Product/Interaction Specification in Known Fate DM

[bacollier]

All,
I suspect I am missing something obvious, but I have been trying to insert a interaction term into the DM for known fate type analysis using FillCol-Product and product(X, X) and I always get a column of zeros and the same AIC and K as the initial main effects model, as well as non-sensical real parameter estimates. After extensive searching in the MARKBOOK and help files, I have hit a wall. Could someone point me in the right direction?

As a mini-example I have a DM with an intercept a 2 Period Time Trend (breeding and non-breeding) and Sex as follows:

B_0 B_1 B_2
1 1 Sex
1 2 Sex
1 2 Sex
1 2 Sex
1 2 Sex
1 1 Sex
.
.
. as above for 30 months.

But, I have not been able to figure out how to build the interaction B_1*B_2 (Breeding period * Sex) to be included as a single column (should it be a single column) in the known fate approach. Do I have to go back and insert an time dependent covariate for the trend and time dependent interaction covariate in the .inp file? I figure I am missing the obvious, so a gentle nudge (or shove) would be appreciated.

TIA,
Bret

[gwhite]

Bret:
You have to enter a column with each entry in the interaction column a product function. Thus,
B_0 B_1 B_2 B_3
1 1 Sex product(1,Sex)
1 2 Sex product(2,Sex)
1 2 Sex product(2,Sex)
1 2 Sex product(2,Sex)
1 2 Sex product(2,Sex)
1 1 Sex product(1,Sex)

Alternatively, you can enter
B_0 B_1 B_2 B_3
1 1 Sex product(col2,Sex)
1 2 Sex product(col2,Sex)
1 2 Sex product(col2,Sex)
1 2 Sex product(col2,Sex)
1 2 Sex product(col2,Sex)
1 1 Sex product(col2,Sex)

See the MARK help file, under 'Design Matrix Functions'.

Gary

[bacollier]

Gary,
Thanks for the nudge, I missed it.

Bret

[cooch]

<snip>

See the MARK help file, under 'Design Matrix Functions'.

Gary

Time for me to add this to the individual covariates chapter. Consider me similarly 'nudged'.