San Antonio, Texas
June 10, 2012
June 10, 2012
June 13, 2012
ASEE Board of Directors and Corporate Members Council
25.261.1 - 25.261.12
AC 2011-2920: IMPLEMENTING SOCIAL LEARNING STRATEGIES: TEAMTESTINGRebecca A Bates, Minnesota State University, MankatoAndrew Petersen, University of Toronto Mississauga c American Society for Engineering Education, 2011 Implementing Social Learning Strategies: Team TestingABSTRACTThis paper describes how to provide collaborative learning opportunities and fast feedback onexam performance by adding a team component to examinations. The method is supported byresearch in collaborative and active learning pedagogy and has been applied to computer sciencecourses ranging from first-year programming to graduate-level artificial intelligence. This paperrelates the use of team tests in two different university settings, with a range of implementations.Furthermore, it offers suggestions for customizing the technique to fit a specific classroomenvironment.1. INTRODUCTIONFinding the time and opportunity to incorporate active and collaborative learning in your classescan be challenging. Team testing is a collaborative learning activity with low implementationcosts and multiple advantages for both students and faculty. Along with the obvious benefit ofdeveloping team problem solving and discussion skills, students receive fast feedback on theirperformance, the instructor spends less time reviewing the exam (in class and with individuals),and the classroom environment benefits from the added value placed on collaboration andreciprocal learning. This paper describes a number of variations on the team testing idea anddiscusses how factors in the course affect the type of team test to develop.In a team test, the students complete an individual test paper as well as a group test paper. Theindividual component enforces individual accountability and allows the instructor to askquestions in formats that do not naturally benefit from a group discussion. The group componentasks the students to evaluate others’ ideas and to synthesize a solution that incorporates the bestideas generated by the members of the group. Due to the need to compare and defend ideas,questions on the group test naturally elicit and evaluate higher-level cognitive functions likeanalysis, evaluation and transfer.1 By doing so, team testing converts the evaluation environmentinto a learning environment. Depending on instructor goals, team tests can be structured toreinforce evaluation goals or learning environment goals.Conventional wisdom holds that test periods are lost instructional time, but we recall that in ourown experience as students, good exams often led to a more complete understanding of thematerial by forging relationships between domains and encouraging deeper insights. One studentwho participated in a team test reported, “You can collaborate and discuss what the generalconsensus of the answer is. … You [can] see why their answer might be more correct.” Byleveraging the motivation that an examination conveys, team testing encourages activediscussions between well-prepared individuals and fosters an ideal collaborative learningenvironment.In addition to evaluating and encouraging higher levels of proficiency, team testing reduces thelatency between evaluation and feedback. Exams give faculty an important opportunity toevaluate students, but they also serve to inform students of their progress and to point out areaswhere work is needed. Feedback should be timely to reduce exposure to content misperceptionsand reinforcement of correct understandings,2,3 but the time required to provide feedback tostudents is dependent on faculty time and the availability of grading support. The pressures onfaculty time are discussed at length at conferences and meetings, and institutional budgets dictatewhether grading support is available. In practice, this means that feedback is provided moreslowly than desired, but team testing provides immediate feedback by exposing students to theirpeers’ understanding of the material and testing their ability to contribute to the group’s solution.By providing students with more timely feedback, team testing reinforces foundationalknowledge that the remainder of the course requires.In the next section, we describe the baseline team testing implementation. Section 3 provides thepedagogical theory supporting the idea, and Section 4 presents our experiences using variants onthe baseline team testing implementation in the classroom.2. IMPLEMENTING TEAM TESTING2.1 Giving the ExamThe exam format depends on the ultimate goal of the group exam: evaluating content knowledgein a group context or creating a learning environment. To assess individual ability, an exam isgiven to each student. The individual exam is evaluated and typically forms the majoritycomponent of the student’s score. After the individual exam, there is a group component, whichcan either be a required part of the exam (evaluation goal) or considered a bonus (learningenvironment goal). Groups of 3 to 4 students collaborate on the group test. Larger groups reducethe impact of individual voices, and groups of two often