It seems to be a fairly common belief on here whenever there are polls or important topics about FRC that most of the teams who compete in FRC aren’t on CD. “CD is only good teams” and " CD is a small percentage of the actual teams who are experiencing this issue." I thought I would look into that a bit, here’s what I found.
Using the Users tab on CD, I pulled out the blue alliance team information from the profiles who had at least 100 site visits between 7-2-19 and 7-2-20. I realize this is probably a poor range due to COVID-19, but it is what it is. I used @Caleb_Sykes FRC ELO document to get the ELOs for teams from end of season 2019. I also used this document to determine the total number of active teams in 2019 (3760).
Belief #1: “most teams who compete in FRC aren’t on CD” - 939 teams represented on CD vs 3760 active teams is ~25% representation. Confirmed.
Belief #2: “CD is only good teams” - The ELO histogram comparing all FRC teams to teams represented by CD users shows that the distribution of CD users’ teams skews much higher than the overall FRC population - Confirmed.
Belief #3: “CD is a small percentage of the actual teams who are experiencing this issue.”. No data presented to address this myth but given that CD is only 25% of FRC teams then if an “issue” is widely affecting a lot of teams then it is likely that CD users only represents a percentage of those experiencing it - Plausible
Edit - by the way, this is mostly tongue in cheek as most of the terms like “good teams” and “small percentage” are subjective at best…
Yeah I thought it was interesting. Going into it i thought it would skew the other direction, that CD would actually have a fairly decent representation of FRC, but that doesn’t seem to be the case, at least for this data set. I manually clicked on each profile then typed the team numbers in, so it took awhile… but I’m sure someone could extract the data more efficiently somehow.
I think it’d be interesting to see a heat map of representation by state/district. I am not sure how you could do it but I am imagining something where each region is colored a shade of red to green based on percent representation (region teams on CD/total region teams). The middle/cross over point would be the total percent (teams on CD/total teams).
In the Elo graph, there are “CD Users (20+ replies)” and “CD Users (ALL)”. Did the criteria for active user change at some point from 100+ site visits to 20+ replies? Or is this something else?
Personally, I’d argue that both of these criteria are too strict to be considered an “active member” if “active” means “still in use” and not “regularly contributing”. A user on a lower-performing team who checks CD twice a week through their build and competition season is active IMO, but might only visit early Jan until late March which is 25 or so times. 100+ visits means you check CD at least twice a week for the whole year, which I think is more than the minimum for being an active user. And filtering by number of replies isn’t great either because an active user might read a lot but only post something if they have a question so they wouldn’t get counted (I was like this for my first year or two on here).
Since the threshold was 100 visits, perhaps “most teams who compete in FRC aren’t ACTIVE on CD” would be more accurate. I wonder how this data would look if the criteria was like 25 visits from kickoff to covid close-down.
I don’t disagree, but OTOH, if you are asking “does this survey represent FRC?”, then chances are you won’t catch the team who visits 2/week only during build (since the survey is like not then). You are most likely to get a sampling of people who visit moderately often. So, this is still a useful measure.
If that’s the question, then you need to measure team number metrics for those that answer the survey, which would be different from those that are active in any given time period.
I think both of these would make valid criteria, depending on what you’re trying to measure. People that visit often are going to be receiving information from CD, while those that post often are providing information, in a broad sense. If you’re doing a PSA, you’ll know how far that reaches with the first group, while if you’re gathering opinions the second group is more important.
The threshold for how many visits/posts are needed to be considered active is something that needs more thought - the numbers used here seem arbitrary. I would propose looking at non-binary groupings. Instead of just active/inactive, have 5 groups: inactive (no activity), slightly active, somewhat active, very active, and extremely active, corresponding to the quartiles of the distribution of activity. Doing that would require a bit more analysis of the raw data, though. It would be interesting to see how the bell curves differ between those different groups.
With the Internet/Search and the fact, you don’t need to be a member of CD to read, I am not sure what you can draw out of FRC participation/representation on CD. I would assume almost any interested Mentor or Student would come across CD as one of the top searches on day one of searching for info.
I suspect there are a lot of lurkers from about every team here when it’s important. People like information and CD it is where the information mostly is.
Must be careful to draw conclusions about the engagement of non-members who use this site regularly when needed.
Those that have accounts do so for a reason, my reason was to improve scouting and gain more insights. Does that make my team better? In a way yes as it keeps me engaged.
But there’s also teams who don’t even know the official software docs or examples exist. I wouldn’t put too much faith in teams looking for resources in lieu of just winging it. Even my team does that (which has existed for 10 years) and I have to constantly remind them resources exist. Of those members that know about them, even fewer actually use them.
I get a lot of team folks won’t search, some will, and I find it hard to believe none will. I think on many teams there are one or two that do (you might be the one that does) . Then the team relies on them to bring back info they can use. The persons on CD learn more, then that gets filtered to the team
For me at least it’s a learning event and giving event. I see that a lot here.
The criteria for all the data I used was 100+ site visits in the past year. The 20+ replies just drills down a little bit further. I thought it gave an interesting perspective on how many users post vs. Just view posts. Out of the 1949 teams with 100+ site visits, 386 had no replies.
This is a good idea. The numbers I picked were arbitrary, and somewhat based on how many values I wanted to manually pull. If there was an easier way to get all the user data with the team numbers exported, this would be a simple, interesting look at the data.
This is optimistic. I know a fair number of teams that have one or two students that do “discover” the broader FRC community and the resources and best practices therein, but very often their team is so ingrained in their existing way of doing things that those one or two people fail to affect meaningful change.
Alternate interpretation: Most teams who compete in FRC don’t have CD users who self-identify as being on the team.
With a 2/3 majority of users not listing a team identification we can hardly make assumptions about how many teams are represented. I am no longer associated with a team, and others may be like me. I’m sure there are some who would not want their team affiliation known. Others simply declined to fill in their profile.
