Tiebreakers

[jaggie34]

I was recently an ES/test-writer for a tournament that had a large number of ties in the standings (14 to be precise). The exam was ~60 questions with 5 questions specified ahead of time to be tiebreakers, however, due to the high number of ties, in a few instances, the tied teams had the same results on the 5 tiebreaker questions. I ended up checking time turned in as the next tiebreaker, but since all tied teams used the full time and were timed out of the test, I settled on using questions attempted as the final tiebreaker in these situations. I'm curious if anyone else was in this situation, what would you have used as the tiebreakers?

[gz839918]

It's nearly impossible for teams to tie in Sounds of Music thanks to musical instrument sections, but my tiebreaker procedure is to designate a few questions as tiebreakers, and for each tiebreaker, I assign a certain number of tiebreaker points. The tiebreaker points don't affect a team's score unless there is a tie. If team C01 has 10 normal points and 2 tiebreaker points and team C02 has 10 normal points and 3 tiebreaker points, then team C02 wins the tiebreaker. (Alternatively, each tiebreaker point may be thought of as 0.01 normal points, so C01 would have 10.02 normal points and C02 would have 10.03 normal points, which is how I'm doing tiebreakers in Scilympiad.)

I weight the tiebreaker points so that the harder tiebreaker questions are worth more tiebreaker points. My tiebreaker questions are often open-ended, and I forgo a rubric so that I can break ties based on small nuances in the quality of a student's response. For normal questions, all teams who reach a certain threshold get the same amount of points, but for tiebreakers, I can award different amounts of tiebreaker points for the team that just barely has the right answer compared to the team that goes above and beyond in their response.

For teams who usually leave questions blank, including tiebreaker questions, I usually have a silly tiebreaker so that I can break ties from them if absolutely necessary. Mostly that's just to make people laugh though.

[Unome]

I generally tiebreak by comparing all questions in order. I used to select specific questions but I've found that selecting questions in advance to break ties often ends up not working too well because it's really hard to pick good tiebreaker questions - since they have to separate teams with the same score, and also work across all score ranges, it's quite a challenge to distribute them in such a way as to be effective.

[BrownieInMotion]

For events with stations/slides, I calculate the team's subtotal on each section, then compare each team's highest section score against each other to break the tie. I do this to avoid having a team's ranking affected by a single question, as well as rewarding teams who have deeper knowledge.

I do a similar thing with test events. I'll aggregate scores from FRQs or a section then compare those. I'm not sure about the format of your exam, but I probably would've picked some set of questions to aggregate scores for and compare, though that may not have worked in your case.

In general, it would be good practice to write your exams and design your rubric so as to avoid this kind of situation in the future. Writing more questions, writing questions that span a wide range of difficulties, and designing problems/questions that allow for a wide range of partial credit options are all good ways to do that.

[ScoutViolet]

I agree with @BrownieInMotion, you could value some sections higher. Like for water quality, the classification of animals could be lower point values because well you just need to know the picture. For something like Ornithology, I would have a hard time... Maybe you could designate the birds that are rarer and/or more similar to other birds (like, Common Cactus Finch and Medium Ground Finch, though I am pretty sure those aren't going to be tested on). I'm obviously not going to make a whole list of these, but I think you should value sections at different points.

[MorningCoffee]

I agree that point values should be distributed according to difficulty.

Side note though: please do not test birds that are not on the list (an exception might be when in relation to birds already on the list). Haven't seen too many issues with that, but just throwing it out there.

(While I'm at it, please please please check the sources of your images. Allaboutbirds has an amazing Compare Similar Species feature; unfortunately, this means just googling "chipping sparrow" can pull up American Tree Sparrow images because they were found on a webpage with "Chipping Sparrow" in the title; I have seen this on tests before)