Windows kill a Billion (with a B!?!) birds annually? Probably not.
I just came across this article with the following headline:
As many as 988 million birds die annually in window collisions.
Wait, what? Almost a BILLION (with a “B”) birds are killed every year by flying into windows? That’s an outrage! Let’s tear down all of our building and get rid of all our windows in defense of our feathers friends!
No wait. Actually, let’s take a step back and see what’s happening here. First, where does the 988 million number come from? It comes from this study: Bird–building collisions in the United States: Estimates of annual mortality and species vulnerability in some journal called “The Condor”. And actually what they say in their abstract is:
Based on 23 studies, we estimate that between 365 and 988 million birds (median = 599 million) are killed annually by building collisions in the U.S.
So 988 million (nearly a BILLION!) is the upper threshhold of their interval estimate of annual bird deaths by window collisions. But why did the Washington Post leave out the interval estimate and only mention the upper bound? I’d guess because it’s a more sensational headline. (If I’m wrong, please let me know what the reason is.) C0me on media, you’re better than this.
So just to be clear, what the studied actually “showed” is that they estimate that somewhere between 365 and 988 million birds with a median of 599 million are killed in window collisions annually. Even if their methods are completely sound their median estimate is about 40% lower that the sensational upper limit. (Which I suppose would translate into 40% fewer clicks.) So even if they are correct, I feel like the media is being sensationalist about their “findings”.
BUT! BUT! I suspect strongly that their methods are not only not completely sound, I suspect their methods are barely acceptable. In fact, I’ve actually reviewed the statistical methods employed by this same exact author, Dr. Scott R. Loss, before for a similar study that claimed that cats were killing incredibly large numbers of birds. My complete report can be viewed here. (Full disclosure: I was a paid consultant for Alley Cat Allies when I wrote that; They are not paying me for this blog post.) So, Dr. Loss doesn’t have the best statistical track record in my opinion, though he may be a fine ornithologist (he was, after all, the Outstanding Conservation Biology Student of 2009-10. Congrats!). In my professional opinion, as I have stated before, the methods employed in the article “The impact of free-ranging domestic cats on wildlife of the United States” have almost no statistical validity.
While I haven’t fully reviewed the methods in the new paper about bird collisions, I suspect that the same or very similar methods have been employed to reach these astronomically sensational numbers. Again, as I have said before, these numbers may be completely correct (though I suspect they are not), but the statistical methods used to arrive at these numbers are extremely shoddy.
And guess what?!?! Dr. Loss looks like he has a whole series of these papers coming out! Check out his CV under submitted papers. Vehicle collisions, power lines, wind farms!
So Dr. Loss if you’re reading this, please PLEASE consult with a statistician. Also, if you’d like to respond to my criticism, I’d be happy to post your unedited response on my blog.
P.S. By the way, one BILLION (that’s 1,000,000,000) birds per year would mean that almost 32 birds PER SECOND were dying on average in every second, day and night, of every day all year round JUST IN THE UNITED STATES!
Posted on February 6, 2014, in Uncategorized. Bookmark the permalink. 1 Comment.
Pingback: The media hates cats and statistical rigor | Stats in the Wild