paper: FRC Growth numbers

Thread created automatically to discuss a document in CD-Media.

FRC Growth numbers
by: Jon Stratis

A look at how FRC has grown geographically since 1992.

This is a collection of data series and graphs depicting team growth for FRC, broken up geographically. Each US State has its own graph, as does each district, parts of Canada and parts of Israel. The data is there to do break downs for any other country, state, or province that’s desired, but this would quickly spiral out of control if I tried to do everything. Gotta know when to stop!

A description of the data sources:

All data used for these graphs was obtained through the TBA API. There are some interesting points to take into consideration for anyone looking to duplicate this effort:

  • The State/Province field can contain EITHER the full name (ie Minnesota) OR the abbreviation (ie MN). This requires some work after pulling the data to merge the name and abbreviation datasets.
  • There are a LOT of missing zip codes. It seems that only teams that have competed in the past couple of years have zip codes listed, otherwise that field is null. This limits how precise we can get looking at the data, unfortunately.
  • There are some teams with a null value for their rookie year. This requires some extrapolation based on events teams were listed as competing in.

I don’t fault TBA for these items, it’s likely a result of receiving different amounts of data and different data formats from FIRST at different times (See the TBA blog for more info on that!). Still, it would be extremely helpful for those wishing to run analysis like this if the data could be standardized somewhere, and kept up to date each year :slight_smile:

FRC Growth.xlsx (1.14 MB)

This little project grew out of a seemingly simple question from Andrew:

Over the past few years, I’ve seen a bunch of people (not just Andrew!) hold Michigan up as an example of “Districts = more teams”, or as Andrew put it, “If you build it, they will come”. So, time to stop picking on Andrew :slight_smile:

I wanted to see how growth in all of the districts was happening, and have some actual numbers to back it up with instead of my general impression. So this spreadsheet was born. It gave me an excuse to get into the TBA API (the FRC Event API, unfortunately, doesn’t have the full history available, as far as I know. Just the past couple of years).

Once that was done, it was just a bunch of Excel skills and formulas to datamine everything. What we ended up with are (if you don’t want to download the spreadsheet or wait for the bazillion formulas to calculate) a bunch of graphs. The few I picked out to include below:

So, the data is here for people to look at and speculate on. Personally, my big question is… what drives growth in FRC? Going through the various graphs, you’ll see some sudden growth spikes, and other places where growth dips negative. Is it from the creation of new events? The availability of new funding? A change in the competition itself (introduction of 3v3, districts, split champs, etc)?

My second question is… is there a natural plateau of FRC teams within a given area? If so, what is the driver behind that - competition availablity, demographics, local economy, something else? I doubt we’ll ever get 100% penetration (major sports don’t even have 100% penetration), but is there a percentage we should be aiming for?

FL was a huge spike in 2011/2012 due to JCP grants. I believe I’ve had a LOT of comments about the JCP grants on here and don’t need to rehash why I believe large cash rookie grants are counterproductive to FRC goals in the long term.

I was more theorizing on the impact that “play slots” had on team growth. My gut feel would be to take this data as a function of “slots” in the state but that breaks horribly in regions like NE where a state is what would constitute a county in other regions (I’m looking at you Rhode Island). Any chance of adding a new line to each graph representing “number of plays in state”. Downside, might require a second scale on the vertical axis though.

I never said a whole lot about growth because I’d run numbers and found growth was way too complicated to predict and had way more going into it than “if (districts) growth = 2*growth;” Plus, why do you think I asked it as a “seemingly simple question”? I KNEW it quickly grew into a project, wanted to data, and wanted to see who I could convince to do it for me :stuck_out_tongue:

I’ll have to take a bit to digest this info. Thanks.

There are a number of problems when considering number of plays available to teams. For example, until this past year, many (most?) Western Wisconsin teams considered the Minneapolis events their “home” event, as they were closer than the event in Milwaukee. You almost need to avoid splitting it up by state, and instead create a list of every team’s closest event. That would let you look at the “local carrying capacity” of each event. That’s probably realistic to do for a single year (the TBA API provides Latitude and Longitude for teams and events), but growing that out historically would be complicated. You also have to consider event capacity - I know some events that are still running under their total capacity, yet there’s no way to pull that capacity number to get accurate results.

Besides, if its just a question of filling capacity at the events, then that would imply that adding an event anywhere would cause teams to pop up in that area. I think lack of plays could dampen the number of teams in an area, but the true limit of penetration is caused by other factors.

One of the big drivers of the FiM 2013-2014 spike:

FiM is exploding due to the low cost structure and the grants available. People will play if they don’t have to pay. :wink:

Hey, I didn’t say it was easy… I’ve done studies on team travel in the past and used POSTGIS + geolocation of events based on venue address. But the issue I keep running into is, what do I measure from. Like, sure I could say “give me events under 300 miles from the centroid of teams in this state” but is that a useful metric? Would it be better to cluster teams into events and see which regions are over capacity? I really don’t know, geo stuff is hard.