suffer from a dominant (but notnecessarily stronger) partner. The group exam can be given in the same class period, if time isavailable, or in the following period. Providing a gap between the two exams can be beneficialbecause students can shore up weaknesses in their understanding brought to light by theindividual exam. However, not too much time should be allowed, since more misconceptions inunderstanding will be uncovered by the group exam than by the individual exam, and it isimportant to provide fast feedback.Before the exam, students must prepare sufficiently to be ready to actively engage during theexam. Without adequate preparation, students may find it too easy to passively rely on the groupconsciousness and may not be capable of identifying gaps in their own knowledge during thediscussion. To encourage individual accountability and active discussion during the exam, thestudents must follow two rules. 1. Each student must write some of the answers. 2. All students must agree on every answer submitted.When there is a “hung jury”, students can be encouraged to record the top two positions with asupporting argument for each. This becomes an answer the group can agree on.During the group exam, the instructor’s job is to facilitate discussion. As with any type ofclassroom problem solving, the instructor should move from group to group to keep students ontask, to refocus or raise questions when necessary. If the primary goal is to facilitate furtherlearning rather than evaluate current knowledge, instructors can provide hints or feedback abouttrains of thought. At the same time, the instructor can encourage individual accountability bydrawing out students not actively participating or by keeping time and calling for groups to selecta new student to record answers.2.2 Setting the QuestionsThe individual component and group component should test the same domain of material so thatthe students gain the benefit of fast feedback. The questions on the two components may be thesame or may differ in the level of proficiency exercised and evaluated. The type of questionsfeatured on a group component will differ based on the level of the students in the class and theinstructor’s learning objectives. The instructor may generate new questions for the group exam,but in many cases, asking the students to compare, analyze, or evaluate their answers from theindividual exam enables a deeper exploration of the question in the collaborative context.In early classes, where goals are often related to content knowledge and application within adefined setting and where students are less experienced learners, the questions may focus onfactual or procedural knowledge and might not differ from the individual exam. Presenting thesame problem on both exams prepares the students for discussion and encourages them to have astake in a specific answer. Students at this level benefit from comparing answers and debatingwhich is most correct.1For more advanced courses with mature learners, questions that focus on factual knowledge areless useful on the group exam as they do not foster in-depth discussion. The format favors open-ended problems that require students to evaluate and compare several possible solutions to reacha solution or that ask them to evaluate the context in which their knowledge is being applied.3. SUPPORTING IDEAS3.1 Theoretical FoundationsThe theoretical foundation for this teaching method stems first from the work of Vygotsky whodescribed a social constructivist framework for learning. In this type of setting, learning isenhanced by interaction with others beyond that which is possible individually.4,5 Buildingcommunity via discussion and collaborative learning can support the integration of academic andsocial experiences upon which student success has been shown to depend.6,7The ideas of active, collaborative, and reciprocal learning stem from Vygotsky's framework. Ingeneral, active and collaborative learning promote greater positive attitudes towards learning andthe subject matter.8 In reciprocal learning, students take turns teaching others. This is built intothe structure of this approach, which strengthens student understanding of the material for bothstrong and weak students. Guided instruction,9 where scaffolding built by the teacher is reduceduntil the student has responsibility for learning, also motivates this approach: exams are used as asource of learning and content discussion, with students building community through theirinteraction with each other. Along the way, they build communication skills and communicationabilities. In terms of team building, this approach has a low time-cost. By using the two simplerules presented in Section 2.1 to guide the process, we address the finding that students succeedbest in teamwork when there are specific instructions about listening, leadership, consensusbuilding and conflict resolution.10 ,11Even though collaborative learning has been supported by the National Science Foundation andrecommended by the education community, it is still not widely practiced in engineeringclassrooms.12 When implementing active learning, faculty sometimes find it difficult