US deaths in Iraq pass 2000

[Kuciwalker]

Originally posted by Gibsie
Can you recall a time where MOBIUS hasn't been laughed at for the stuff he posts?

I'm asking for the cite on his casualty claim.

[Kuciwalker]

Originally posted by MOBIUS
Well by looking at the US casualty trends from the CNN figures, it seems like that drop in the casualty rate remains a distant pipe-dream...

Given the variance on those numbers I wouldn't trust that trendline.

[MOBIUS]

That's because you don't know how to do trendlines...

[Dinner]

Originally posted by Jon Miller
Mobius.. I agree that the drop in casualty rate is a distance pipe-dream

but that is a BS graph

JM

Have to agree with you Jon.

[Kuciwalker]

Originally posted by MOBIUS
That's because you don't know how to do trendlines...

Yes I do. At least, I assume you're just doing a linear regression - and the r^2 is probably pretty small (bad), looking at the graph.

[MOBIUS]

"Lies, damned lies and statistics..."

Yeah, I was using the linear one because it most supports my point...

[MOBIUS]

Originally posted by Kuciwalker
I'm asking for the cite on his casualty claim.

I would but Iraq threads that aren't archived only go back to the beginning of this year and I seem to be encountering problems searching for them in the archives...

[Kuciwalker]

Originally posted by MOBIUS
"Lies, damned lies and statistics..."

Yeah, I was using the linear one because it most supports my point...

The point is that it's a bad fit because the data is too spread out.

Of course, I don't really see why you would do a different type of regression - I doubt a polynomial equation for it would be particularly meaningful.

[MOBIUS]

Actually it would, it suggests current casualties have levelled off, with a projection that the US should be 'casualty free' in Iraq by March 2009 - so Bush is technically right that the coalition is winning in Iraq...

[MOBIUS]

Originally posted by Kuciwalker
I'm asking for the cite on his casualty claim.

Originally posted by MOBIUS
with a projection that the US should be 'casualty free' in Iraq by March 2009

Now remember, I said it here first...

[Kuciwalker]

Equally spaced data points give absolutely terrible values for polynomial fits outside the range already given.