Being good at one role of a partnered dance like WCS does not make you good at the other role! Anyone who has ever danced both roles can tell you that, but so can data!
UPDATE: On 3/7/2021, I found a few errors (I think they came from poor data cleansing methods when I first manually cleaned it several years ago. I just went back to the raw data and re-cleaned and analyzed using better practices. As of ~9am CST 3/8/2021, it’s all fixed! In the process, I also changed some of the text to more accurately reflect the updated results, though they didn’t actually change much.
I’ve put background info in the gray boxes below, but if you are already intimately familiar with how WSDC treats roles and levels and/or you don’t care and just want to see the pretty graphs, skip the gray boxes!
Also yes, this is long. And yes, I did delete whole massive paragraphs and side-tangents just to get it as short as it is. I think you can get plenty just by looking at the graphs though. But if you want detailed info (and extra info that isn’t represented graphically) and/or my thoughts and how they relate to WSDC rules, you’ll need to read all/most of it.
[Warning: I started writing a background on West Coast Swing (WCS) for any non-westies reading, but I quickly gave up because, again, this is already the short version, haha. But feel free to ask any questions and I’ll try to explain succinctly (or family, you can just email me, or hopefully we’ll all be vaccinated and I’ll be able to see you at Thanksgiving and we can chat about it!)]
First things first, WSDC Jack & Jills (the random partner, random music competition in WCS) currently require you to compete at the same level as both a leader and a follower. So if you compete as an Advanced follower, you are also considered an Advanced lead; i.e. you can enter as a lead in the Advanced division even if you have never competed as a lead before (or ever danced as a lead before). You can petition the head judge at any event to compete at one level below your WSDC level (e.g. that Advanced follower could submit a petition prior to the competition to ask to compete as a lead in Intermediate). There are two major issues with this:
1. These petitions don’t carry over. You have to petition every single time; even if that Advanced follower got dead last in Intermediate at an event one weekend, they’re assumed to be Advanced for the next event. So it’s just a hassle to petition every single time. And again, you can only go down one level, so even if that Advanced follower had never led a dance in their life, they can’t start at the beginning in Novice.
2. Granting the petition or not is entirely up to the head judge. Some judges grant them, some don’t. So again, you can get a different answer every weekend for no real reason.
There’s also one big assumption: that your level in one role is highly predictive of your level of dance in the other role. Now first, let’s set aside that WCS competitions don’t have any external metric and so it’s all just rank ordering and there’s no objective way to define ‘how good’ any given level is (and therefore the Objective Goodness Level of an Intermediate follower has no inherent relationship to the Objective Goodness Level of an Intermediate leader). But even if WSDC level were totally objective, as I said before, anyone who’s danced both roles can tell you they’re different skill sets and working at one role doesn’t inherently bring your skills in the other role up to match. But that’s just subjective feeling about the experience of learning and dancing those roles. If that’s true, we should be able to find evidence that people aren’t equally good in both roles, right?
Fortunately, 1) WCS is big on publicly accessible data and 2) the competitions that could tell us something about what level people dance both roles at already exists and that data is on the internet! Two competitions took place that were perfectly aligned in time with the WSDC having not already limited the level of people dancing a second role (in fact, I believe the rule change happened in response to these competitions) and with a few event directors being open to competitors dancing in both roles and choosing their level.
At Boogie by the Bay 2014 and Liberty Swing 2015, competitors were allowed to enter two skill level J&Js, one in each role. Additionally, they could choose what level they wanted to compete in for their secondary role with no restriction*.
*For Boogie, I know there was no restriction; I’m actually not sure about Liberty. Nobody competed more than 2 levels down, so it’s possible that was a limit set by the event. This is where I welcome clarifications and corrections from anyone who knows!
If it were true that people were generally just as good in one role as the other, we would expect anyone who competed down a level (or more) in their secondary role to have done exceptionally well, basically dominating those lower divisions. But that’s not what we see at all!
But before you get to see those pretty, pretty graphs, we need some caveats:
– This is not a random sample of WCS event-goers/competitors.
– They are likely more serious/experienced competitors (considering most of the people we’re interested in analyzing are Intermediate-Champion)
– They are people who chose to sign up for an extra J&J in their other role at a time when this was not common (and not allowed at most events), making it likely these are people who danced their secondary role much more regularly and/or worked on it much more than your average competitor (and the average competitor probably danced one role exclusively).
– This was only from two events, both in the US, both within a 6 month period. This doesn’t necessarily extrapolate to other events, other countries, and other time periods (but that also doesn’t mean it can’t tell us anything applicable to other events, countries, and times!). If this were repeated now (well, post-pandemic anyway), maybe we’d see something slightly different! A lot more people dance both roles now; maybe that means there are more casual dual role dancers, so maybe they’d do a little worse. Or maybe people who focus on a secondary role now are more likely to start both roles at the same time and therefore actually have worked on both and be more equally good at both, so maybe their levels would be more similar in each role. We can never know without testing it, so for now, the results from these two events are what we have to go on.
