At the risk of sounding like a creaky wheel, I will repeat something I wrote
a year or so ago when a similar topic came up - I think that ecologists
worry too much about high-powered statistics and too little about the
fundamentals. For example, almost none seem to understand propagation of
error, even though this is the starting point for any data analysis. Most do
not realise that statistics is based on models and picking the correct
statistical approach involves picking a suitable model - you cannot just do
a linear regression and say (I have often heard this) "I don't need a model,
I used statistics".
I recently found a box containing some of my undergraduate work, including
physics lab reports going back to 1954, and I was impressed by how good my
data analysis was back then compared to what I have to put up with when
reviewing papers in ecology and marine biology. All the stats I had learned
then was from a little pamphlet by Yardley Beers called (I think) "Theory of
Error". I wish that could be worked into an ecology course!
Bill Silvert
----- Original Message -----
From: "Nathan Brouwer" <[email protected]>
To: <[email protected]>
Sent: sexta-feira, 30 de Abril de 2010 16:53
Subject: [ECOLOG-L] stats for undergrads - what about Gotelli?
An assortment of comments and questions regarding stats for undergrads:
