This event has ended. Create your own event → Check it out
This event has ended. Create your own

The 2017 Seventeenth Annual UMM Undergraduate Research Symposium (URS) celebrates student scholarly achievement and creative activities. Students from all disciplines participate in the URS. Types of presentations include posters, oral presentations, and short or abbreviated theatrical, dance, or musical performances. 

Presentations are accompanied by discussions and multimedia.


View analytic
Saturday, April 22 • 10:00am - 12:00pm
Using Correlation of Order Statistics to Discern Directional Dependence

Sign up or log in to save this to your schedule and see who's attending!

Pearson’s correlation is used throughout the scientific community to measure how strongly two variables are related to each other. However, Pearson’s correlation can miss relationships between variables if the relationship is too complex, and when it does it doesn’t give information about how two variables are related. To examine these types of problems, we investigated a new kind of correlation based on ordering variables. We created it by ordering pairs of variables from smallest to largest based on the first variable, then from smallest to largest based on the second variable, and finding Pearson’s correlation between a variable when it is ordered from smallest to largest and the same variable when it is ordered based on the second variable. We examined if and when these new correlations differed from Pearson’s correlation to determine if the new correlations could reveal underlying trends in the data that Pearson’s correlation didn’t and if the new correlation could correctly predict the direction of the relationship between two variables. We tested our method using real world data collected from a variety of laboratory and field experiments where there was a known causal relationship and data that we generated using statistical computing software. We found that our new correlation showed some aptitude for discerning relationships that Pearson’s correlation missed in certain types of data. Our method had a low success rate with correctly predicting the direction of a relationship in the data we generated; however, our method outperformed existing methods when applied to real world data.

Saturday April 22, 2017 10:00am - 12:00pm
Student Center, Oyate Hall 600 E 4th St., Morris MN 56267