search

maxbiostat / FMDV_AMERICA

Tracking the spread of Foot-and-Mouth Disease virus in South America
1 stars 0 forks source link

R2MC11 #95

Closed maxbiostat closed 4 years ago

maxbiostat commented 9 years ago

Page 5 Lines 34-38 and following paragraph (Lines 41-57): “For serotype A…”. This sentence is really confusing and need to be better formulated. For type A you find that geographical distance drives the diffusion, whilst this has a higher statistical support for type O but not like the cattle exchange. Now, the question is, what cattle exchange means? This implies geographic distance as well (because you are defining trade between countries, which in South America are not so very close), isn’t it? So, the geographic distance is the main effect of FMDV diffusion or a confounding effect for cattle trade? I am really struggling to find a logic behind this results (or its analytical approach) considering that you have a strong bias in your data (both spatially and temporally) and you are analysing the geographical distance and trade (both cattle and swine) variables separately…have you checked for multicollinearity?

maxbiostat commented 9 years ago

This is a really good point. Our approach (testing the marginal likelihood of each predictor individually) may be rather inefficient and also fail to account for correlation. There are ways round it, though.