BA, the statistician who did the "Overall gi Ranking" for the survey at
Aesopian's Brazilian Jiu-Jitsu Journal has sent me graphs which analyze the data from my survey.
"I have run a multinomial logit on the whole dataset. After cleaning up a few entries and removing (Your Academy's / Association's gi) as I wasn’t clear what to do with those, I was left with 742 responses with a favorite gi and more than one gi owned, for a total of 3097 gis.
The graphs only report gi brands that have more than 10 occurrences. With 2 major exceptions (Padilla & Sons and Sirius), it is reasonably similar to Aesopian’s results with minor re-ranking (by the way his study was based on 563 responses meeting the criteria for being used for a total of 1694 gis). Some new brands are popping up."
In the following graphs, the yellow dots reflect the brand's score for the rate at which each brand is preferred (declared as a favorite) relative to Atama (the most commonly owned brand). The green bars reflect an 95% confidence interval (which is a measure which reflects the possibility for error between the sample and the total population).
This first graph reflects the
Overall results among all the respondents:
A few noteworthy things about the Overall results:
1. One of the biggest differences between my survey and the Aesopian survey were the scores for the brands Sirius and Padilla&Sons:
Padilla&Sons 0.698 -> 0.484
Sirius 0.695 -> 0.286
BA suggested that my polled audience might be less price sensitive than the Aesopian audience and perhaps this is what caused the extreme drop in these two ratings.
2. Perhaps more noteworthy was the appearance of Breakpoint as the overall most preferred gi. In the Aesopian survey, Shoyoroll was the overall most preferred gi with 9 of the 16 Shoyoroll owners calling it their favorite. In my survey, 19 of the 35 owners of Breakpoint called it their favorite gi. The chart above, though, takes into account not only how many people owned the gi and how many people called it their favorite, but also what other gis they owned which they preferred Breakpoint over. Even among different groups within my survey, Breakpoint's rating is very competitive with that of Shoyoroll.
3. While Breakpoint took over the top of the board, many of the same brands from the Aesopian survey remain strong. As mentioned, Shoyoroll is still very high. Isami also remains at the top. Senki and Naja are noticeably high in the rankings, but have very wide confidence intervals (high possibility for error in their ranking) - only 12 respondents owned Senki and only 13 owned Naja.
4. Addidas, BadBoy, Kikskin and Krugans remain at the bottom of the board. They scored low on the Aesopian survey and their scores remained similarly low in my survey. Many generic gis also appeared in my survey at the bottom of the board such as Proforce, TigerClaw and Century.
BA also sent me many graphs which show which gis are preferred among various segments of my sample. Examining the differences in these graphs, we can see if a brand is more or less suited to a certain type of BJJ practitioner.
This graph reflects the rate at which each brand is preferred among people who are "short" (shorter than 5'10):
This graph reflects the rate at which each brand is preferred among people who are "tall" (taller than 5'10):
The most significant difference in score between small and tall people was for Keiko:
KeikoRaca 0.486 -> 0.707
BA had this to say:
"Keiko Raca scores well (prop = 0.579) for small (lighter than 175lbs & shorter than 5'10") people, very well (prop = 0.659) for big (heavier than 175lbs & taller than 5'10") people, extremely well (prop = 0.852) for lanky (less than 175lbs & taller than 5'10") people but very poorly for stocky/squat (heavier than 175lbs & shorter than 5'10") people, so it seems to fit slim people."
This graph reflects the rate at which each brand is preferred among people who are "light" (less than 175lbs):
This graph reflects the rate at which each brand is preferred among people who are "heavy" (greater than than 175lbs):
Clearly there were quite a few things that separated my survey from the survey by Aesopian's:
1. My poll occured about a year and three months after the poll that Aesopian conducted.
2. The sample populations were certainly different (mine coming almost exclusively from online forums which discuss BJJ, or more specifically, BJJ/MMA gear), otherwise we wouldn't have had so many discrepancies in our results.
3. Aesopian's poll had 905 respondents, there were 563 unique respondents who owned multiple gis for a total of 1694 gis. My poll counted 1004 respondents, there were 742 unique respondents who owned multiple gis for a total of 3097 gis.
My survey polled a slightly larger number of people, but far more of my respondents owned multiple gis. The total number of gis taken into account for the measurements of preference was far greater in my survey than the number from the Aesopian survey.
4. The way I phrased my questions and the answers I allowed as options were significantly different from Aesopian's Questions and Answers. Aesopian's survey only had 19 choices of gi brands as options (but did give space for "other" to be written in) while my survey listed 49 gi brands as options (but did not allow for "other" to be written in. My feeling was that I listed all the major gi brands and that even if I allowed for "other," there was a good chance that people who owned other gi brands might not take the time to write them in and would thus skew my responses. Perhaps people who had good experiences with other brands would be likely to write in that brands name while people who had bad experiences would choose not to write in the unlisted brands name - then the results would show that brand to test falsely high. I felt that it was best for me not to allow "other" to be written in, but to try and list the gi brands which were available on retail websites.
One thing I did not account for were a few brands which cater to audiences that I wasn't paying attention to. The female audience complained that I did not list the brands which cater specifically to women (Fenom was mentioned by name a few times). Also, the British audience complained that I did not list some UK companies (Black Eagle and Faixa Rua).
I made my follow up survey far more open ended so respondents would be able to discuss exactly what information they felt was going to be lacking from my initial survey.
One thing that could have been done to improve the survey (and which I will perhaps implement in a later survey) would have been to have people to give a score for each brand they own rather than simply list their favorite. This would lead to more accurate overall rankings and allow us to identify more differences between the brands.
I want give a major thank you to BA, the statistician who did the rankings graphs for both my survey and the Aesopian survey. He went out of his way to process this data for the BJJ community - it's a labor of love for him!
