Announcement

Collapse

Forum rules and no-advertising policy

As a participant on this forum, it is your responsibility to know and follow our rules. Please read this message in its entirety.

Board Rules

1. You're responsible for what you say.
As outlined in Section 230 of the Communications Decency Act, The Chronicle of the Horse and its affiliates, as well Jelsoft Enterprises Ltd., the developers of vBulletin, are not legally responsible for statements made in the Forums.

This is a public forum viewed by a wide spectrum of people, so please be mindful of what you say and who might be reading it--details of personal disputes may be better handled privately. While posters are legally responsible for their statements, the moderators may in their discretion remove or edit posts, though are not legally obligated to do so, regardless of content.

Users have the ability to modify or delete their own messages after posting. Moderators generally will not delete posts, threads or accounts unless they have been alerted and have determined that a post, thread or user has violated the Forums' policies. Moderators do not regularly independently monitor the Forums for such violations.

Profanity, outright vulgarity, blatant personal insults or otherwise inappropriate statements will not be tolerated and will be dealt with at the discretion of the moderators.

Users may provide their positive or negative experiences with or opinions of companies, products, individuals, etc.; however, accounts involving allegations of criminal behavior against named individuals or companies MUST be first-hand accounts and may NOT be made anonymously.

If a situation has been reported upon by a reputable news source or addressed by law enforcement or the legal system it is open for discussion, but if an individual wants to make their own claims of criminal behavior against a named party in the course of that discussion, they too must identify themselves by first and last name and the account must be first-person.

Criminal allegations that do not satisfy these requirements, when brought to our attention, may be removed pending satisfaction of these criteria, and we reserve the right to err on the side of caution when making these determinations.

Credible threats of suicide will be reported to the police along with identifying user information at our disposal, in addition to referring the user to suicide helpline resources such as 1-800-SUICIDE or 1-800-273-TALK.

2. Conversations in horse-related forums should be horse-related.
The forums are a wonderful source of information and support for members of the horse community. While it's understandably tempting to share information or search for input on other topics upon which members might have a similar level of knowledge, members must maintain the focus on horses.

3. Keep conversations productive, on topic and civil.
Discussion and disagreement are inevitable and encouraged; personal insults, diatribes and sniping comments are unproductive and unacceptable. Whether a subject is light-hearted or serious, keep posts focused on the current topic and of general interest to other participants of that thread. Utilize the private message feature or personal email where appropriate to address side topics or personal issues not related to the topic at large.

4. No advertising in the discussion forums.
Posts in the discussion forums directly or indirectly advertising horses, jobs, items or services for sale or wanted will be removed at the discretion of the moderators. Use of the private messaging feature or email addresses obtained through users' profiles for unsolicited advertising is not permitted.

Company representatives may participate in discussions and answer questions about their products or services, or suggest their products on recent threads if they fulfill the criteria of a query. False "testimonials" provided by company affiliates posing as general consumers are not appropriate, and self-promotion of sales, ad campaigns, etc. through the discussion forums is not allowed.

Paid advertising is available on our classifieds site and through the purchase of banner ads. The tightly monitored Giveaways forum permits free listings of genuinely free horses and items available or wanted (on a limited basis). Items offered for trade are not allowed.

Advertising Policy Specifics
When in doubt of whether something you want to post constitutes advertising, please contact a moderator privately in advance for further clarification. Refer to the following points for general guidelines:

Horses -- Only general discussion about the buying, leasing, selling and pricing of horses is permitted. If the post contains, or links to, the type of specific information typically found in a sales or wanted ad, and it's related to a horse for sale, regardless of who's selling it, it doesn't belong in the discussion forums.

Stallions -- Board members may ask for suggestions on breeding stallion recommendations. Stallion owners may reply to such queries by suggesting their own stallions, only if their horse fits the specific criteria of the original poster. Excessive promotion of a stallion by its owner or related parties is not permitted and will be addressed at the discretion of the moderators.

Services -- Members may use the forums to ask for general recommendations of trainers, barns, shippers, farriers, etc., and other members may answer those requests by suggesting themselves or their company, if their services fulfill the specific criteria of the original post. Members may not solicit other members for business if it is not in response to a direct, genuine query.

Products -- While members may ask for general opinions and suggestions on equipment, trailers, trucks, etc., they may not list the specific attributes for which they are in the market, as such posts serve as wanted ads.

