Mi papi by the numbers

Now that I've had a night of real sleep, I've done what I usually do when I try to make sense of something.  I rounded up the data.

Here is Theo's show record at USDF santioned shows to date:

Not a lot of data to date.  He's only been to two shows and had eight tests.  I can't draw a lot of conclusions from such a small data set, but hey, that's never stopped anyone from making decisions off of inconclusive data.

So our average test score when up over two percentage points from the first show to the second.  Nice!  Our submission score improved greatly due to a lack of broncing and frightening the judges.  I also got an uptick in my position and his gaits.  We got a hit in impulsion, but that makes sense when three of our five tests were judged by someone that thought Theo was lazy.

Broken out by judge, yeah, we had two very different reactions this past weekend.  Janet Briggs saw us twice at GMHA and we averaged 65% in her ring.  Susanne Handler saw us three times and we averaged 60% in her ring.  The Kristi submission score is highlighted because that includes the bronc test with a submission score of 3.  The Susan Buchanan test was also a bronc test, though a small one.

We are pretty much the same regardless of test.

Breaking out the three Susanne tests, it was interesting that when we upped our gait score, it did nothing about her energy comments but we took a bigger hit to submission than what we gained in gaits.  In the end, regardless of her comments, she scored our most relaxed test the highest (our comments included 'much too relaxed').  Our most electric test was the lowest.  Evidently accuracy and submission really does trump all in Training level and I need to ignore the constant judge comments about need more energy and more forward.  Well, not ignore, but not strive to push Theo toward explosion land.  Impulsion and submission are the two collectives with coefficients.  Our impulsion score never moves, so I better focus on that submission score that swings from 3 to 7.  I need more 7s in submission and position.  Those I can fix.  I can't do a dang thing about the fact he's an earthbound type of a horse.

I feel better after going through the numbers.  I was on the right track.  I don't regret my decision to experiment with my last test.  I wouldn't know what the results would be if I didn't try.  Now I can dial in my approach and give Theo the most appropriate ride possible.

I love data.

Findings

I didn't really think I would find anything significant when I started this project. I figured I'd find no relationship between the number of horses ridden at a trial and horse/rider falls. There are just too many other things going on, like the individual riders, the levels of competition, etc. Lo and behold, I think I actually found something. From my write up:

Summary

The question of how many horses one rider can safely compete with at a horse trial has come up repeatedly within the eventing community. This analysis is an attempt to show some of the trends and relationships that occur when a rider competes with multiple horses.

Three questions were asked:

1. Do riders with more than one horse have more horse falls/rider falls and Did Not Complete (DNC) than riders with just one horse? The answer is no, riders with multiple horses had a smaller percentage of falls and DNCs than those with one horse. This data is skewed, since most amateurs are grouped in one population (one horse) and most professionals are grouped in the other (multiple horses), making the results inconclusive. There is no correlation between horse/rider falls and having two or more horses. However, there is a correlation between the number of horses ridden and falls when there are six or more horses with the data showing a change in the relationship at eight horses entered in a horse trial.

2. Is there a trend between horse falls/rider falls and DNCs when the number of horses ridden increases? There is a trend from one horse ridden to seven horses ridden with decreasing percentages of falls and DNCs. With eight to twelve horses ridden at one horse trial, the results become unstable due to the small population but show a general upward trend. This suggests that there is an effect on DNCs and falls when eight or more horses are being ridden by one person in one trial.

3. What is the percentage of horse falls/rider falls for the riders with the top average number of horses shown in a horse trial? Is there a trend within their own riding as the number of horses shown increases? The results were mixed. Some riders showed increasing numbers of DNCs and falls while others showed no effect. It appears to be individual, with riders that have the appropriate support crew and experience are able to safely handle larger strings of horses. With eight horses or more, it appears to be more likely to have a fall or DNC, and the level of the competition becomes very important.

In conclusion, there is little correlation between rider falls or horse falls and the number of horses being ridden when the rider has seven horses or less. An experienced rider with seven horses at a trial is still less likely to have a DNC or a fall on any one given ride than a less experienced rider with one horse. The experience of the rider and the level of the competition appear to be stronger indicators than the number of horses ridden. However, when there are eight or more horses, there is a relationship between the number of horses and the possibility of a given ride ending in a fall or DNC. These rides account for only 0.4% of the rides in 2010, with only five riders competing with eight or more horses at a horse trial in 2010.

Regulating the number of horses at a trial will most likely not result in a significant change in the overall number of falls. Only 0.4% of the population falls into the area where there is a correlation between falls and the number of horses being ridden. That aside, safety is paramount to the sport, and preventing any horse or rider falls is valuable. Capping the number of horses at seven or eight entered at a horse trial would have minimal effect on the events, as having eight or more horses is already a rare occurrence. This would also set a precedent and would prevent the rare occurrence from becoming more commonplace.

What the heck does all of that mean? It means that if they're riding seven horses or less, they're just as safe or safer than a rider with just one horse. Eight horses or more, there is an increase in the percentage of falls and DNCs, but that is pretty rare. Not even our most prolific pros show up with eight or more horses very often. Should we put a maximum number on horses at a horse trial? Eh, we can, and it's not a bad idea. Seven or eight horses per trial shouldn't put a major dent in anyone's training business. Other than a handful of occasions, no one will even notice. But it is a good idea in terms of stopping a trend before it can even get started.

