I have had students work in groups for my whole teaching career. I love having students share their thoughts collaboratively, in investigative activities, through practice, to discover and to review ideas.
In my six years of teaching, I have always worked hard to engineer the perfect groups. I would ask myself questions like:
Who seems like they would make a good pairing?
Who do I want to avoid ever putting together?
Who would have a good learning experience by needing to slow down or speed up with certain people? who are the students who keep a group on task and who are the ones who lead a strong leader in their group?
Do I want the groups to be more homogeneous or heterogeneous this time?
At Twitter Math Camp this summer, Alex Overwijk gave a presentation on Visible Random Groupings. http://slamdunkmath.blogspot.com/2014/08/vertical-non-permanent-surfaces-and.html Although I didn't attend, I did read his post and hear lots of people talking about it. It got me wondering if all of the effort that I put into grouping students is really helpful. Students are very perceptive and even when I'm not intending to send a message with my groupings, they often pick up subtle ideas that I didn't even think of consciously. A student believes that I put them in a certain group because they're the "smart one" or the "off-task one."
Although it needed some time to ferment in my mind, I also read Lani Horn's book, Strength in Numbers, last winter. She really encourages teachers to get out of the business of grouping students. I wasn't quite ready for her message yet then.
So, this fall I created my first seating chart by having students get in order of how far they live from the school and counted off by 5s to create groups. Next, we used the numbers in addresses to get in order. Then, we got in order by height and counted off. Tomorrow I'm going to have students jot down their favorite number, get in order and count off.
So far, I've noticed an incredible difference! The students don't look to me and try to figure out why I placed them in a certain group. There is absolutely no complaining about groups. Social barriers seem to break down. I have way fewer behavior issues (this is a much more well-behaved class overall, but I think it helps).
I think that the visible part is key. Even if you do it randomly yourself, students won't trust it unless they can see that it's truly random.