Seventeen thousand blank forms, and 20,000 more in which the vote was unclear, were among the 12.7 million posted back to the Australian Bureau of Statistics (ABS) during the same-sex marriage survey.
These figures were revealed in a report about the observer program, under which people nominated by the "yes" and "no" sides were able to lodge disagreements – which were noted, but not acted on – with how votes were counted.
According to the ABS, taking those disagreements into account would have changed the survey margin by a tiny 56 votes.
The report said that of the 12.7 million responses, observers had viewed a sample of 606,991, amounting to 4.8%.
Observers were not able to formally challenge the decisions made by the ABS, but rather note whether or not they agreed with the interpretations. This is different to how a regular election would run, in which scrutineers are able to watch, and challenge, the counting of every single vote.
The report says the the vast majority of survey responses – 95% – were sufficiently clear to have the "yes" or "no" response marked via machine. This means the people had followed the instructions – clearly marking just one box, with no extraneous marks, comments, pictures or material on the form.
The other 5% were done manually by staff, who looked to an extensive set of guidelines about how various marks on the form should be interpreted.
The report provided by the ABS suggests observers largely agreed with the interpretation of the surveys, and that even if their disagreements had been taken into account, it would have had an infinitesimally small effect on the numbers.
Of the sample of 606,991 forms that were scrutinised by observers: 311,318 were automatically counted yes/no votes, 261,678 were manually counted yes/no votes, 15,659 were blank forms, and 18,336 were forms marked "response not clear".
Unsurprisingly, the "response not clear" forms were the most consistently contested by observers, with both "yes" and "no" observers disagreeing with the ABS determination on 625 forms, or 3.4% of those viewed.
Of the other categories, 337 (0.1%) of the manual forms elicited disagreement from both observers, along with 29 (0.2%) blank forms and a single one (0.0003%) of the forms that had been automatically counted.
"These rates of agreement indicate, that if the observers were to have provided feedback on all 12,727,920 survey responses, then observers would have agreed with 99.99% of forms processed, which equates to 9 survey return disagreements per electoral division," the report reads.
"There were comparable numbers of Yes and No disagreements, such that if all of the disagreements were added up, and the observer's recommendation adopted, there would be a net decrease in the survey margin of just 56 survey responses at the national level."
Coalition For Marriage partner Australian Family Association has been sharing a petition on its Facebook page calling on the government to hold an "independent audit" into the survey.
"We believe that the results provided by the Australian Bureau of Statistics are unreliable and do not represent the true numbers reflected by the Australian people," the petition – with just over 2,000 signatures at the time of writing – states.
The ABS has said before that the survey is not open to legal challenges.
Kim Mansfield-Reid, from Warrawong in the Illawarra in NSW, told BuzzFeed News via Twitter she had posted back a blank form because she opposed the process, but felt it was important her non-vote was counted.
"The form sat on my kitchen bench (with all manner of kids craft, bills, pens and hair pins) as I couldn't decide what to do… either 'yes' or 'informal'," she wrote.
"I thought not sending it in would give the wrong impression – that I didn't care – but that wasn't true. I did care. I saw it there, amongst all that other stuff, every day and I knew the deadline was approaching. I sent it off on the second last day (and also contacted neighbours offering to drop their completed/sealed envelope in our local post box near our local school) as I wanted my 'non-vote' to count."
Mansfield-Reid was happy with the "yes" victory and had no regrets about her decision to send back a blank form.
In the Quality and Integrity statement accompanying the results, the ABS said there had been fewer than 500 instances of survey fraud, theft, and forms for sale reported to the Australian Bureau of Statistics (ABS) throughout the survey period.
Lane Sainty is a reporter for BuzzFeed News and is based in Sydney, Australia.
Contact Lane Sainty at firstname.lastname@example.org.
Got a confidential tip? Submit it here.