Issues with a ratio as dependent variable?

Discussion created by jcbrewster on Apr 27, 2018
Latest reply on Apr 27, 2018 by Dan_Patterson

Hi All,

As many of these questions start, I am new to spatial statistics.... I feel like I know just enough to be dangerous, and can't tell if what I want to input into the various Esri tools is legit (OLS, GWR, Moran's I, etc).


I have survey data regarding deformities found in fish in a small alpine watershed. There are a number of variables that I would like to analyze to see if they may be contributing to the deformities, such as elevation, stream depth/width, gradient, discharge, and snowpack.The survey reaches were randomly chosen and have been surveyed over 5 years, and none of the reaches were surveyed twice (a total of 64 reaches were surveyed, so not a big dataset). I have some important categorical attributes as well, like aspect, stream order, habitat condition, and stream bottom substrate, but those can't be tossed into the mix, right?


Given that each survey had a different number of total fish observed, I want to use a ratio (deformed fish to total fish per reach) as my dependent variable. Can I do that legitimately? Even though the explanatory variables aren't ratios?

[I'm using ArcMap 10.6]


Any thoughts would be appreciated. Thanks!