Event Announcements -- Members may post one notification of an upcoming event that may be of interest to fellow members, if the original poster does not benefit financially from the event. Such threads may not be "bumped" excessively. Premium members may post their own notices in the Event Announcements forum.

Charities/Rescues -- Announcements for charitable or fundraising events can only be made for 501(c)(3) tax-exempt organizations. Special exceptions may be made, at the moderators' discretion and direction, for board-related events or fundraising activities in extraordinary circumstances.

Occasional posts regarding horses available for adoption through IRS-registered horse rescue or placement programs are permitted in the appropriate forums, but these threads may be limited at the discretion of the moderators. Individuals may not advertise or make announcements for horses in need of rescue, placement or adoption unless the horse is available through a recognized rescue or placement agency or government-run entity or the thread fits the criteria for and is located in the Giveaways forum.

5. Do not post copyrighted photographs unless you have purchased that photo and have permission to do so.

6. Respect other members.
As members are often passionate about their beliefs and intentions can easily be misinterpreted in this type of environment, try to explore or resolve the inevitable disagreements that arise in the course of threads calmly and rationally.

If you see a post that you feel violates the rules of the board, please click the �alert� button (exclamation point inside of a triangle) in the bottom left corner of the post, which will alert ONLY the moderators to the post in question. They will then take whatever action, or no action, as deemed appropriate for the situation at their discretion. Do not air grievances regarding other posters or the moderators in the discussion forums.

Please be advised that adding another user to your �Ignore� list via your User Control Panel can be a useful tactic, which blocks posts and private messages by members whose commentary you'd rather avoid reading.

7. We have the right to reproduce statements made in the forums.
The Chronicle of the Horse may copy, quote, link to or otherwise reproduce posts, or portions of posts, in print or online for advertising or editorial purposes, if attributed to their original authors, and by posting in this forum, you hereby grant to The Chronicle of the Horse a perpetual, non-exclusive license under copyright and other rights, to do so.

8. We reserve the right to enforce and amend the rules.
The moderators may delete, edit, move or close any post or thread at any time, or refrain from doing any of the foregoing, in their discretion, and may suspend or revoke a user's membership privileges at any time to maintain adherence to the rules and the general spirit of the forum. These rules may be amended at any time to address the current needs of the board.

Please see our full Terms of Service and Privacy Policy for more information.

Thanks for being a part of the COTH forums!

(Revised 5/9/18)
See more
See less

1/2 points in dressage: how's that working for you?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    1/2 points in dressage: how's that working for you?

    I'm really appreciating the half points. I think I'm seeing 6.5s much of the time when before it was more than a 6 but not quite a 7 and the judge's only option was to give me a 6. I think the introduction of half points have improved my overall scores a bit. However, when I scribe and when I look at class results, it does not seem like scores in general are changing any.

    What is your experience and impression?
    Hindsight bad, foresight good.

    #2
    Originally posted by Badger View Post
    ..... However, when I scribe and when I look at class results, it does not seem like scores in general are changing any.
    .....
    Because statistically it doesn't make any difference.

    Think of a collection of scores distributed in a "bell shaped curve." This can be for an individual show at a specific level, or for a year's worth of scores at a level.

    This is bell shaped curve describes a histogram, with "bins" or "buckets." When we had integer scoring (eg., 0,1,2,3,4,....etc.) there were ten bins and each "bin" was an integer corresponding to each official score descriptor.

    Now there are half point scores....eg., 0.0, 0.5, 1.0, 1.5, 2.0, 2.5.....etc for a total of 20 bins....eg., each prior bin has been sliced in half.

    So, now you have the same distribution of scores, except, the "bins" are smaller.

    Overall...if one were to look at a large distribution of scores, the shape of the bell-curve (mean and standard deviation) probably hasn't changed much.

    These half-points came from the FEI....and based on a half-baked idea that is not grounded in any accepted statistical methodology.
    Do not confuse motion and progress. A rocking horse keeps moving but does not make any progress.
    Alfred A. Montapert

    Comment


      #3
      Apparently straight dressage has also been using them for a while. I scribed for a judge the other day who was very excited about it. I haven't really seen much of a difference in my scores though.

