Negative standard errors

questions concerning analysis/theory using program PRESENCE

Negative standard errors

Postby sloeb » Thu Feb 04, 2016 6:42 pm

I am running a single-season model. When I add an interaction term between Site (a categorical variable) and a standardized continuous variable I get negative standard errors with some of my estimates. I also get a warning that there are negative standard errors in the VC matrix. Any suggestions on why this may be happening and how I may correct it? Many thanks.
sloeb
 
Posts: 1
Joined: Thu Feb 04, 2016 1:17 pm

Re: Negative standard errors

Postby Lea » Tue Feb 23, 2016 11:40 am

I am pretty sure I have come across that at some point as well. I think I tried an alternate parameterization of the model and it fixed the problem but I was also running multi-season models. The only thing I could find in previous posts that might be helpful was this thread viewtopic.php?f=1&t=1102&p=2957&hilit=+vc+matrix+#p2957

sorry this is not more helpful. hopefully someone can better address your question.
Lea
Lea
 
Posts: 33
Joined: Thu Oct 24, 2013 1:09 pm

Re: Negative standard errors

Postby Lea » Tue Feb 23, 2016 1:20 pm

Oh and i came across this:

Negative values in VC matrix may indicate:
A not positive definite input covariance matrix may signal a perfect linear dependency of one variable on another. In those cases, sequential analysis of the covariance matrix, adding one variable at a time and computing the determinant, should help to isolate the problem. Starting Values- The model-implied matrix Sigma is computed from the model's parameter estimates. Especially before iterations begin, those estimates may be such that Sigma is not positive definite. Then it is up to the researcher to supply likely starting values. Missing Data-Large amounts of missing data can lead to a covariance or correlation matrix not positive definite. My Variable is a Constant! Sometimes, either through an error reading data or through the process of deleting cases that include missing data, it happens that some variable in a data set takes on only a single value. In other words, one of the variables is actually a constant. This variable will then have zero variance, and the covariance matrix will be not positive definite. Simple tabulation of the data will provide a forewarning of this. If this is the problem, either the researcher must choose a different missing-data strategy, or else the variable must be deleted.
Lea
 
Posts: 33
Joined: Thu Oct 24, 2013 1:09 pm

Re: Negative standard errors

Postby darryl » Tue Feb 23, 2016 4:04 pm

This can also happen if you start getting boundary estimates; estimated probabilities that are either 0 or 1. Is that the case?
Darryl
darryl
 
Posts: 495
Joined: Thu Jun 12, 2003 3:04 pm
Location: Dunedin, New Zealand


Return to analysis help

Who is online

Users browsing this forum: No registered users and 12 guests

cron