to give uptheir role as controllers of knowledge.13 However, giving an exam is an expected part of bothtraditional and active classrooms. As such, team testing may be a relatively comfortable step totake towards a more active classroom. Within computer science, team programming has beenshown to enhance learning and community (e.g., McDowell et al.14). Team testing can augment aclass using team programming or provide similar benefits when used alone.Feedback is generally considered important in learning and there is a growing body of work (andtechnology*) related to providing fast feedback (e.g., Chen et al.2 and Mehta3). Unfortunately, notmuch, if anything, has been published on the use of discussion as a fast feedback mechanism orthe return speed of graded exams and assignments. However, structured discussion with peers isa well-known method for increasing engagement and enhancing learning.15,16 Psychologyresearch has shown that reinforcement, through punishment and/or rewards, is important tolearning and that learning is easier when the reinforcement closely follows the action.17 Teamtesting provides the reward of rapid feedback in two ways. First, it provides students withsolution feedback and secondly, it provides feedback about their own performance and how itrelates to the performance of their peers.3.2 JustificationCooperative learning in a testing environment is beneficial because students develop groupinteraction and individual reasoning skills with the encouragement of a reward. The studentshave spent time considering the problems before discussion, which encourages them to becomemore involved. In fact, team-testing discussions are amongst the most lively and energetic in ourcourses, in part because of emotional reactions to learning answers but also because each studentis prepared for the discussion. The students are clearly connecting the content to a social learningenvironment, and as a result, they are better able to recall and reason about course material.18Many teachers spend a significant amount of class time going over the correct responses to anexam, or worry that they cannot take the time to do this without losing content coverage. Ratherthan spend the same amount of time going over the exam, a team test allows students to discover,justify and own the answers. By having students spend time working through the exam with eachother, the students become teachers, participating in reciprocal learning.For planning teacher-to-classroom feedback, the group exams give a better picture of whichconcepts were globally missed or were difficult. As with any exam, this can be useful feedback* The Web-based Interactive Science and Engineering Learning Tool at Oregon State University is an example of fast-feedback technology. (Koretsky, M. Web-based Interactive Science and Engineering Learning Tool. https://secure.engr.oregonstate.edu/che/WISE/. )about the teaching of the content, with even more weight because the results have been filteredthrough both individual minds and group consideration. When the teacher discusses the exam,the concepts missed after the group exam can be the focus, thus reducing faculty review time.3.3 BenefitsOne model of active learning separates "doing" from "observing" and differentiates betweendialog "with self" and "with others".8 Group tests contain room for both types of dialog. First, thesolo test forces to the student to hold a dialog with self, with the instructor as audience. Next, thegroup section asks the students to enter into a dialog with others -- potentially with problems theyhave already considered and prepared answers for. While this dialog enhances learning, it alsoenhances the sense of community and has been shown to improve student attitudes toward thediscipline and course.19 Rather than prolonging a competitive evaluation atmosphere, acollaborative environment is available for students who prefer collaboration over competition.In terms of feedback, after taking a team test, especially one identical to the individual exam,students leave the room knowing which of their answers are likely to be correct. One problem intest-taking is when smart students apply their own logic to problem solving but do notnecessarily have the grounding knowledge. The logic may be solid but the answer is not correct.Discussion and skeptical questioning leads the group closer to a correct answer. As one studentreported, “It was pretty productive because you could give an answer … but someone could givea rebuttal. … It would help you out in your understanding.”Comparing performance on team and individual tests has made it possible to give studentsfeedback not only on their content knowledge but also on their confidence in their knowledge.One student reported, “Having someone else’s word that something is correct feels like you aregiven more confidence in what is right and wrong.” Another was the only one of his group to getthe answer correct on the individual test but was unable to persuade his peers to change theirminds. After written and verbal encouragement, this student began to advocate his knowledgemore strongly. At other times, we have observed students who are