Sample size

questions concerning analysis/theory using program PRESENCE

Sample size

Postby Fiona.Pamplin » Wed Jul 10, 2013 7:00 am

Hi,

I know that this is a contentious topic, but what is the latest view on how to define sample size when you are working with multiple surveys across multiple sites. Sample size /no. of parameters determines whether you use AIC or AICc so that is why I am betwixt and between.

In my case, I have 61 sites that were surveyed 12 times each. My detection histories vary considerably depending on the species (I am looking at 6 different species within the one study).

I have read that it is possible to define sample size as:
1) no. sampling units (sites)
2) no. of detections
3) no. of surveys at occupied sites

Do you select one of the above depending on which gives the best fitting model?

And is it OK then to use different sample sizes for each species ('tho obviously you would use the same sample size for all the models for one species).

Any thoughts appreciated

Thanks
Fiona
Fiona.Pamplin
 
Posts: 3
Joined: Sun Jul 07, 2013 11:14 am

Re: Sample size

Postby darryl » Thu Jul 11, 2013 9:55 pm

You should be consistent in which definition you decide to use. I'm not aware of any updated advice and what I generally suggest is try a couple of definitions and see how much it affects your final inferences. If they're relatively robust, then it doesn't really matter which you choose, but if it does I tend to pick a number somwhere between the extremes of number of sites and total number of surveys arguing is as arbitary as any other value and you're hedging your bets by picking a number in the middle. No claims that's the smartest thing to do though. Others may have some better suggestions.
Cheers
Darryl
darryl
 
Posts: 495
Joined: Thu Jun 12, 2003 3:04 pm
Location: Dunedin, New Zealand

Re: Sample size

Postby Fiona.Pamplin » Fri Jul 12, 2013 8:35 am

Dear Daryl

Thanks so much for the advice - will go for no. of detections as this falls between the two and seems logical in my mind.

Cheers
Fiona
Fiona.Pamplin
 
Posts: 3
Joined: Sun Jul 07, 2013 11:14 am

Re: Sample size

Postby J.Vanek » Thu Jun 01, 2017 1:16 pm

Sorry to bump this, but have there been any advances in this realm? Suggestions/references?

It's my understanding that RPresence uses the number of units based on the help file:

use.aicc

logical argument for whether small-sample adjustment should be used with AIC. Effective sample set to number of units.


Can we change the effective sample size in RPresence at this point in its development?

Thanks,
John
J.Vanek
 
Posts: 7
Joined: Sat Jun 04, 2016 9:23 pm

Re: Sample size

Postby jhines » Thu Jun 01, 2017 9:02 pm

No, not that I know of. I'd stick with Darryl's suggestions (pick something between the two extremes).

In Presence, you can set the effective sample size. The next update of RPresence will allow you to do it as well. In the meantime, you can copy the AIC table to a spreadsheet (or use R function "write.csv"), then re-calculate AICc based on the formula:

aicc=table$AIC+2*k*(k+1)/(n-k-1)

where "n" is the effective sample size, and "k" is the number of parameters.

Jim
jhines
 
Posts: 599
Joined: Fri May 16, 2003 9:24 am
Location: Laurel, MD, USA


Return to analysis help

Who is online

Users browsing this forum: No registered users and 12 guests