– Nobody danced their secondary role at a level higher than Intermediate. From Liberty, there were secondary role dancers competing in Newcomer, Novice, and Intermediate. And at Liberty, there were fewer dancers competing in their secondary role, and they all danced 0-2 levels down from their primary role level (i.e Novice competed in Newcomer; Novice, Intermediate, and Advanced competed in Novice; and Intermediate, Advanced, and All Star competed in Intermediate). From Boogie, there were secondary role dancers competing in Novice and Intermediate. Every Boogie competition had at least one secondary role dancer whose primary level was each possible level above the competition level (i.e. Intermediate, Advanced, All Star, and Champion competed in Novice; and Advanced, All Star, and Champion competed in Intermediate). There were no dancers who competed in the same role for both their secondary and primary role.
Oh, and I also need to tell you a bit about the data source. I got the full results from all leveled J&Js from Boogie 2014 and Liberty 2015 from StepRightSolutions.com, which has full results from a lot of events! This was incredibly valuable because having names (including those who didn’t make it past prelims) allowed me to find every pointed competitor in the WSDC points registry to figure out if they usually competed in a different role and what level they competed in that role in when these events took place. Note that I use “primary role” to refer to the role that someone competed in more often at the time and/or was a higher level in, and “secondary role” to refer to the role that someone competed in less often and/or was competing at a lower level in than their WSDC pointed level.
Edited to add some extra info someone asked about that you may be curious about too:
– In total, there were 660 entries into the 5 contests. 53 entries were in the dancers’ secondary roles, while 607 entries were in the dancers’ primary roles.
– Of the 53 entries in dancers’ secondary roles, 46 also competed in their primary role while 7 did not also compete in their primary role. Of those 7, 3 of them competed in their secondary role at the same level as their primary role, so they possibly didn’t compete in their primary role because of that conflict. Of the other 4 who didn’t compete in their primary role, 2 had earned their last points 3-5 years prior, so it’s possible they were no longer competing in their primary role regularly.
Now for the pretty, pretty graphs! Note that these are alluvial plots and I’ve set them up so the x-axis basically flows through rounds of a competition, while the y-axis shows how they did in the competition (with yes’s and higher placements at the top and no’s at the bottom). Most graphs have the level of peoples’ primary role on the far left column, and then you can see how dancers in the same group (i.e. all of the same starting level) flowed through the competition in their secondary role.
I’m going to start with the money graph though: How well does someone do relative to how many levels down they danced their secondary role?
For reference, the expected outcome is that about half (47%) would get a yes in prelim, about 1 in 5 (19%) would get a yes in semis, and about 1 in 13 (8%) would place top five, 8% would get 6-10th place, and 4% would get 11th+. Also I’ll note that I ran the statistics on it (Chi Squared test), but due to the low number of observations (only 1 dancer 4 levels down, 7 at 3 levels down, 12 at 2 levels down, 30 at 1 level down, and 4 at 0 levels down), there was such a high probability that these results were due to random chance that the only result that seemed to have a less than 5% chance of resulting from random chance (i.e. p<0.05) was the 4 level down competitor placing 1st. Also having secondary role competitors did not at all change the likelihood that the primary role dancers would get any given placement (i.e. just as many primary role dancers made semis and finals and got each of the placements as you would have expected if only primary role dancers competed in the contests).
More important context: nobody danced their secondary role above Intermediate level. This means that the 1-level-down dancers are all primary-Novices in Newcomer (3 dancers), primary-Intermediates in Novice (23), and primary-Advanced in Intermediate (4). And a the dancers who made finals dancing one level down were Novice-dancing-in-Newcomer (2) or Advanced-dancing-in-Intermediate (1); Intermediate-dancing-in-Novice seem maybe a little less likely to make semis and none made finals.
I think this speaks to some level-specific differences that may mean setting a limit on how many levels down someone can dance may not be an approach that works for all divisions. For instance, Newcomer may be too easy a division for even Novice secondary role dancers (because just having those basic movement and timing skills puts them at the head of that pack)**.
**Though I’ll add here that I’ve heard a lot of Novice dancers express that they’d feel more comfortable dancing their secondary role in Newcomer for the first time they compete in it. This is where you have to balance who that division is for and why (and recognize that maybe that primary-Novice is giving a just-learned-WCS-yesterday-afternoon an amazing competition dance in Newcomer, and that has value too). Story time! I actually danced my first secondary role competition (leading) at a comp that didn’t have Newcomer. And I totally bombed. This was when I was aggressively prepping tuck turns the wrong direction, but was completely oblivious to how badly I was dancing. I probably would have done better if I could have done that in a Newcomer comp (and my partners might have been ignorant enough to not have a terrible time *laugh cry emoji*). But also getting straight no’s in prelims there did not dissuade me from continuing to dance (or even compete). I think by the time you have someone who’s competed before, you don’t need to convince them to keep dancing *shrug emoji*. [I assume emojis are possible on wordpress, but I can’t find them 😦 ].
