Why Duck? I never thought I would write a quizzing program. I really think that 'drill and kill' software is evil. I think it encourages students to do exactly the wrong things. Most software I know like this encourages students: (1) to focus on getting the 'right' answer rather than to construct a 'good' answer and (2) to relinquish the authority of knowledge not just to a teacher, but to the teacher's machine surrogate. There are other evil things about this software, but these were two that I thought I might be able to do something about. The next few paragraphs represent a diatribe about transmissionist vs constructivist teaching paradigms. You may want to skip a bit if you're familiar with these ideas. If you want, skip to the paragraph that starts with 'Duck.' Much of current educational practice is like a conspiracy between students and teachers in avoiding meaningful learning. It is as though we tell students, "It would be nice if you could actually learn to be a biologist or whatever, but that would be expensive. Instead, we're going to ask you some questions and if you can answer those questions the way a biologist would, then we will give you a piece of paper that says you're a biologist." Students mostly try to play the game by the rules and cram as many of the answers into their heads as possible. Many even think that the enterprise is valuable and are enthusiastic about being able to remember all the trivia. By 'question' here, I mean (for the most part) the multiple choice questions that students respond to with a bubble sheet. These questions are usually 'causes to effects' and are fairly carefully constrained so as to have a single 'correct' answer. Questions like this include: "the mitochondrion is (a) an organism, (b) a species, (c) an organelle, (d) an organ" and "Using a model of simple dominance, a mating between heterozygous individuals will produce what ratio between dominant and recessive phenotypes: (a) all dominant, (b) 1:1, (c) 3:1, (d) 1:2:1". Questions like these usually represent some fact or skill that a domain expert would know or be able to apply. The problem is that being able to guess the right answer just isn't a very useful goal. It doesn't say anything about whether the student could apply the idea in another context. It doesn't say anything about whether the student understands why the answer is right. It doesn't say anything about whether the idea has affected how the student actually thinks or whether they have just remembered abstractly that in this course they have to answer the question a particular way to get a point. It may represent learning, but it tells us little about meaningful learning. By meaningful learning, I mean learning new ideas in a way that they are integrated into a student's personal understanding of the world. This means that in learning, the student has explored their prior conceptions and extended them to account for new phenomena. Meaningful learning means that the student, rather than trying to simply remember ideas transmitted to them by an instructor, becomes an active 'constructor' of new personal knowledge by engaging in a dialog with an instructor and peers. Ideally, we might challenge students with complex, open-ended questions that require developing an educated opinion regarding what a 'good' answer might be like. This is what biologists actually try to do (and at the graduate level, they follow that up with experimental research to test the veracity of their answers). We mostly don't ask undergraduates to wrestle with these kinds of questions. They're hard to grade. When you ask students to solve these kinds of questions, they hate them. Why? Because they don't have a right answer, and students know that the test that will evaluate their performance in the course will only be concerned with questions that have right and wrong answers. A final point in my diatribe is that when someone says that these questions have 'right and wrong answers' they're just being silly anyway. If we asked a bunch of different people to examine many of the questions used, we could find a whole range of interpretations that would cause different responses to the questions to be true or false. It is in this give and take among people regarding assumptions and models that wisdom lies. It is the human negotiation and interchange that should be at the foundation of our knowledge. Truth does not reside in a book: if it resides anywhere, it is approximated in the discourse of the communities of people who seek answers to biological (and other) questions. Giving computers the authority to tell students that they are right or wrong encourages students to see the truth as black and white. Duck was written mainly as a response to the Owl program, a web-based quizzing program developed by the UMASS Center for Computer Based Instructional Technology. I wrote Duck mainly because Owl was written only for Windows NT -- a platform we have avoided supporting in the Biology Department. If Owl had been available for Unix/linux, I probably wouldn't have bothered writing Duck. Since it wasn't, and our faculty wanted something like it, I decided to put something together. In a week, I wrote a set of scripts in PHP (an open source scripting language) that interact with mSQL (an open source database package) and Apache (an open source web server) to provide a web-based quizzing package. I didn't try to replicate Owl -- Owl has a bunch of powerful features that I doubted my faculty would use. As far as I could tell, my faculty were mostly interested in making sample exam questions available for students to look at. In writing Duck, I just tried to ameliorate the negative aspects of having students do this. We made a bunch of minor changes and extensions to Duck over the first semester we used it, but it has remained pretty much the same. It is neither elegant code nor a fancy interface, but it works very reliably for us. Duck questions can be multiple-choice, short answer, or extended response. For multiple choice and short answer, the instructor can define unique feedback for each item (up to 9 items) that a student might select or enter. There is also a default feedback which is returned if nothing matches a short answer (and is also returned for the extended response questions). The student input for the extended response questions can be either emailed to the instructor or discarded by the system (so that the instructor could have the student answer three questions only one of which is evaluated by the instructor, but where the student doesn't know which). The goal of providing feedback for each item is that a student can try to guess which is 'right', and perhaps get it on the first try. Afterwards, however, they may wonder why some of the distractors were wrong. Students are encouraged to interact with a question until they're confident they can eliminate every distractor. The feedback that the system provides is clearly labeled as feedback from the instructor, not just an evaluation by the machine. The system discourages anyone from taking the data too seriously. The login system doesn't actually keep anyone out, but it does log the login name and password, when a session starts, and all of the responses to the system. This allows an instructor to look at any individual student's performance, see which questions they completed, which items they selected first, which distractors they explored, etc. I tried to make it relatively difficult, however, to pull out a 'score' from the system because I feared it would encourage a shift toward using 'right/wrong' sorts of questions. This was probably unnecessary. The biggest difficulty of the system is writing meaningful feedback. It is easy to get bored explaining over and over again why distractors are wrong, and just putting page references to the textbook where the students should find the right answer. If faculty don't bother to write good feedback, then the advantages of Duck over other systems are probably small or none. I have been intending to conduct a study in which students interact with a similar set of questions: one group with meaningful feedback and the other with only "right" and "wrong". I predict that the group that gets meaningful feedback is more likely to explore items after finding the correct item than the group that gets only token feedback. Looking over the transcripts of student sessions supports this, but I haven't managed to conduct a controlled experiment yet. Faculty have used the system in several ways that I hadn't thought of. One faculty member puts all of the actual exam questions with many correct and incorrect responses. Students are encouraged to explore the responses. The actual exam uses only a subset of the responses. Students were initially discomfited by having questions in the system with multiple correct answers. Although grades in the class were not substantially higher than in the previous semester, students reported increased satisfaction with the 'fairness' of the evaluation in the course and gave Duck high marks in the end. Another faculty member gives students quizzes on paper, and provides the questions with feedback on Duck for students to explore on their own, rather than using class time to go over the quizzes. Other faculty have used the system to post questions posed to students using Classtalk (a classroom communication system) in class. I am releasing Duck as an open source project under the GNU General Public License. I don't know whether anyone else will want to use it or even whether anyone else will find it useful. I hope people find it useful and ideally, I would hope to build a community of developers who would be interested in contributing to the project and moving it forward. My goal is to make the project compatible with the Horde, an open source project that provides a 'wrapper' and libraries that would give Duck a cleaner look and feel and make it easy to write versions in other languages. If anyone has questions about Duck, please don't hesitate to contact me. Steven D. Brewer Biology Department University of Massachusetts Amherst MA 01003