View Single Post
  #19   Report Post  
Old 19-08-2003, 12:32 PM
Torsten Brinch
 
Posts: n/a
Default Animals avoid GM food

On 19 Aug 2003 10:18:51 GMT, Brian Sandle
wrote:

Torsten Brinch wrote:
On 18 Aug 2003 22:44:13 GMT, Brian Sandle
wrote:
[quoting:]
These reports from farmers and seed dealers can easily be dismissed as
anecdotal evidence from which no conclusions can be drawn.


Indeed. And that will remain the case unless anecdotes are followed up
by scientific studies of the matter.


E.g. Shawn S. Donkin, Ph.D. Animal Sciences Department, Purdue
University, fed corn/cornsilage to 16 dairy cows. 8 cows were fed
from Bt corn, and 8 from a near isogenic line of non Bt corn.


Average dry matter intake was 52.7 and 55.9 lb/d, for Bt and non-Bt
respectivley (SE=1.12, p=0.06), average milk yield was 84.2 and 86.9
lb/d, (SE=1.20, p=0.15). Curiously the author concludes that his
results show no differences.


http://www.agry.purdue.edu/ext/forag...y_fed_spec.htm


p=0.15 means that there is 15% chance that the result is spurious.


No,no, you can't conclude like that. But, it is a common
misunderstanding of p-values. p=15 means there is 15 % chance of
observing a spurious difference of the observed magnitude or larger,
-if- the situation is such that any difference observed will be
spurious, i.e. when reality is that there is no difference.

This is really a non-experiment on purpose I would say. The number
of subjects has been kept down to 8 pairs so it is really impossible
to get the p=0.05 needed to conventionally say you have
significance.


I couldn't be harsh as that. But of course, if a relatively few more
animals had been used in the experiment, and the same difference had
been observed in e.g. avg. dry matter intake between groups, that
difference would have been considered significant when tested at the
p=0.05 level.

Once you have set up the experiment I think that would
be the major cost and going to 16 pairs would have allowed the
required significance and not cost much more.


I am not sure that is true. Rather I should think the cost of the
present experiment would be very significantly related to the number
of animals. I mean, these are big animals, cows, they are not fed,
kept and housed cheaply. If they were mice or rats, I would tend to
agree.