Dr. Andrew Barnas
@AndrewBarnas
Followers
9K
Following
13K
Statuses
10K
Postdoctoral Fellow - University of Victoria with @JasonTFisherLab #ACMELab @UVicEnvi. Boreal and Arctic ecology. I like birds, bears, models, and cameras.
Mission, British Columbia
Joined August 2016
Interested in using #drones for wildlife research? Unclear on how much detail your manuscript should provide? Check out: A standardized protocol for reporting methods when using drones for wildlife research - Jof Unmanned Vehicle Systems
4
44
206
Always nice to be mentioned by name in a news article! (Everyone has to call me the statistics from now on)
0
0
4
Biased sample of course, but is there any part of science twitter that is still active here?
0
0
4
If you ever want to annoy your grad school supervisor, you can change their article types to "Ancient Text" in EndNote
0
1
19
@MelanieDickie @taaltree Yes, but you know that is not why I provided the simplified model. It was to demonstrate the noise in predictions introduced by your interaction terms in an under powered global model.
0
0
0
Too much to respond to here, but I encourage interested folks to read all three papers (links at bottom) and draw their own conclusions. But: 1. Data transformations matter 2. I never suggested they use model selection. Only used to demo "noisy" interactions. 3. n = 53, too low.
Recently, we received a response to Dickie et al 2024, with a number of critiques. Including that the effects of data transformation affected our results. They did not. /1
0
0
4
@ClaytonTLamb @taaltree Fortunately it is rainy as hell at my field site, so I am trapped at a computer. Here is the predictive plot from a reduced model without noisy interaction terms (remember n = 53). Same Y-axis.
0
0
0
@taaltree @ClaytonTLamb Not sure on original use of Tweedie, nor the use of min-max scaling of variables. Not specific to this exchange, but we should all justify statistical choices in our methods sections. Different stats decisions lead to different outcomes, which is what I highlight.
1
0
0
@ClaytonTLamb @taaltree Emphasis on "seems to". My concerns on small sample size are not fairly addressed. The reduced model I provide was only to demonstrate the noise introduced by interaction terms. And our focus on the coefficients follows what the authors presented in the original paper.
1
0
1
Author's original paper here: Author's response to our response here:
0
0
1
Hey thats me
Former @iBioUWindsor postdoc Andrew Barnas from @PredictEcoLab explored the potential affects of #ClimateChange on the foraging habits of eider duck predators in the Canadian north. Join us for the full story.
0
0
22
@masonfidino Yeah, tough to do but sometimes warranted. I think a key point is to offer advice on how the shortcomings can be fixed/overcome. So its not just pointing out flaws, but offering constructive advice.
0
0
5
A nice little write up from our latest paper on mammals of the boreal. This paper was a long time coming, so happy to have it out in the world
0
1
12
RT @glukianoff: We can make literally zero progress as long as declaring something conservative is taken to mean “this is wrong or evil & h…
0
80
0
@evornithology Entirely context dependent of course, but to me there is zero expectation of publishing or continued unpaid work by students after graduation. They can write if they want, but the PI might have to pull things together. And he states the student will be first author.
0
0
3