Have an idea?

Visit Sawtooth Software Feedback to share your ideas on how we can improve our products.

ACA .acd file

When I generate .acd file for my ACA study, the utilities fo rthe caliberation concept come as zero. This happens for all the respondents. Below is a copy of one record.

8 7
   18    7   27   11   33   42   21    0.00000    0.00000   20
   13    2   26    9   32   39   20    0.00000    0.00000   80
   18    7   27   11   33   39   21    0.00000    0.00000   20
   13    2   25   11   33   37   19    0.00000    0.00000   80
   18    7   27    9   33   39   21    0.00000    0.00000   20
   13    2   26    9   33   37   19    0.00000    0.00000   80
   18    7   27   11   33   42   21    0.00000    0.00000   20
   13    2   26    9   32   39   19    0.00000    0.00000   80
   0.00000    0.00000    0.00000    0.00000
asked May 24, 2012 by mal_saw2001 Bronze (875 points)

1 Answer

0 votes
Back in the "old days" (ACA v4), we used to use these two columns  (those currently showing as 0.00000 in your example above) to store the prior (self-explicated) utilities, and the Pairs utilities (those estimated only using the Pairs data).  Then, ACA v4 used to try to estimate an "optimal" weighting of the two sets of utilities to best predict the calibration concepts.

Now, we do things differently.  And, the only program that uses the .acd file is the ACA/HB estimation module.  The ACA/HB module estimates utilities in a very different way, using hierarchical Bayesian analysis.  But, to make sure ACA/HB software can read both new and very old .acd files, it was necessary to keep the same file structure as before.  Hence, we just pad those columns with arbitrary 0s, because HB isn't going to use that information anyway to estimate the utilities.

So, this is all as expected, and you shouldn't be worried that anything is going wrong.
answered May 24, 2012 by Bryan Orme Platinum Sawtooth Software, Inc. (134,015 points)
...