I have an ACBC study with 265 completes. All 265 completes show up when I run counts, but 9 are missing in the utility estimation. I initially thought I messed something up with my filters, I also ran the analysis for all respondents (including incompletes). The same respondents are missing in both utility runs.
I checked the individual records in the counts summary and the missing respondents made it all the way through the tournament and calibrations, so I'm not sure why they are missing from the utilities.
(note: the missing respondents are distributed throughout the sample i.e. they are not all in one spot / there are good respondents before and after)