      Comment


        #4
        Love them.
        Gives a better idea of "not quite an 8, but close"

        Comment


          #5
          Originally posted by FlightCheck View Post
          Love them.
          Gives a better idea of "not quite an 8, but close"
          I agree.

          It may make no statistical difference (because the times a judge gives a 7.5 instead of a 7.0 are probably pretty much balanced by the times she gives a 5.5 instead of a weak 6.0), but as a competitor I appreciate the opportunity for the judge to fine tune the feedback she gives me on the test sheet.

          Comment


            #6
            I love it, too. I think it helps judges be a bit more rewarding-- when a movement is better than a 6, but not quite a 7, the judge can give you a 6.5. I can't complain about my scores thus far this year.
            A clever person solves a problem. A wise person avoids it.
            ? Albert Einstein

            ~AJ~

            Comment

              Original Poster

              #7
              Hindsight bad, foresight good.

              Comment


                #8
                Originally posted by pluvinel View Post
                These half-points came from the FEI....and based on a half-baked idea that is not grounded in any accepted statistical methodology.
                So business as usual for the FEI then?

                Comment


                  #9
                  And on the administrative side you no longer have to worry about a judge who usually just does dressage shows using half marks and then having to go back and edit them!

                  It's easier for the judges to have one system to work with although they're still thrown by the no halting on the first centerline for BN-T!

                  Comment


                    #10
                    Originally posted by pluvinel View Post
                    Because statistically it doesn't make any difference.
                    I understand what you're saying, but there's an assumption that judges see every half step as equal in size. Since judging is somewhat subjective, I think it could make a difference - you'd have to have the numbers to know for sure. Maybe judges are now giving out a disproportionate number of 6.5's or something. It would for sure be interesting to know if the distribution of marks has changed in mean or shape at all.

                    I would think rankings by score shouldn't change because of a slightly different marking system, but saying that it makes no statistical difference without any evidence just seems not quite right to me.

                    Comment


                      #11
                      Originally posted by pluvinel View Post
                      Because statistically it doesn't make any difference.

                      Think of a collection of scores distributed in a "bell shaped curve." This can be for an individual show at a specific level, or for a year's worth of scores at a level.

                      This is bell shaped curve describes a histogram, with "bins" or "buckets." When we had integer scoring (eg., 0,1,2,3,4,....etc.) there were ten bins and each "bin" was an integer corresponding to each official score descriptor.

                      Now there are half point scores....eg., 0.0, 0.5, 1.0, 1.5, 2.0, 2.5.....etc for a total of 20 bins....eg., each prior bin has been sliced in half.

                      So, now you have the same distribution of scores, except, the "bins" are smaller.

                      Overall...if one were to look at a large distribution of scores, the shape of the bell-curve (mean and standard deviation) probably hasn't changed much.

                      These half-points came from the FEI....and based on a half-baked idea that is not grounded in any accepted statistical methodology.
                      um... how can it not matter statistically? Are you sure about that? What if they decided to go from a 10 point system to a 3 point system? Using your logic, that would not matter statistically either. So, I beg to differ. I am not saying that it necessarily does matter but it certainly could....

                      Similar example: what if a college prof decided to go from a 100 point grading scale to a 10 point grading scale? Would that matter? Indeed, it would.

                      Because our range of scores is quite narrow (most of my tests have movement scores that have a 1-2 point range), introducing 1/2 points may be statistically warranted. I will not get into a discussion of error, variance, and the like, as I think the point can be clear without introducing these terms here.

                      Comment


                        #12
                        Originally posted by Ash0815 View Post
                        I understand what you're saying, but there's an assumption that judges see every half step as equal in size. Since judging is somewhat subjective, I think it could make a difference - you'd have to have the numbers to know for sure. Maybe judges are now giving out a disproportionate number of 6.5's or something. It would for sure be interesting to know if the distribution of marks has changed in mean or shape at all.

                        I would think rankings by score shouldn't change because of a slightly different marking system, but saying that it makes no statistical difference without any evidence just seems not quite right to me.
                        (1) - Yes....judges do not use the judging scale as intended. In the scale of 0,1,2,3,4,5,6,7,8,9,10, the original assumption for its use is of LINEARITY....eg., every step between each score is EQUAL.

                        I agree that the judges don't use it that way.....which perturbs the objective of the scale. The scale has descriptors (good, fairly good, etc) which would clarify how to use the scale.