Thanks so much! Now if you can do this for FTC… (lol!)

The growth numbers are extremely helpful to our pitches!

The following are general musings and not facts.

It would make sense that there would be a plateau-- one of the traditional methods of modelling population growth is with a logistic function, which has this characteristic. It also generally makes sense-- growth is limited by available resources. Resources being money, people, and experience. It would make sense that different areas hit different resource caps than others.

I might also hypothesize that “value” has something to do with team growth-- in particular, Michigan has probably one of the best “values”-- between state grants and districts, participating is monetarily cheap and you get more of it, compared to any regional area or even district areas that aren’t subsidized. Or perhaps Michigan is so much of an outlier that trying to generalize from there doesn’t make sense.

In any case the data is quite messy. I’d imagine that there isn’t quite so much a dominating factor as a number of factors that may be more or less important in different areas.

Here are the overall retention numbers (active teams / all time teams) for each region based on the posted dataset.

You see an overall retention rate in the US of 60%, with MN, MI, & WI, leading the way with the highest retention rates for regions with at least 50 teams.

Correction, the cost structure is actually not low. The average cost per team in Michigan is actually higher than in regionals. The grants that are available there are what really drive the affordability of the program. Of course it’s more affordable for the individual teams when you have the state government chipping in to help cover the costs!

I kind of question using the metric of total teams to look at growth. I would prefer total plays. At some point it seems like the growth of an area comes from teams moving from attending one event to attending two events. Israel and Ontario didn’t really add a lot of teams this year but I bet they added a lot of plays.

Do additional plays for a team increase the number of people exposed to the program? Number of plays and program growth are two separate metrics that measure different things. Increasing growth increases exposure and inspires more people. Increasing plays affects the experience for the same number of people as before. Sure, there’s an argument to be made as to the importance of both cases, but frankly the argument for number of plays is so exhausted on here it’s not even funny any more. There hasn’t been much discussion on program growth, though, other than the annual thread on registration numbers, which doesn’t really look at long term trends for different geographic areas.

Jon, thanks for these metrics.

Just wondering - does MN conduct any survey regarding teams reporting their annual budgets and/or number of sponsors?

I’m wondering if there’s an easy way to find out if the state grants and sponsors are nearing a saturation point if teams can self-report trends in their annual budgets or anything like that.

Any other states/districts try something like this? Can resource saturation be measured in other ways? Mostly in relation to areas of high team density?

A rough number of sponsors may be found by breaking down the “Team Name” field. Obviously there are issues with this, but I think it might be the closest approximation we can get without an actual survey.

With regards to play count, I might agree that there are some issues with it as a metric, but I also think there’s an argument that the overall quality of an area’s FRC community and the experience of students and mentors on those teams is not directly coupled to how many teams there are in that area. I don’t really have a good metric for that, but I think the right concept is there with play count if you accept the underlying assumption that more FIRST is a good thing for participants and teams.

We do have an annual end of year survey, it closed earlier this week actually. Some of the questions we asked on it were centered on budget (total team budget, robot budget, travel budget, etc), but I don’t think we asked about number of sponsors. We’ve only just started looking at the results, so I can’t really say any more than that. One of the problems you have with these sorts of optional surveys is your response percent - how do you extrapolate to cover those that don’t respond? Are some questions going to show inaccurate results because the demographics of those that responded are different from those that did not? How many responses do you need before you can say the results are accurate enough?

And since you mention “high team density”… I think that’s not just a measure of teams per mile, but rather a measure of teams per population for a given area. Major cities can support a larger number of teams per mile than small rural towns.

Total Team Count is a crap metric, as is total plays. Retention and sustainability, however, are likely linked into total plays in a very round about way. We’ve seen that more plays in a season is linked to more on field success (which way this relationship goes is an open question) and teams that are successful are, intuitively, less likely to fold. But this is all based on intuition rather than hard numbers. (See rant about how nobody collects data on why teams fail)

How about total number of miles traveled in the region?

I think it does.

29% of Michigan high schools have FRC teams, compared with 16% in Minnesota and 11% in Wisconsin.

Michigan held 23 district competitions and a very large DCMP this season, providing opportunities for many new folks to get exposed and inspired.

The metric I would really prefer is total number of kids.

When I look around MN, most teams I see going to one regional are 15 or less kids. Most teams going to 2 regionals are 20 or more kids. I know the survey the MN RPC put out this year had the questions to collect this data (at least in MN). My opinion is at some point it gets easier to add kids by growing the existing teams over forming new teams. MN has a pretty deep pool of teams that currently go to 1 regional.

What is DCMP?