# Back transforming coefficients from a deltal lognormal GLM

I'm fitting a delta lognormal GLM to catch rate date that has many zero observations using the R function deltaLN (package FishMOD). Catch rate is my independent variable, while my dependent variables are the categorical variables year, month, and area. Specifically, I'm interested in extracting the year effect as my "standardized" catch rate. This model is particularly useful for my type of data because is contains fits a binomial component that uses a logistic link function (capturing presence/absence) simultaneously with a lognormal component (catch rate conditional on presence).

The model form in R is: catch.model <- deltaLN(ln.form = catch.rate ~ year + month + area, binary.form = ~ year + month + area, data = catch.rates.dat)

I'm running into problems trying to back-transform the coefficients to obtain the year effect, however. The model allows you to extract the lognormal and binomial coefficients for each variable, which I have back transformed as follows:

    binom.coef <- inv.logit(catch.model$coefs$binary[2:38])
ln.coef <-   exp(model.N.north$coefs$ln[2:38])
year.effect <- binom.coef * ln.coef


Where the coefficients for the years are contained in position 2:38 in both of the coefficient lists. I really thought that this was the correct back-transformation, but the units appear to be very far off of what was expected. Does this look correct?