Hi,
I'm referring to the last email on ztest. I'm trying to do
the same, by comparing the density estimates of several birds (multiple
species) at two different levels of disturbance. I used CDS with
poststratification based on disturbance level and density was
estimated for each species. In the distance book (Buckland et al 2001)
at page 353, there is the z test calculation. The example was from
fin whale survey and the z test tested the differences of mean size of
two different pods (cluster size). Would that be a problem if I use
the same calculation to compare the density estimates at two
disturbance levels?
Thanks.
Best,
Nurul
Nurul Winarni
Email1. [log in to unmask]
Email2. [log in to unmask]
http://noonathome.wordpress.com/
http://wildlifewisdom.wordpress.com
 Original Message 
From: Jeff Laake <[log in to unmask]>
To: [log in to unmask]
Sent: Monday, June 23, 2008 11:06:58 PM
Subject: Re: [DISTSAMP] Ztest for density
Sergio
You are confusing two different types of tests. The one you describe is
testing whether a sample mean is different from a population mean or
whether it is different from a hypothesized value (eg. 0). The other is
testing whether two sample means differ from each other and this is the
more common. The first is a one sample test and the second is a 2sample
test. There are many different types of 2 sample tests. They can be
paired or nonpaired samples and nonpaired tests can assume a common
variance or different variances. The test we have been discussing is
nonpaired with different variances. Also, note that the formula you
gave for the one sample test was incorrect because the denominator
should have been the standard error of the mean and what you wrote was
the sample variance/ sqrt (n). It should have been sqrt(sample
variance/n) which is the standard error of the mean. On this point, the
standard errors for density (from DISTANCE) are equivalent to the
standard error of the mean and they are not standard deviations, so you
do NOT divide the std error by the sqrt of n. This often confuses
folks. For parameters (eg density) we use standard errors and not
standard deviations. A standard deviation describes the spread in the
data and a standard error is the uncertainty about a parameter (eg
mean). The confusion comes from the fact that the standard error for a
sample mean is the standard deviation divided by the sqrt of n. That
does not apply to all parameters and there is typically no relationship
between the standard deviation (of the data) and the standard error of a
parameter, except for a mean.
With regards to the contents of Snedecor and Cochran, my reference to it
was a wry joke at myself because it is so old. Any basic stats text
should give the formula for the hypothesis test we described. Len
Thomas replied to me individually that my book (used) can be bought for
$4.80 at
http://www.abebooks.com/. Note that I'm not recommending either this
site or book.
On a related subject, for what you want to do you are probably less
interested in hypothesis testing than you are in estimating differences
in densities between various guild. In the ecology and other fields
there is a move away from hypothesis testing towards model selection.
David Anderson or others may want to add more here. Unfortunately, the
current DISTANCE software does not provide a lot of options (at present)
to represent the abundance data by models. We have models for detection
probability but the estimates of density are based on sampling concepts
and are not model based for most of the analysis techniques in
DISTANCE. I believe in one of the past list server messages Steve
Buckland mentioned a paper that he and others have put together on
analysis of experimental distance sampling data, so he may have more to
contribute here.
regards jeff
SERGIO . wrote:
> Dear All,
>
> The equation: D^_1  D^_2 Z =  sqrt[{se(D^_1)}**2 + {se(D^_2)}**2]
>
> that Claire Meid referes to in his email, it is not the classical ztest equation to compare a sample mean with the population mean. How can i cited this formula to compare density between guilds and/or habitats and how can i consider this a validated equation (aproximation of ztest?) for my bird census analysis?
>
> The original formula at statistics books and statistics tests is this one:
>
> x  Uo
> Z = 
> var / sqrt [N]
>
> where x is a sample mean, Uo the population mean, var is the variance of the population and N is the sample size.
>
> In the density case the N could be always the same so its not consider and i can use densities from 2 samples to be compared, but the acceptance range could be always the same for ztest with this modifications?  1,96<Z<1.96 for two tailed test (for alfa = 0.05).
> Thanks to all,
>
> Best Regards,
>
> Sergio Nolazco Plasier
> Bach. Biology
> LimaPeru
>
>
>
> _________________________________________________________________
> Express yourself instantly with MSN Messenger! Download today it's FREE!
> http://messenger.msn.clickurl.com/go/onm00200471ave/direct/01/
> 
> distancesampling list home page:
> http://www.jiscmail.ac.uk/lists/distancesampling.html
> To unsubscribe, go to the home page and click "Join or leave the list",
> or email [log in to unmask] with the following message body:
> leave distancesampling
> Before posting to the list, read the list guidelines at
> http://www.jiscmail.ac.uk/files/distancesampling/
> distancesampling.welcome
> Any questions, email the list owner at
> [log in to unmask]
> 
>

distancesampling list home page:
http://www.jiscmail.ac.uk/lists/distancesampling.html
To unsubscribe, go to the home page and click "Join or leave the list",
or email [log in to unmask] with the following message body:
leave distancesampling
Before posting to the list, read the list guidelines at
http://www.jiscmail.ac.uk/files/distancesampling/
distancesampling.welcome
Any questions, email the list owner at
[log in to unmask]


distancesampling list home page:
http://www.jiscmail.ac.uk/lists/distancesampling.html
To unsubscribe, go to the home page and click "Join or leave the list",
or email [log in to unmask] with the following message body:
leave distancesampling
Before posting to the list, read the list guidelines at
http://www.jiscmail.ac.uk/files/distancesampling/
distancesampling.welcome
Any questions, email the list owner at
[log in to unmask]

