D72R1 mcds engine not modelling multiple covariates

[rjcamp]

Running single covariates the mcds engine works correctly, but get the error ** Warning: Invalid or missing covariate. ** when running two or more covariates in a model.

dst64B7.txt
dst64E8.txt

[LHMarshall]

Investigated the Distance projects and concluded that this is most likely a data issue rather than a software issue.

The project which gave an error in 7.2 also gave an error in 7.1. A project which ran successfully in 7.1 with 2 factor level covariates was tested and still ran successfully in 7.2.

[LHMarshall]

Actually... have found something pretty weird. Which factor covariate goes first appears to affect the success... these are the same analyses I just switched the order of othe two covariates. In Distance one factor is stored as text the other is stored as integer. Data files show that only difference is that the two covariates have been switched in order.

[dill]

Some of the sensitivity is due to some of the covariate classification not quite being right, @rjcamp is following-up on this and we'll see how much this is an issue...

[dill]

Also unclear here if the issue is that ObsYr only has one level in the data (in my understanding of the MCDS command file at least, see below), yet is being included as a covariate but I don't know if that would raise the error Rick reports.

from the command file attached above:

Factor /Name=ObsYr /Levels=1 /Labels=1987;

then later

Estimator /Key=HA /Adjust=CO /NAP=0 /Covariates=Obs, ObsYr;

[lenthomas]

I don't think this is the problem per se as a simple example (below) with just one factor covar with one level runs okay with the warning

** Warning: There is only one level for factor covariate COV. A minimum of two levels is required for estimation; hence this covariate will be omitted. **
dstB47.txt
dstB68.txt