracy of their answers. Students are given access to all ConcepTests used in class prior to exams.
 |
Graph of data from four Earth Science classes taught by the module author in Fall 2007 and Spring 2008 with similar general eduation student populations, in similar classrooms, using the same teaching strategies. |
I have used ConcepTests in his courses over a period of years. Anonymous student comments have been almost uniformly positive about the experience with this teaching strategy in Earth Science and Physical Geology classes. The average student success rate (the daily proportion of student correct answers averaged over the semester, see graph) on ConcepTests in the Earth Science class was between 50-70%. In other words, students correctly answered from half to two-thirds of the ConcepTests they attempted. Out of a population of more than 400 Earth Science students, no student averaged above 90% for the semester. The highest recorded student average score was 86% and the lowest score was 28%.
ConcepTests and Learning. The key learning principals behind the use of ConcepTests are that the instructor can use them to monitor student learning and that students will understand concepts better after discussing these concepts with their peers. A review of over 200 questions posed over multiple semesters reveals that most were answered one time but that 63 were asked twice. For those questions that were answered twice, an average of 45% of students responded correctly on the first attempt and 63% answered correctly after peer instruction. The degree of improvement increased steadily relative to the initial number of correct answers (see graph).
 |
Change in proportion of students obtaining correct answer on ConcepTests before and after peer instruction. Dashed line represents no change. Five ConcepTests had fewer correct answers after peer instruction. In all cases, less than half of the class initially answered these questions correctly. |
Ongoing Research. Beginning in Fall 2009 we will be comparing student responses on ConcepTests to specific exam questions. In particular, we are testing the following hypotheses:
- Students who change their answer from wrong to right after peer instruction will score better on related exam questions than students who get both ConcepTests incorrect. In other words, do these questions make a long-term difference in student learning or is it just a short-term effect? How do wrong-right students compare to right-right students on exam questions?
- The type of group interaction is more important than the ability of the students in the groups in leveraging collaborative learning. Group interactions that involve all group members speaking and listening will result in more students changing incorrect responses to correct answers after peer instruction than group interactions where students just share their answer choices with little or no discussion. How do low scoring students from different group interaction protocols perform on the same questions on the final exam?
If you are interested in working on this research, contact me at david_mcconnell@ncsu.edu.