I'm not surprised that a person can ride seven horses at one show and do it safer than little ol' me and my one horse. Just because I can't get around Rolex once doesn't mean that Phillip Dutton can't do it three times in a day. By the same logic, just because I can't safely show seven horses in one day, that's no reason to believe someone else couldn't. I don't have an army of grooms and working students to help me out, for one. And let's keep in mind, when they have a big string, they're not all going around at the FEI level. A lot of them are green beans working their way up. Skipping around Novice on a new horse isn't going to make a top of their game pro break a sweat.

In the end, pros know what they're doing. They aren't going to risk their bodies more than they have to because that's their business. It's how they pay the bills. I would support a cap of seven or eight horses more to take the pressure off of them. They won't have the pressure to slip one more horse in to get the extra miles, then can point at the USEA and say that it's not possible. It would also be a safety against less experienced pros overreaching. It would have little effect on entries, with only 154 rides being grouped in with a rider competing on a string of eight or more.

Will it solve world hunger and bring world peace? Not even close. There are so few rides in this category that it probably won't even make a dent in the overall statistics, but saving one horse or rider from a fall is still one fall prevented.

The full writeup is in a Word document that refuses to become a Google doc (darn charts and SPSS objects), so it has been loaded to the dear husband's website so that it can be downloaded by anyone that's geeky enough to want to see the details. It's nothing fancy, but it's an objective view of the issue. I have no say in this whatsoever, being a LL rider with no connection to any UL rider. Hopefully it will be of interest to someone.

For science!

This is my mad scientist expression, complete with my evil sidekick

Today is a red letter day for geeky eventing analysts everywhere. Even though I'm pretty sure I'm actually the only one.

I have finished the data entry for my riders with multiple horses project. All 40,000+ rows of data are done, putting every ride in the US in 2010 into one spreadsheet. I have never been so happy to see the end of a list of events. So far I've learned that we have some very creative namers out there. I want to know the story behind the horse named Silent Donut. There has to be a good story behind a name like that. I also liked Dad's Empty Pockets and My Tuition.

Now that I have my database, it's time for the actual data mining to start. The question is whether or not riding multiple horses makes a rider's fall or horse's fall more likely. Of course, after spending months building this database, I'll probably check out some other trends as well. I don't have a lot of data fields, just what's posted on the USEA's official results:

Horse name
Rider name
Dressage score
XC score
Jumping score
Final placing
Number of horses ridden by the rider at the trial
Horse trial name
Level
Area
Month
Year

I want to look at this question of multiple horses a couple of different ways. Some are really simple, some are a bit more complex. I don't think just one view is going to answer the question:

1. Do riders with more than one horse have more horse falls/rider falls and DNCs (did not complete) than riders with just one horse? This is the most literal way of looking at the question, but will probably be skewed due to riders in the lower levels that are just learning the game.

2. Is there a trend between horse falls/rider falls and DNCs when the number of horses ridden increases? This will take out all of the rides where there was just one horse and look at the trend as the number of horses increase. This should give a clear picture of whether or not increasing the number of horses actually has an effect.

3. What is the percentage of horse falls/rider falls for our riders with the highest average number of horses per trial? Is there a trend within their own riding as the number of horses increases? This will probably be the top twenty examples, compared back to the overall average. These are the extreme cases that can show what is actually happening when up to 12 horses are being ridden at a trial by one person.

Does anyone think I'm missing a question or a view? Also, should I use rider names or not? This is all public information to begin with, anyone can go get this, but it feels weird to use actual names after so many years of never, ever leaving identifying information in an analysis.

I'll also do some descriptive statistics, taking a look at what the database shows, but most of it is stuff the USEA publishes already, anyway. I'm already kicking around the idea of adding past years so that I can do year over year trending, but we'll see if I can convince myself to do it. For science!

Data entry

Yes, I did actually ride my horse today, but that isn't what's on my mind. Could be that I'm blotting out that horrible four beat thing we were calling a canter in my lesson tonight . . .

But more to the point, I'm still thinking about the amount of data I'm collecting. I had no idea how many entries there were in 2010. None. And as I collect them all and put them in a spreadsheet, I have to touch each one to get the data cleaned up. I haven't even made it to Chatt Hills yet! I keep plugging away at it (I'm about half way through Area 3) since I think it's something worth doing. It would be terrible if everyone got up in arms about a petition over something that was, in the end, a non-issue. So far the most horses I've seen at one horse trial with one rider is twelve. I can't even imagine handling that, even if it's over two days and you have an army of grooms. That's a lot of rides!

I can barely handle getting the princess ready for her lesson. Mud season is in full swing and she's taking full advantage. She was just covered in mud today when I pulled her out. Of course she was, because I was running late. I swear they have some sort of traffic radio being piped out to them.

Traffic is slow on 495 northbound, keeping riders from showing up with adequate groom time. We recommend all horses with riders coming from the south and lesson times around rush hour to go get as much mud on them as possible. And next is sports, with an update on the scores from the stupid trotting around in circles phase from the Fork.

It's time to get back to the data entry grind, even if I'm going a bit cross eyed from it. Areas 1,2, and 3 are a large chunk of the trials, so I'm feeling somewhat accomplished that I've made it this far. Hopefully I'll have this part of the project done by the end of the weekend so I can get to the fun part.