1. Evaluating the Effectiveness of
Behavior Monitoring Applications
in the Red Panda
Ashley M. Fortner & Elizabeth W. Freeman
School of Integrative Studies, George Mason University
14. Statistics & Validation
Within-observer reliability: Tests my ability to get the same results when scoring
the same video multiple times.. This value is expressed as a correlation coefficient
(r): +1.0 = high linear association, 0 = no linear association.
Kendall coefficient of concordance: Will be used to test the time in nest box
between the previous student Brittany, myself, and the new Ruby for Good
application.
Inter-method reliability: Will be used to test the results gotten from the hand
scoring and the results from the Ruby for Good Application.
16. Kendall Coefficient of Concordance (W)
H0: There is no agreement amongst the three
judges.
H1: There is agreement amongst the three
judges.
k = 6
m = 3
W = 0.8
r = 0.7
χ2 = 12
df = 5
p-value = 0.03478778
α = 0.05
α > p-value = Reject H0
There is significant evidence showing that there
is agreement amongst the judges.
18. Conclusions & Recommendations:
Based on our correlation coefficients, the
application appears to work successfully!
HOWEVER…...
It seems to work best when nest box
movement is spread out.
Need a better method of data exportation
from the application.
Data analysis would be expedited with the
ability to calculate total time in nest box.
19. Importance:
BEHAVIOR ACTIVITY PROFILES:
More likely to know if something is “out
of character”.
Will notice these things faster → can
react quicker → SAVES LIVES!
RUBY FOR GOOD APPLICATION:
Speeds up data analysis process.
Allows researchers to concentrate
efforts elsewhere.
20. Project Continuation
CURRENT BEHAVIOR MONITORING APPLICATION:
Behaviour Pro
• Lacks the ability to pause during observation.
• Lacks the ability to make notes.
• Lacks the ability to edit data.
• Unable to sync up video times with application.
NEW BEHAVIOR MONITORING APPLICATION:
• Will be upgrading to better fit researcher’s
needs.
• Cosmetic fixes (button size, screen orientation).
21. Acknowledgements:
My Mentor: Dr. Elizabeth Freeman, School of Integrative Studies
Sean Marcia: Founder of Ruby for Good & Creator of the application tested during
this project
The Biology Undergraduate Research Semester: Provided generous funding and
classes to help develop this project.
The Undergraduate Research Scholars Program (OSCAR): Provided generous
funding and introduced me to many resources.
Lastly, I want to thank all of YOU for being here today and supporting the red
panda!
Name: The Red Panda
Scientific Name: Ailurus fulgens
Habitat” Himalayan Cloud Forest ~ Less than 20C
Endangered: Due to Deforestation
Importance: Protects other species as well (Black Gibbon)
Reproductive Problems: Poor motherly care, poor milk production, maternal cannibalism
GAG, GS, HS, Lay, Move, Nest, NV, Other, Scratch, Sit, Stand
.7-.9 = high correlation, .9-1.0 = very high correlation
Taking observations every 10 seconds. When it analyzes the images and determines time on video it doesn't begin recording until it get 3 consecutive positives and it doesn't stop until it records 3 consecutive negatives.
This could lead to some inaccurate results if you don't know what to look for. For example, imagine a panda appears on video at the 3:00 minute mark and then at 5:30 the panda leaves for 10 seconds. Since it was only 10 seconds that the panda was off video it would only be 1 negative observation in a bunch of yes observations so it would change that to a yes.