I too hate "lies, damn lies and statistics", but I didn't want to overload the original post with too much text.
As to details on the method - I used the same basic approach as ATSB. I was focusing on RPTs, because that is where the greatest concern lies. Clark and I looked at relative rates of airspace-related occurrences and airmisses for RPT aircraft in CTAFs and MBZs. The analysis was "all CTAFs in Australia that have had RPT movements in the last 8 years" and "all MBZs that have had RPT movements in the last 8 years". The lot. No carefully 'chosen sites'. No selective memories. We got the ATSB recorded airspace-related occurrence and airmiss data off their database (and well done to ATSB for maintaining this database).
Putting it simply, there are more RPT movements in MBZs than CTAFs in Australia, because MBZs tend to be used at busier airports and at many airports carrying RPT traffic (and ATSB mentioned this as well in their report). This was a concern to me when I first woke up to the issue because if an MBZ is busier than a CTAF, then it could easily have more airspace-related occurrences (or airmisses) without being more risky.
In practice, we found that almost all the RPT traffic in and out of aerodromes happens in MBZs. If there were 104 RPT airmisses in 8 years at MBZs (ATSB report Table 3) despite having lots of RPT traffic, and 41 RPT airmisses in CTAFs even though there was little RPT traffic, we suspected and then proved that it was the CTAFs which were more risky.
Finally, we are not "of NAS" nor salaried line pilots (despite various ATPL and CPL and PPL licences) nor management nor ATC nor AOPA nor members of any "large and influential" organisation. Neither for nor against. We've got no vested interest, and only got into this issue because I thought that something basic had been messed up and as a result people were taking decisions based on erroneous assumptions that could err to danger. I still worry about it.