Blocking

This forum is for posts that specifically focus on Ngene.

Moderators: Andrew Collins, Michiel Bliemer, johnr

Blocking

Postby Benjamin » Fri Sep 30, 2016 5:57 pm

Hi!

When using an efficient design, with only signs for parameter (i.e. [0.000001])

Is it possible (necessary) to block? Want to maximum give 4 choice sets to each respondent, but need at least 13 choice sets. Block in 4 from 16 ?
Or is blocking only from orthogonal?
Benjamin
 
Posts: 13
Joined: Tue Sep 27, 2016 9:04 pm

Re: Blocking

Postby Michiel Bliemer » Fri Sep 30, 2016 6:12 pm

True blocking is only possible in orthogonal designs, HOWEVER, we have implemented a heuristic in Ngene that lets you blick any design. Ngene aims to minimise correlations between attribute levels within a block, which results in blocks that have a high degree of attribute level balance. So you can use blocking also with an efficient design in Ngene, and indeed block a design of 16 into 4.
Michiel Bliemer
 
Posts: 1705
Joined: Tue Mar 31, 2009 4:13 pm

Re: Blocking

Postby Benjamin » Fri Sep 30, 2016 6:21 pm

Thanks!

When only using +/- for priors, what would be best in Ngene, orthogonal 16/4 or efficient design 16/4, or both (i.e. orth=seq + eff=(mnl,d)) ?
Benjamin
 
Posts: 13
Joined: Tue Sep 27, 2016 9:04 pm

Re: Blocking

Postby Michiel Bliemer » Fri Sep 30, 2016 6:29 pm

Efficient, since this allows Ngene to remove dominant alternatives by setting ;alts = alt1*, alt2*. If you do not want to check for dominant alternatives then both orth = seq or eff is good in case your priors are (near) zero.
Michiel Bliemer
 
Posts: 1705
Joined: Tue Mar 31, 2009 4:13 pm


Return to Choice experiments - Ngene

Who is online

Users browsing this forum: Google [Bot] and 11 guests

cron