prepared but not confident intheir knowledge be pleasantly surprised to find that they are at the same level, if not higher, thantheir peers.4. TEAM TESTING IN THE CLASSROOMTeam testing has been employed in a total of nine different courses at two separate institutions:Minnesota State University, Mankato (MSU) and the University of Toronto Mississauga (UTM).The classes spanned the curriculum, ranging from first year programming to third year operatingsystems to graduate level natural language processing, so they display a variety of sizes (5-50students), learners (novice to independent), and content (logical and algorithmic problem solvingas well as design of large systems).In this section, we introduce factors that affect the effectiveness of team testing and influencehow it should be deployed. We report anecdotal evidence from our own experiences with teamtesting, relating students concerns and feedback wherever possible. The discussion focuses on theconditions in the classes that caused us to alter the team testing implementation.4.1 Class SizeThe authors have successfully used team testing in classes of up to 50 students. However,colleagues have successfully employed various collaborative testing techniques in courses withmore than 500 students. The primary consideration is physical space, which should be plentifulenough to allow groups to interact comfortably. Because discussion can get lively and facultyinteraction is helpful, we have found these techniques to work especially well in classes from 20-30. On the other end of the spectrum, we have employed team testing in classes with as few as 5-6 students. In this case, a single discussion group works, even though there are more than thepreferred 3-4 students, because in very small classes, a group dynamic of discussion and problemsolving often exists already.4.2 Reward SystemsThe most important variable that affects how students perceive team testing is its impact on theirfinal grade. In its original form, we imagined team testing to be an opportunity to practice gooddiscussions and to convert an evaluation environment into a learning environment. Students in alearning environment should feel comfortable practicing new skills, which suggests that theevaluation weight should be low or nonexistent. At the same time, team tests work because thestudents come to the activity prepared and motivated, so we cannot simply eliminate the “test”component of team tests.The original versions of team testing at MSU made the team test voluntary and offered extracredit for participating. In our experience, offering just 5% bonus credit on the individual examcauses an overwhelming fraction of students to remain for the team-test. The bonus points offer atangible reward to the students, countering the stress that comes with exams and grades, whileproviding motivation to invest energy and preparation in a good outcome. As a result, studentsdid not appear to feel stressed by the addition of the group component of the exam, but they didactively engage in the exercise.In contrast, the academic culture at UTM made it difficult to offer bonus credit on an exam, sothe group component of the exam was weighted as a small fraction (10%) of the overall exammark. As a result, initial student reaction to the team test was quite different at UTM, withstudents expressing concern about being placed in a group with perceived “weaker” membersand feeling stress preparing for the group activity. Instead of viewing the team test as anopportunity to collaborate, the students reacted negatively to a personal evaluation being affectedby a factor outside of their control.The instructors also perceived noticeable differences between the two exam environments. Byoffering extra credit for the team test, the MSU instructor was able to treat the exercise as acollaborative learning activity. She circulated among the groups, asked questions, and redirectedgroups that strayed off track. The UTM instructor, however, felt uncomfortable intervening inindividual groups and felt that the exercise was an evaluation.Nevertheless, in both cases, discussion within the groups was vigorous, and after the experience,students were pleased with the group exam. Both sets of students viewed the group discussion asa means for improving their overall mark, and many students enjoyed the activity.4.3 MarkingThe reward system chosen also impacts how the team test is marked. When assessingcollaborative learning, the basic approach is to provide independent and interdependentassessment for group work.8 The independent component of the exam maintains individualresponsibility. If the reward is minimal and the focus is on learning, rather than evaluation, thenthe grades for group exams can be coarse, reducing grading time and ignoring small differencesbetween groups. If the focus is on evaluation, then more care needs to be spent on the marking toprovide formative feedback on the group’s effectiveness.In our experience, due to the low weight of the group component and the focus on learning ratherthan evaluation, grading on group exams can be generous, not requiring a 100% correct answerto get full marks, making