                        (2) - I don't have time to do a mathematical proof for you....but I'm sure you have technical basis for your statement and can share....I would be interested in being corrected.

                        For those of you who may not have been around during the Nerd Herd times, here is a link to the original paper.

                        Unfortunately PVDA changed their web site and will link you to a, Google drive, not to a pdf link, but look for the paper called, "2010 Scoring Variables and Judge Bias in United States Dressage Competitions."

                        https://drive.google.com/folderview?...tWOGtGSzg#list

                        This was written during the days when the USDF attempted to implement a qualifying rule because USDF/USEF leaders were saying there was such "bad riding" out there. A group of nerdy COTHers thought to investigate such statements with data.

                        And There is also another article on the list calld 2008 Score Analysis which was precursor to the 2010 published work.

                        And a link to a USDF Connection Magazine article discussing the Journal for Quantitative Analysis in Sports paper.
                        http://www.usdf.org/EduDocs/Competit...ssage_comp.pdf

                        The JQAS paper is the only peer-reviewed analysis of dressage statistics out there.
                        Last edited by pluvinel; May. 26, 2015, 07:46 AM.
                        Do not confuse motion and progress. A rocking horse keeps moving but does not make any progress.
                        Alfred A. Montapert

                        Comment


                          #13
                          Originally posted by pluvinel View Post
                          (1) - Yes....judges do not use the judging scale as intended. In the scale of 0,1,2,3,4,5,6,7,8,9,10, the original assumption for its use is of LINEARITY....eg., every step between each score is EQUAL.

                          I agree that the judges don't use it that way.....which perturbs the objective of the scale. The scale has descriptors (good, fairly good, etc) which would clarify how to use the scale.

                          (2) - I don't have time to do a mathematical proof for you....but I'm sure you have technical basis for your statement and can share....I would be interested in being corrected.
                          I had no idea that analysis had been done, thank you for sharing, I'm excited to do some reading later today.

                          Regarding (2), if it was a completely objective measure then I would agree that there would be no statistical difference in the distribution. The thing is though there are people involved, so there's a definite chance of skew being introduced to the marks, due to psychological factors. Some previous posters mentioned they thought judges prefer going up a half point rather than down a half point - would that mean that the overall average score is now higher? My point is, there may or may not be a difference introduced, but saying no difference without using some statistical test on the actual numbers seems naive.

                          Comment


                            #14
                            Originally posted by Ash0815 View Post
                            .....

                            Regarding (2), if it was a completely objective measure then I would agree that there would be no statistical difference in the distribution. The thing is though there are people involved, so there's a definite chance of skew being introduced to the marks, due to psychological factors. Some previous posters mentioned they thought judges prefer going up a half point rather than down a half point - would that mean that the overall average score is now higher? My point is, there may or may not be a difference introduced, but saying no difference without using some statistical test on the actual numbers seems naive.
                            Your post is your first post.....This is really, really old news and has been discussed (can we say beat a dead horse) extensively. Pls do a COTH search on "dressage statistics" or some variant of "qualifying rule" and you will get more than you ever wanted for your reading pleasure.

                            (1) - Every time you have a doctor read a mamogram or ultrasound, or build a nuclear reactor, there are humans evaluating "qualitative" tests. There are industrial tools available to the dressage folks which can be used and adapted for dressage to both quantify and reduce "inter-rater variability" (eg., judging variability).

                            This is stuff which has been discussed ad nauseum here. There was no interest in (a) learning about or (b) implementing concepts for statistical robustness of dressage scoring....from USDF/USEF to FEI.

                            (2) - Who said there is "no statistical test?" This proof is easily done. You can prove this for yourself using Excel and the RANDBETWEEN and TTEST functions. You will find that a random distribution between 1 and 10 in integer steps and between 1 and 10 in half-steps has the same mean of 5. The standard deviation varies slightly because of the half steps. A robust statistical test called a 2-sample t-test of the two distributions shows no statistical difference.

                            QED
                            Do not confuse motion and progress. A rocking horse keeps moving but does not make any progress.
                            Alfred A. Montapert

                            Comment


                              #15
                              I have a question/comment about how the judge's mind works.

                              I have two mental models.

                              Under the old "whole number" system, for a movement which is "between a 6 and a 7 - .

