Study on fights

[JusticeZero]

http://jiujitsu365.wordpress.com/2008/03/11/do-most-fights-go-to-the-ground-research-i-conducted/

Read this the other day. It generally confirmed a lot of things I had heard, but a few things came up that caused some thought.

the first person to hit the ground.. either lost the fight (59%) or there was no discernible victor (33%)... This finding recurred repeatedly even if only one person went to the ground or if both people went to the ground. It even applied to situations where both fighters ended up on the ground and the person who initiated the takedown or pushed or punched someone in that direction landed on the ground first. In this study, fighters who hit the ground first were the clear victors in less than 5% of fights observed.

emphasis mine..

Seems to explain the TMA position of "Don't let yourself go to the ground!", as well as giving some validity to those arts which focus more on knockdown than on infliction of damage. After all, if an opponent who falls is almost certainly going to be defeated by followthrough, it is entirely reasonable to concentrate your attention on learning to unbalance an opponent and cause them to fall down - in fact, since techniques to knock an opponent down are not as damaging in and of themself, they can more easily be trained in a resistant context! Sort've like Judo or Taekkyon or various other arts... Just make sure to work in a handful of techniques to finish off a downed enemy.

[bushido_man96]

I think I've seen that somewhere before. It appears that in the case of a takedown, where both go to the ground, the results appear somewhat murky, but it seems to me to imply that if you take someone down, with say a Wrestler's takedown, then ideally they land first, so you have a better chance of winning. Makes sense to me.

Even with this in mind, its important to note that having a good ground game or takedown game will help to defeat the chance for someone else to take you down, thus raising your chance of being the victor.

[honoluludesktop]

The conclusion appears to be based on people in street fights on youtube. It may not have anything to do with skilled ground fighters.

Youtube contains a lot of pretend fights, kid fights, arranged fights, and what gets posted may not be based a true sample set of the general population. You need data like age, size, education, etc. to make a non-spurious correlation between winning fights, and being taken to the ground.

[JusticeZero]

It's still a reasonable sample. It's very hard to find ANY sample of fights that are in any way a good representative sample; YouTube is most likely one of the best ways to find a vaguely representative set available.

Furthermore, it was pretty much explicitly trying to AVOID having "skilled ground fighters" dominating the representation, as the original purpose of the study wat to examine the very high "95% of fights go to the ground" claim. Most certainly if you are an expert ground fighter you will find that most of your fights go to the ground, much in the same way that a boxer will find that a lot of their fights will involve someone throwing punches.

[sensei8]

Are many MAists going to the ground or not going to the ground because of what a report or people say?

I'm hoping its neither, because every MAist must do what they feel is the best for them per the situation at hand for themselves.

IMHO!!

[honoluludesktop]

Being an available sample set doesn't make it statistically reasonable. Behavioral statistics is a science, governed by math, and census data, not what can be found on YouTube.

Aren't we all above average fighters here? If so, even if accurate, the survey doesn't apply to us. I would try to stay off the ground for another reason. So that my opponent's friends don't pound on me, and given the chance, can "exit stage right".

[JusticeZero]

I'm used to having to make inferences from astonishingly poor data sets for the pure reason that that's the best you're going to be able to get your hands on. Are you suggesting that it is ethically and practically acceptable to grab random strangers and induce them to start fights with other strangers in order to get a better sample? It's pretty common these days to record whatever odd thing that people see and post them up on social media. Fights are out of the ordinary. So fights get posted online. Not a perfect sample, but fights are inherently an outrageous enough notable event that they are likely not going to be filtered much by the posters.

It's like trying to gather information on troop movements through a country by sampling the diaries of children living in the area. While it won't give a perfect picture, it will give a good amount of data to work with, because "dear diery. 2day i saw a gajillion skary soljers in tanx!" is a pretty good indicator of some kind of troop movement in armored vehicles near that point, even if it's not as accurate or raw as you might prefer.

[honoluludesktop]

Please don't put words in my mouth. I am only stating that it is a mistake to accept the conclusions of this survey based on its evaluation of YouTube videos, as statistically representing what happens in the real world.

Btw, your example has nothing to do with statistics that are used to predict future behavior.

I have no problems with what you personally wish to make inferences from, or believe in. That is your right.

[JusticeZero]

The "prediction" is just the application of the principle of "This happened in the past, so probably it will happen the same way in the future, as we know of nothing at this time that would indicate change." While I agree that it is not a perfect representation, I would also note that one should not fear to use imperfect data to draw conclusions with, when perfect data is not available.

[DWx]

Youtube contains a lot of pretend fights, kid fights, arranged fights, and what gets posted may not be based a true sample set of the general population. You need data like age, size, education, etc. to make a non-spurious correlation between winning fights, and being taken to the ground.

I suggest that perhaps you guys contact the author. He may have tried to include data such as this. Obviously he might have had difficultly gathering information on thing such as education and occupation but he could have included estimates on age and size based on what was seen in the video.

It's still a reasonable sample. It's very hard to find ANY sample of fights that are in any way a good representative sample; YouTube is most likely one of the best ways to find a vaguely representative set available.

300 videos is not a bad sample size and if you're not going to collect it from YouTube, where else can you get the data from? I agree with Justice in that its as good a source as any. Just got to read the report with that in mind.