the reward positive for the students while still allowing for fastfeedback. One student in particular was concerned that the other members of the group would notshare his desire for excellence: “I really have a hard time doing the group thing as I strive to getthe best grade and working in a group seldom allows for that opportunity.” A small rewardprovides motivation, and keeping the reward small enough to avoid further stress about theresults of the group exam reduces complaints and classroom tension if a student feels he or shehas been placed in a “bad” group.4.4 Exam CompositionTeam testing supports many different kinds of questions, and the relationship between questionson the individual exam varies depending on the instructor’s objectives. At both MSU and UTM,the instructors favored reusing questions from the independent component of the test. In ourexperience, relating the questions on the two exam components increases discussion since eachstudent has spent some time thinking about the question and formulating an answer to which theyhave some emotional attachment. MSU had some success reusing exactly the same questions,and they found that solutions to group exam questions were noticeably stronger than typicalanswers on the individual exam. UTM tended to ask follow-on questions and saw similar results.Re-using questions also relieves one concern students reported about team tests – that thematerial on the group component would be significantly more difficult.Regardless of the type of questions used, the group component should contain far less materialthan the individual component. When using a reduced set of problems rather than a full test, wefound that students would delve deeply into related issues, starting discussions about theory andcontext that are rare in a regular classroom. In an electric circuit theory class, team testing wasused on weekly quizzes. A short, two-problem quiz was given and then the same two problemswere given to student pairs. After the first week, the team quiz was reduced to one problem,because students spent so much time discussing the application of theory. As a result, they wereoften able to show better understanding during this phase than on their individual quiz.4.5 Instructor TimeTeam testing does create some additional work for the teacher: creating a second exam andgrading additional questions. This can be minimized by using the same (or at least related)questions on the group component and by coarsely grading the team work. Furthermore, markingtime on the individual component is reduced since less feedback needs to be provided. Since thestudents have discussed the questions, they often better understand how they earned their grade,so students ask fewer questions about marking in office hours and by email. Instead, somestudent groups used office hours to continue the group discussion with instructor adjudication.4.6 Exam TimingWhen giving the team tests, the class structure made a difference in the method and the timeallowed for team tests. When the class session is long enough (1.5 to 2 hours), immediatelyfollowing the individual exam with the group exam works well. When there are recitationsections or labs, a portion of that time can be used for the team test instead of using extra lecturetime. Giving a team test in the next class period, especially with time at the end for classroomdiscussion and questions also works, but it weakens the feedback link.Students who have become accustomed to both the feedback and reward of group exams haveasked for them at the final exam. In these cases, a slightly reduced exam time, e.g., from 120minutes to 90 minutes, allows time for a team test as well as comprehensive coverage of coursecontent. This gives students feedback they might not otherwise receive at the end of a term.We have not yet tried giving a group exam before the individual exam. We feel this wouldincrease exam preparation time since the group exam questions and individual exam questionswould need to be more decoupled for a fair individual assessment. We are also concerned thatdiscussion would be weaker, since the students would enter the group component with lesspreparation.4.7 Group FormationFor any type of team work, the selection of the teams and the roles assigned to the team membershas an impact on the team's success. The goal of group formation is to create effective groups,giving students a fair chance to do well while minimizing the problems that come with groupwork. We have used a wide variety of heuristics including student physical proximity (both closeand distant), splitting up friends/lab partners/long term groups, and coupling weaker and strongerstudents. With one notable exception, the groups functioned well regardless of their method ofcreation. In that case, a student acknowledged as being exceptionally strong inadvertentlydominated discussion, since his group was unprepared to challenge his ideas. Teacherintervention is required in this situation. The other students may be encouraged to enter thediscussion by asking them to justify why they agree with the proposed solution.At UTM, where the group component is a fraction of the