Monday, November 17, 2014

The old man and the wolves

 Reading this: http://www.startribune.com/sports/outdoors/282833431.html, and inside my brain statistical hairs start standing on end, bells go off, and finally, a plague of mathematical modelling locusts descends and places the land into darkness. I read the original article with supplements here: http://pubs.er.usgs.gov/publication/70122639.

Rolling my R's...It is Crrrap. A 90% confidence interval for population estimates varying by up 5000 individual moose using two different survey methods the authors admit are not compatible, then evaluating variables using a "simple simulation study" for comparing the variables by what is experimental computation? Let me say this, I don't know who the reviewers were, but if I tried to pass off these methods, I should be laughed out of the room. When you have those kinds of error estimates with this sort of modest population size, then start using assumptive linear models to get p-values out of this, it is garbage. I The only way this was published was by his old name and employer. Before a reporter even dares start putting this opinion up, he needs a second opinion.

No comments: