Summing Covariates Contribution

[elizabeth_10]

Hey all,

I read the MacKenzie book as well as some journals regarding the covariates effect. According to the journal as well as the book, it says that to study the contributions of each covariate, it is done by summing the Akaike weights for all models in the set containing the covariate in question.

So I did the same, taking the weights given in the journal, I tried summing up the Akaike wgt that have the covariate in question, but the value that I obtained is different from the author of the journal calculated.

So I looked into the MacKenzie book, I tried summing again. But I just couldn't get the same values.

Did I miss out any formula in summing up?

I've been pondering for long, pls help to clear my doubts. Thank you in advance.

Regards,
eliz.

[darryl]

This is a way of looking at the overall importance of particular factors. Which example in particular were you looking at in the MacKenzie book?

Cheers
Darryl

[elizabeth_10]

Hey Darryl,

I was referring to the Mahoenui Giant Weta in Single-species, Single-season Occupancy Models (pg. 116-119).

For p(Day), I managed to sum up to 91% based on the Model Selection table, however for p(Obs) and Browse, it is written 73% and 33% respectively, but when I calculate manually, I obtained 72% and 32% only. Same goes to psi(Browse), it is written 58%, but I obtained 57% only.

Did I miss out some formula or theory?

Regards,
eliz.

[darryl]

This just looks to be rounding error to me. The values in the text were probably summed directly from a spreadsheet with more decimal places than the 2 given in the table. Will clarify in the next edition, sorry for any confusion.

[elizabeth_10]

Alright then. Thanks for your reply.
Regards,
eliz.