test mark, some students requested thatthey be informed of group pairings before the exam to facilitate the formation of study groups.Despite some initial concern that releasing the pairings would result in the responsibility forstudying material being split up among members of the group, we found that the individual examled students to study all of the material. As a side benefit, some of the study groups that wereformed because of the team test continued to meet later in the term.We have not investigated assigning teams on a long-term basis but believe it would be aninteresting avenue for future study. This approach would reinforce the social learning aspect andallow for more flexibility in negotiation of roles within a team.5. SUMMARYTeam testing enhances student learning by providing faster feedback and using collaborativelearning to develop higher order cognitive skills. It leverages student desire to thoroughly preparefor evaluation to increase engagement in a focused small-group discussion. In doing so, teamtesting converts an evaluation environment into a learning environment.Team testing is a flexible idea which we have successfully implemented in many types of coursesfor a broad range of students, learning goals, and content foci. Since the exercise oftenrepurposes material already generated for the individual component of the exam and decreasestime spent reviewing exam solutions, we have found team testing to be a low-cost method forintroducing collaborative learning into the classroom. More importantly, students have respondedextremely positively – even going so far as to request a team test for the final exam!6. ACKNOWLEDGMENTSThanks to our students who were willing– and often enthusiastic! – participants in this process.7. REFERENCES1 Biggs, J.B. & Collis, K.F. 1982. Evaluating the Quality of Learning: The SOLO Taxonomy. New York AcademicPress.2 Chen, J.C., Whittinghill, D.C., and Kadlowec, J.A. 2006. Using rapid feedback to enhance student learning andsatisfaction. Frontiers in Education.3 Mehta, S.I. 1995. A Method for Instant Assessment and Active Learning. Journal of Engineering Education, 84:295.4 Vygotsky, L. 1978. Mind and society: The Development of Higher Mental Processes. Cambridge, MA: HarvardUniversity Press.5 Vygotsky, L. 1986. Thought and Language, revised American Edition, translated and edited by Alex Kozulin,Cambridge, MA: MIT Press.6 Treisman, U. 1992. Studying Students Studying Calculus: A Look at the Lives of Minority Mathematics Students inCollege. College Mathematics Journal 23(5), pp. 362-372.7 Wilson, D. 2008. Does it Matter? Psychological Sense of Community & Belonging in Engineering Education.Frontiers in Education 2008.8 Johnson, D.W., Johnson, D.T. & Smith, K.A. 1998. Active Learning: Cooperation in the College Classroom, 2ndEd. Edina, MN: Interaction Book Company.9 Hausfather, S. J. 1996. Vygotsky and schooling: Creating a social context for learning. Action in TeacherEducation. (18) 1-10.10 Jones, D.W. 1996. Empowering teams in the classroom can work. Journal for Quality and Participation, 19(1),80-86.11 Schmidt, L., Schmidt, J., Colbeck, C., Bigio, D., Smith, P. & Harper, L. 2003. Engineering students and training inteamwork: How effective? Proc. American Society for Engineering Education Conference.12 Smith, K.L., Sheppard, S.D., Johnson, D.W., & Johnson, R.T. 2005. Pedagogies of Engagement: Classroom BasedPractices, Journal of Engineering Education, 94 (1) 87-101.13 Bruner, J. 1985. Vygotsky’s theory and the activity-oriented approach in psychology. In Culture, Communication,and Cognition: Vygotskian Perspectives, J. V. Wertsch, ed. New York: Cambridge University Press.14 McDowell, C., Werner, L., Bullock, H., Fernald, J. 2002. The effects of pair-programming on performance in anintroductory programming course. ACM Technical Symposium on Computer Science Education (SIGCSE '02).15 Schellens, T. & Valcke, M. 2004. Collaborative learning in asynchronous discussion groups: What about theimpact on cognitive processing? Computers in Human Behavior, 21(6), 957-975.16 Wells, G. (Ed.) 2000. Action, Talk, and Text: Learning and Teaching through Inquiry. New York: TeachersCollege Press.17 Ferster, C.B. & Skinner, B.F. 1957. Schedules of Reinforcement. New York: Appleton-Century-Crofts.18 Kapitanoff, S.H. 2009. Collaborative testing: cognitive and interpersonal processes related to enhanced testperformance, Active Learning in Higher Education, 10(1), 56-70.19 Slusser, S.R. and Erickson, R.J. 2006. Group quizzes: an extension of the collaborative learning process, TeachingSociology, 34(3), 249-262.
ASEE holds the copyright on this document. It may be read by the public free of charge. Authors may archive their work on personal websites or in institutional repositories with the following citation: © 2012 American Society for Engineering Education. Other scholars may excerpt or quote from these materials with the same citation. When excerpting or quoting from Conference Proceedings, authors should, in addition to noting the ASEE copyright, list all the original authors and their institutions and name the host city of the conference. - Last updated April 1, 2015