Friday, May 09, 2008

Statistics NZ Email Explains Counting Process

"Why did the Clerk say that the petition calling for a referendum on the Anti-Smacking Law was short by 15,500, when we had expected it to be short by approximately 3,000 signatures?"

This is the question many people have been asking.  You would have thought that the Clerk would have issued a press release, explaining the process used to make the calculation.  One of the political bloggers in New Zealand explained what was the most likely reasoning behind the Clerk's decision.  Now, finally, I have got my hands on an email from the office of Statistics NZ, explaining clearly, the process used to discern the quality of the signatures we submitted.  The selection in bold is the key bit you want to read a couple of times.

"I am replying to your enquiry seeking more details on the estimate of the number of registered electors who had signed the Citizens Initiated Referendum in regard to the "anti smacking" legislation.

The results from the sample, show the number of signatures excluded was much larger than you thought it should be.
The explanation is multiple signatures, and the adjustments that need to be made for them.
Because signatures for the petition are collected over a full year, by different collectors, and in different locations, it is possible that some people will sign the petition more than once.

Only signatures from registered electors count towards the required total. Under the legislation, a sample of signatures is taken by the Clerk of the House These signatures are then checked against the Electoral Roll by the Electoral Enrolment Centre.

In deciding whether enough electors have signed the petition, we need to take into account electors who have signed more than once. Only one of their signatures can count.

It is a rare event for both signatures from someone who has signed twice to appear in the same sample. Most signatures in the sample from people who signed twice will show up only as single signatures. It is more difficult to draw both signatures in the sample.

For these reasons, the proportion of electors in the sample does not represent the proportion of electors in the full petition who have signed the petition twice.

To put numbers into the argument we have taken a 1 in 11 sample. The probability that an elector who has signed twice and has both their signatures appear in the sample is about 1 in 121.

So the 158 double signatures in the sample will not represent 11 times 158 or about 1,738 electors who have signed twice. They will represent 121 times 158, or about 19,100 electors who have signed twice.

That is the main reason that the number of exclusions is larger than you expected..."

While we're on the subject of Referendums, below is an excerpt from Larry Baldock's address to The Kiwi Party Conference in Wellington, 5 April 2008...

"Firstly we will change the threshold for a CIR referendum from 10% of those on the electoral role as it is currently to 5% of those who voted at the previous election. The current high threshold of 10% is why so few petitions ever succeed, and that is probably the way the National and Labour parties like it to be.

In the state of California citizens can achieve a binding, yes binding referendum with a petition of only 5% of those who voted for the successful Governor at the last elections. That works out at less than 300,000 signatures in a population of 36 million.

Secondly if any citizen completes a referendum calling for the repeal of legislation already passed by a deaf parliament as we have done then those referenda must be binding also."

Larry goes on in his third and fourth points, calling for regular referendums to be held mid-term, and also for binding referendums on all private member's bills (such as Sue Bradford's Anti-Smacking Bill).  I disagree with Larry's third point - Citizen's Initated Referendums (CIR) should take place following the collection of signatures, which should not be restricted to within a pre-specified time-period mid-term.  Also, regarding Binding Citizen's Initiated Referendums, these should be used sparingly.  Statistics show that the more elections/referendums that are held, the fewer citizens participate.

Binding Referendums (not necessarily Citizen's Initiated) must be taken on all bills on moral or ethical issues.  You would find that MPs would be far less likely to introduce extreme bills, as there would be no hope of them passing, thanks to this new democratic process.