Statistical testing to see if the % cover assessed for two periods are different

Section for FAQs for i-Tree Canopy

Moderators: i-Tree Support, i-Tree Team

Post Reply
ayonboy
Posts: 1
Joined: Sat Apr 18, 2020 3:08 pm

Statistical testing to see if the % cover assessed for two periods are different

Post by ayonboy » Sat Apr 18, 2020 3:47 pm

Below is a snap shot of the results from a land cover change analyses (years 2010 and 2019) i conducted in i-Tree canopy recently for a campus.


Cover class...........................Area (ha) ± SE.............................................% Cover ± SE
...................................2010..................2019.....................................2010....................2019
Tree............................3.61 ± 0.23........2.45 ± 0.20...........................13.67 ± 0.89 ......9.27 ± 0.75
Grass - other..............5.56 ± 0.28........5.85 ± 0.28.............................21.07 ± 1.05.......22.13 ± 1.07


For instance, % tree cover decreased from 13.67 ± 0.89 (in year 2010) to 9.27 ± 0.75 (in year 2019). What statistical test can i perform to know if the decrease in % tree cover is statistically significant.
Jason.Henning
i-Tree Team
Posts: 340
Joined: Fri Dec 21, 2012 7:45 am

Re: Statistical testing to see if the % cover assessed for two periods are different

Post by Jason.Henning » Mon Apr 20, 2020 8:35 am

This is a great question. The Forest Service is working on a spreadsheet that will help users with these types of comparisons but it likely won't be posted on the i-Tree website until sometime this summer. The statistical test the the US Forest Service uses for these types of comparisons in the McNemar test (https://en.wikipedia.org/wiki/McNemar%27s_test. You can find an Example of the US Forest Service use here, https://www.itreetools.org/documents/55 ... nfield.pdf. If you do a Google Search for "McNemar test" you should be able to find some web applications that can perform that statistical test when you provide a contingency table. The test can also be performed in Excel if you understand how to use pivot tables to create a contingency table and the statistical distributions functions to generate p-values.
I will update this post when we post the spreadsheet.
Thanks,
Jason
A member of the i-Tree Team
Post Reply