I used SI for a couple of years but stopped using it this past fall. The 'motivation' I used was offering them a max of 50% of their in class quiz points back. In class quizzes were worth 5% of their total grade. Just some background. Our SI's were required to attend lectures and run two SI sessions per week. The SI sessions were open to any student in the class (typically around 50 students). That would mean that the SI leader would get swamped with people showing up right before the exam. Not good. Sometimes students from other lecture sections would try and show up too. Also not good. The regulars at the SI sessions usually did not need to be there. Or they felt that going to SI was 'enough' work for the class so they didn't do any work on their own. Having an SI also seemed to make the students less willing to come and see me. I didn't track numbers to have actual data on that, but I do know that when I had the SI I did not have any students I could count on coming to my office on a regular basis (your decision on if that if good or bad). As for training the SI leaders, it was handled through our dean of student's office. They encouraged SI leaders to have minimal interaction with the instructors. Honestly, I think the SI program here was forced on us by people that wanted a feather in their cap for their next job. No one running the program ever asked me what I thought was good or bad about it. When I stopped using it, no one asked why.

I have used SI for five years.  The campus has been running SI for 7-8 years.  We are running three sections of 120 - 150 students of general chemistry.  There is no discussion section assoicated with the course and lab is completely separated from lecture.  In fact, for the two-semester lecture sequence students take a one-semester general chemistry lab course which can be taken with either with gen. chem 1, gen. chem 2 or after they completed the sequence.  This last year it was recognized that SI wasn't working the way it was suppose to work on campus, so the director of the program staged an intervention and had the faculty (and student SI leaders) attend a half day workshop run by people from University of Missouri - Kansas City where SI started.  As to your first question, I teach in a sink or swim environment.  Students that already have the metacognitive sense to realize that they are in over their heads are the only ones that can be reached -- I have learned to accept this.  In the first weeks of class I stop the class and talk about how it is important to recognize what you understand and what you don't understand -- this is typically done after going over a question posed to the class.  On the Monday lecture before SI starts I go over what students can get out of SI.  1.) SI is for the bottom 50% of the class, if the first SI session is too easy don't come back to SI unless you are truly struggling . . spend your chemistry study time more productively.  2.) Network with those that come to SI, where else are you going to find 40 people interested in studying chemistry?  3.)  Ask questions?  Come to SI with questions.  I am not in the room, so there are no judgements about the quality of the questions.  4.) SI is not intended to be a review session, it is intended to be a crutch to allow novice college students to develop into self-sufficient learners.   

This past year we had 12-18 students regularly attending SI (around 10% of the class).   SI leaders are trained by the SI organizer at the beginning of the year.  I meet with my leader 15-30 minutes once a week.  I typically start the meeting asking to see where my SI leader is struggling with running the sessions.  I supplied my leader with POGIL activities out of Foundation of Chemistry by David Hanson.  I gave the SI leader the activity in advance of the meeting and then spent the meeting time going over where students may struggle with the material.  We go over the big picture.  For example, teaching students how to write out and execute the conceptual map of how to solve a limiting reagent problem instead of doing it by rote.  I also give the SI leader information about where students are having problems, this is what prompts my choice of material to supply the SI.  The SI leader is under strict orders not to start giving answers or working problems unless the students ask questions or start to identify their problems.  This worked well this last year, but has not worked well in previous years. 

The biggest pitfall with SI is having students expect it to be a review session or homework session.  If it devolves to that, most of the purpose is gone.  I think having SI for credit as Chip mentioned above really short circuits the entire thing and is a reason why he did not find it useful.  There needs to be intrinsic motivation by the students participating, offering credit kills that.  The other pitfall is finding the right SI leader.  More often than not you don't want the brightest student last year's class to the be the SI leader.  The best SI leaders are the ones that the previous year were weak students at the beginning of the program but then "turned on" during the year.  It is really important that you can find an SI leader that won't give away the answers and makes their fellow students work for it.

I agree, offering SI for some credit was part of the issue. But I do think we had a fundamental problem with the implementation of our program.

These are the goals of SI from office in charge of SI.

Supplemental Instruction was designed with the intention of:

1. increasing retention within targeted historically difficult courses
2. improving student grades in targeted historically difficult courses
3. increasing the graduation rates of students

The first one sounds sort of ok. The problem comes with the second one. Who cares about learning, we just want higher grades. That means high strung students worried that they were possibly going to get an A- would show up as well as students really having a hard time. It did not make for a good mix. As our SI program website states "SI is a non remedial approach to learning as the program targets high-risk courses rather than high-risk students". I could go further, but to sum it up I would say our program had a fundamental design flaw. I know SI programs can be an excellent addition to a chemistry course. I wish ours was. So, I am not trying to turn anyone off to the program. If you want to know how not to run one, feel free to ask me.

We have a Peer Teacher program in which a student can act as an SI for a class, and get a certificate and award at the end of the semester. The SIhas been most helpful during class time, prvoding help to students when completing POGIL exercises. Exam review sessions have been poorly attended, maybe because we also have a tutoring service staffed by trained undergraduates.

In my gen chem course, I post old exams on our course management system for all students (to level the playing field since we know the Greek organizations keep copies).  In order to encourage SI attendence, I do NOT distribute keys-  except to the SI instructors, who are then not allowed to distribute them.