Skip to content

Simulates judging at a DECA Conference to determine whether normalizing scores would increase or decrease accuracy. Can be used for competitions with similar biased judging systems.

License

Notifications You must be signed in to change notification settings

Aditya-Singhvi/DECA-judging

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 

Repository files navigation

DECA Judging Simulator

The DECA Judging Simulator simulates scores in a DECA conference and analyzes the impact of normalizing scores using two methods. Calculates the threshold for judge bias where not normalizing yields better results than normalizing. Assumes that scores follow a normal distribution.

Parameters

relevant_scores: how many scores do you care about? (top 4 qualify, for example).

trials: how many trials to calculate the average error? Balance speed with accuracy.

judges: how many judges do you have?

kids_per_judge: how many kids will each judge evaluate?

avg_score: The average true score on a roleplay

st_score: The standard deviation in participant ability (of the true scores)

min_bias_tested: The minimum judge bias to be tested in the simulation

max_bias_tested: The maximum judge bias to be tested in the simulation

How well trained are your judges? These refer to the standard deviation for a single judged score. For instance, if 95% of your judges would judge within 14 points of a 70 (true score) role-play, your standard deviation would be: (14 / 2) / 70 = 0.10. From 2020 DECA Launch Judging Data, this value seems to be around 0.13 in the real world.

Usage

Inspired by debate over whether to normalize scores at DECA Launch 2020 at The Harker School. Can be used to simulate any similar competition-based system that includes biased judging and potential normalization.

About

Simulates judging at a DECA Conference to determine whether normalizing scores would increase or decrease accuracy. Can be used for competitions with similar biased judging systems.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages