Reddit MMR and other stuff

The raw data wasn't optimal for the things I wanted to check, this is why I only used parts of the data for most of this, with a focus on differences in MMR in relation to various other variables gathered.
The first thing I did was removing all entities with no MMR (Solo or Team). Besides that I also removed the ~100 people who listed themselves at 7200 MMR - in case one of the few 7,2k MMR people participated - sorry :p
After that I ended up with roughly 15,500 entities that I worked with.

1. The average MMR

The average MMR for the remaining entities is 3617 for Solo MMR and 3558 for Team MMR.

2. Average MMR by role

This was the part I was most curious about. As a support-player myself I often feel that even if I'm playing a perfect game, the impact it has on the result is much lower than for example a dominating mid. In the initial poll it was possible to select multiple values, I decided to focus on the people only selecting one role. The results show that with 3901 Solo- / 3798 Party-MMR the mid-players have the highest MMR out of the different roles, followed by offlaners with 3754/3683 MMR. With MMR around 3500 both support roles are at the bottom of the list.

3. Average MMR by server-region

This result is interesting if you play in different regions which will result in a worse ping in some region. I listed all servers and server-combinations that had enough entities to give proper results. Players who play in different server-regions seem to struggle more than players who just play in a single region, most likely the one where their ping is best. The biggest difference is between players who play in both, US East and EU West regions compared to those who only pick one of them.

4. Average MMR by hero choice

This one is a bit tricky as the amount of entities who pick by current meta is much smaller than the other, the MMR difference could come from just that.

5. Average MMR by drafting adjustments

Seeing the result on this surprised me a little. I expected a balanced team to have better chances of winning than a team with weird setups. But it seems that people run better with the roles they are familiar with instead of doing something different.

I'd love to do another survey at some point to get different and more specific data. Feel free to leave suggestions!