                              Model A. I won't give a 7 unless it is truly "Fairly Good". If it is anything less than "Fairly Good", I will give it a 6, even if it is ALMOST a 7. (Truncation)

                              Model B. If it is closer to a 7 than a 6, I will gve it a 7. if it is closer to a 6 I will give it a 6 (Rounding)

                              It seems to me that, if the judge's mental model is A (truncation), then going to half marks WOULD increase the average.

                              If the judge's mental model is B (rounding), then going to half marks would NOT increase the average.
                              Janet

                              chief feeder and mucker for Music, Belle and Tiara. Someone else is now feeding and mucking for Chief and Brain (both foxhunting now). Spy is gone. April 15, 1982 to Jan 10, 2019.

                              Comment


                                #16
                                I'm loving them. The .5 tells me "you're almost there" and lets me know which things I'm really making good ground on improving.
                                "Do what you can't do"

                                Comment


                                  #17
                                  Originally posted by Janet View Post
                                  I have a question/comment about how the judge's mind works.

                                  I have two mental models.

                                  Under the old "whole number" system, for a movement which is "between a 6 and a 7 - .

                                  Model A. I won't give a 7 unless it is truly "Fairly Good". If it is anything less than "Fairly Good", I will give it a 6, even if it is ALMOST a 7. (Truncation)

                                  Model B. If it is closer to a 7 than a 6, I will gve it a 7. if it is closer to a 6 I will give it a 6 (Rounding)

                                  It seems to me that, if the judge's mental model is A (truncation), then going to half marks WOULD increase the average.

                                  If the judge's mental model is B (rounding), then going to half marks would NOT increase the average.
                                  Actually, the whole concept of reducing "inter-rater variability" requires that the "raters" eg., judges, actually be aligned in their interpretations.

                                  This "alignment" is done every day in industry. It allows quality inspectors, engineers, doctors, ultrasound techs, psychologists....etc.....to apply the rules in a consistent basis.

                                  So the two models proposed above would not be permitted. There would be one model....THE MODEL....and everyone would be trained to THE MODEL and expected to follow THE MODEL in application of their ratings...er judging.
                                  Do not confuse motion and progress. A rocking horse keeps moving but does not make any progress.
                                  Alfred A. Montapert

                                  Comment


                                    #18
                                    Originally posted by Badger View Post
                                    I'm really appreciating the half points. I think I'm seeing 6.5s much of the time when before it was more than a 6 but not quite a 7 and the judge's only option was to give me a 6. I think the introduction of half points have improved my overall scores a bit. However, when I scribe and when I look at class results, it does not seem like scores in general are changing any.

                                    What is your experience and impression?
                                    I'm thrilled for exactly this reason. Tons of 6.5s when I would have just gotten tons of 6s.

                                    Although one judge I scribed for last year or earlier said she kept a tally in her head of how many half marks she would have given, and would occasionally throw in a 7 (for example) to make up for it. I don't think most judges did that though.

                                    Comment


                                      #19
                                      FWIW, I do believe that it is working well. I had a test recently with several .5's thrown in and I found it much more informative. I also believe that due to the narrow range of scores employed by most judges, the .5 marks will lead to a much needed improvement in the sensitivity of the scale, much as going from a 4 point scale to an 8 point scale might help. In other words, if most scores range from 5-7, which is essentially a 3 point scale, then they are now on a 5 or even a 7 point scale (4.5-7.5). We needed that differentiation, given that most movements are acceptable or better with few that are close to outstanding. Hope this makes sense - it is late...

                                      Comment


                                        #20
                                        Originally posted by Janet View Post
                                        I have a question/comment about how the judge's mind works.

                                        I have two mental models.

                                        Under the old "whole number" system, for a movement which is "between a 6 and a 7 - .

                                        Model A. I won't give a 7 unless it is truly "Fairly Good". If it is anything less than "Fairly Good", I will give it a 6, even if it is ALMOST a 7. (Truncation)

                                        Model B. If it is closer to a 7 than a 6, I will gve it a 7. if it is closer to a 6 I will give it a 6 (Rounding)

                                        It seems to me that, if the judge's mental model is A (truncation), then going to half marks WOULD increase the average.

                                        If the judge's mental model is B (rounding), then going to half marks would NOT increase the average.
                                        I would assume that judges are trained to use one model over the other, and I also would assume that it is a rounding model.

                                        ??

                                        Comment

                                        Working...
                                        X