Warning message ** Warning: Parameters are being constrained to obtain monotonicity. **

[Bryony Tolhurst]

Dear Eric et al

I am a Distance sampling beginner, and am trying to run a basic CDS analysis on a set of survey data without post-stratification. I have 6 study sites surveyed using point counts with 10 points per site repeated three times each. The size of the study sites vary and I have put this variable in a column entitled 'Block size'. There are 662 observations, each of single dogs (i.e. no clusters). I followed the instructions in the Distance 7.3 user guide page 25 under the section Example 2: More Complex Data Import. My data follow an identical structure to the example given - House Wren, with the one difference that only one observer collected data (hence I did not post-stratify). The analysis runs but the output gives me the error message ' Parameters are being constrained to obtain monotonicity' and has an orange ball instead of a green one when the analysis is computed. I assume this means the reliability of the output is compromised? I actually only need to calculate a strip width in order to retrospectively truncate my data (I am not particularly interested in number of dogs per se but more detectability over specified distances in the day versus the night). 

Are you able to a) confirm that I do need to reanalyse my data (i.e. can't ignore the error message!) or b) advise as to what might be the problem.

Many thanks in advance,

Bryony

[Eric Rexstad]

Bryony

Welcome to the list.

Your interpretation of the warning message is approximately correct.  The software is confronted with two conflicting instructions: fit the data the best you can and ensure the function you fit to the data such that the fitted detection function is at its highest at distance 0 does not increase anywhere as a function of increasing distance.

The message arises because Distance cannot do both.  Often this arises because the histogram of observed distances tails off at large distances, but there may be a couple of detections at very large distances.  The software would like to fit a function that declines, but then increases to fit the pod of distant detections.

The customary way of getting rid of the warning is, as you would guess, truncate the data to exclude those most distant detections; they usually carry little information about the shape of the detection function in the region where shape of the function is most critical.

I hope that answers your question about the warning message.

--
You received this message because you are subscribed to the Google Groups "distance-sampling" group.
To unsubscribe from this group and stop receiving emails from it, send an email to distance-sampl...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/distance-sampling/3cc1794c-6ba6-4700-8721-df657e5079a5n%40googlegroups.com.

-- 
Eric Rexstad
Centre for Ecological and Environmental Modelling
University of St Andrews
St Andrews is a charity registered in Scotland SC013532

[Bryony Tolhurst]

Dear Eric

 

Many thanks for your incredibly prompt message. That makes sense  - thank you so much. I would be interested to see whether the EDR (which from the analysis I ran was around 80m) changes appreciably after I have removed outlying locations.

 

Thanks again,

 

Bryony

---

CAUTION: This email may have originated from outside of the university. Do not click links or open attachments unless you recognise the sender and know the content is safe.