For people dancing 2-4 levels down, they’re generally doing a little better than you’d expect the average dancer in that division to do (though with the major caveats that 1. again, not an unbiased sample and 2. there were two few dancers to make any statistical conclusions). People dancing 0 roles down all got no’s in prelims (they consisted of 1 Novice dancer and 2 Intermediates), so they’re not doing quite as well as we’d expect. Though again, the numbers are so small that these outcomes could just be due to random chance and they all actually performed equally as well as would have been expected if number of levels down had no actual impact on how well you did.
What does this say about what secondary role WSDC policy should be? Personally, I really like the approach Phoenix Grey has proposed: let people choose a level to start their secondary role in, then for that first competition in the new division, if they get points, great, that’s recorded and they can build up to get out of that division as usual. But if they don’t get points, give them a WSDC record of “Entered – 0 points” to show that they did *not* do well enough to dance higher than that level. This way, an All Star can start at Novice, and if they don’t even make semis, they have a record that future head judges can look at to say “Yes, this dancer does belong in Novice for this role.” And then they wouldn’t have to petition or anything for future competitions, but would just work on getting points like anyone else. I think the massive variation we see means a one-size-fits all policy just won’t be the most effective.
Though the other solution would be “everyone starts in Novice”. I would personally be fine with that option, but that’s where people complain about sandbagging/a year of All Stars/Champions crushing Novice. I will say that these datasets do suggest that *could* theoretically be a problem (look at those 3-level-down and 4-level-down flows). However the sample is so small, and I think this is where I think it’s really important to realize that that group who decided to compete in their secondary role at Boogie 2014 and Liberty 2015 very likely to not represent every All Star or Champion who might try competing in their secondary role if WSDC changed the rules so everyone could start in Novice. I expect these were dancers who were more interested in their secondary role and/or had worked on it more than your average All Star/Champion. Also I’m not so worried about sandbagging because both Champions took 1st (one in Novice, one in Intermediate), which means they would both be out of those divisions either at the next event or maybe the one after that. There just aren’t enough Champions and All Stars to clog up those divisions for long. Plus, they’re bringing their partners with them! So if they’re really that good, at least they’re giving amazing dances to their partners who get to move up too! Also that Champion who got first in Intermediate? She now competes in both roles at the All Star/Champion level (and has been doing so for several years; if someone should be in a higher level, the point system is really good at fixing that quickly).
Okay, enough words, time for new graphs! These show outcomes of secondary role dancers by the level of their primary role, and I’ve split it by role. This first graph shows how secondary follows did in their follower competitions (i.e. the left column is their level as a lead). The second graph shows how secondary leaders did in their leader competitions (i.e. the left column lists their level as a follower).
Interestingly, only All Star or Champion dancers seemed to do much better than you’d expect (and remember these are all Newcomer, Novice, or Intermediate competitions). And the number of crossing flows for leaders versus the relatively static flows for followers could suggest that the level of primary leads is more predictive of their following skill, while primary follower level is not as predictive of leading skill. This all is also a little bit of repetition of the story about levels danced down. Intermediate dancers did not do well (nobody made semis or finals) and they were all dancing their secondary role in Novice. The Novice dancers that made it past prelims were all competing in Newcomer.
I also find it interesting (again, caveats, small samples, etc) that it seems the higher level secondary leaders may be slightly more likely to make prelims, but not semis or finals. I wonder if this speaks to the importance of personal movement skills/quality of movement in the prelim judging when you’re only seen for 10-30 seconds. But I don’t have data addressing that specific question, unfortunately…
And that’s basically it! I’m going to put a few slideshows here of the competition-level plots and event-specific plots, for anyone interested! But that’s getting so granular that I don’t think it gives us any useful information.
But thank you for reading (or maybe skimming) this far! I’d love to hear if anyone has thoughts or questions or anything! Especially if anyone has more data science experience with data like this and has anything they’d like to add. Also I haven’t bothered to figure out github yet, but I’m happy to send you my cleaned datasets and/or my code. Everything was analyzed with R and tidyverse and the graphs were generated with the ggplot2-based package ggalluvial.
Both events combined, separated by competition level and role:
Liberty only, separated by level and role:
Boogie only, separated by level and role:
If you’re still here, you probably like graphs and data, and you get to watch read this after-the-credits easter egg: thanks to Sam Hartzell from Minnesota (*super* fun to dance with if you ever get the chance and just an all around great human being!), I got curious what the drop-out rate among Novice competitors (honestly I’m curious about non-competitors, too, but that data doesn’t easily exist the way it does for competitions). So I copied a bunch of prelim results from various years from Step Right Solutions (note: their robots.txt file says no web scraping! I did it manually), and I’m going to use that to try to follow a cohort of dancers from ~2014/2015 and see if they show up on the WSDC points registry, if they remain unpointed but show up in prelims in 2018-2020, or if they disappear! If you want to steal that idea, do it! But also if you want to just see the results, I will probably get to posting it here eventually (but in the meantime I have a totally-unrelated-to-dance PhD to finish, and then I have to get a job and start my career and move states, and of course we still have a pandemic to dig our way out of; all that to say: that analysis and post may not come anytime soon… OR I’ll use it to procrastinate on writing my thesis. Who knows? But don’t hold your breath.)
ScienceSewingSwing 