The last time I taught introductory statistics, I calculated final grades according to the higher of the following two methods.
1. 20% graded group projects (4 projects), 30% for in-class 65 minute exams (3 exams, 10% each), 30% for take-home exams (one at midterm and one at final, 15% each) and 20% on a 2 hour in-class final exam.
2. Method 2 is similar to above with 20% from homework, 30% from the take home exams, and then 50% from the in-class final.
I take the higher of the two scores calculated above. For students who "bomb" an in-class exam, method 2 gives them a chance to redeem themselves if they do well on the final. It is rare that students end up with higher averages as a result of method 2 above, but it gives me a way to offer hope to students who say they bombed a test and there are "late bloomers" who sort of "get it" closer to the end of the course.
The next time I teach the course, I may add a clause that students need to obtain at least a score of 50 on the in-class final in order to pass the class. (Note the in-class final is cumulative of material from the entire semester.) Because the group grade counts for 20% of the score, some students who had scores below 40 on the in-class final earned D's or even a C in the course. I would like to try to prevent that from happening in the future.
My in-class exams are 65 minutes, open notebook, and open textbook. I find this means I don't have to worry about copying tables or giving them formula sheets. I generally have 35-50 students in a class. I do not take steps to give multiple copies of my exams and cheating is not a major concern. It probably occurs, but I have not taken any extra efforts to prevent it.
The structure of my in-class exams and in-class final is not multiple choice or short answer. Usually it involves interpreting a word situation. I give lots of partial credit. The hardest part for me is figuring out what to ask in the age of technology. I am trying to ask more concept and interpretation questions. Right now, I do give SPSS output a good deal and ask students to use that output to answer the questions asked.
o I give them 65 minutes (we are on semesters, but our class is 4.0 credit hours). I try to write the exam so that everyone will finish, but that is not the case. I try to warn students they need to know the material well when they arrive for the exam. There is a tendency not to prepare well with open book exams and I think my students suffer from that malaise. I also give two take home exams (midterm and final.) I have about 20 unique data sets and questions that I give to the students. These are coded in such a way that each student thinks he or she is getting a unique exam. I give students at least a week to complete these take-homes. They are scored 50% on the statistical accuracy and 50% on the writing components. It is the same rubric that is used to grade their group projects. I think the students do "cheat" on these take home exams. I encourage them to come to me with their questions, but I think they ask their classmates. I am "ok" with simple questions that involve generating the correct SPSS output. I do think that having their own data set to analyze discourages having a friend to do the whole project for someone. I am not sure what more I can do to prevent cheating on the take-homes.
o I use exams because it is tradition. I have not thought about this a great deal. I am leaning towards not doing traditional exams in the future. I am becoming a firm believer in using authentic assessment where students are tested in situations similar to what they will face in their subsequent courses and careers. I have not let go of the traditional paradigm of testing though. I want to start designing in-class exams that test concepts, ideas, and interpretation. I want to get a way from calculation questions. I do think the in-class exams do provide a structure that students are familiar with and that is a good thing.
o As I said earlier, I do not take any extra steps for cheating. I think it occurs, but I do not know how to prevent it without a great deal more effort than I am already putting forth. I think the cheating that is occurring on the take-home exams is somewhat benign. Since some of my weakest students continue to submit take-home exams that are still weak, I think if they are getting help, they do not know how to properly execute or interpret that help.