Thursday, December 2, 2010

Message to Antony Green ABC blog

Your calculator does not include the preference allocation for ungrouped independents.

In Northern Metropolitan Adrian Whitehead received 309 primary votes before the VEC pulled the publication of the election results.

If you look at Adrain Whitehead's HTV card your will see that he preferences directly to the Greens then exhausts.

Your calculator was showing incorrectly Stephen Mayne in a winning position when in fact he was behind the Greens' second candidate by 120 votes at the crucial juncture in the count. Mayne would have been eliminated and the Liberal Candidate was placed to win the fifth seat by a margin of 6,800 votes.


Antony Green, ABC said...

COMMENT: I consider the remarks at your post to be irrelevant and other remarks at your blog are personally offensive, but as I am completely sick of your accusations of bias, I will include the link to your post.

The calculator is an automated process. It treats below the line votes as if they were ticket votes. There are no tickets for ungrouped candidates so they cannot be assumed to follow any ticket. There were also so few of them that it would be impossible to know if they mattered until the count was well progressed.

Early in the count there are always variations in the level of support for parties. But as the count progresses, it becomes clearer whether below the line votes will matter or not.

At that point I provide additional commentary beyond what is provided by the automated calculator, as I have done in detail for Western Metropolitan Region over the last two days where the flow of below the line preferences as well as the rate of exhausted preferences will play a part in the final result.

If Northern Metropolitan Region had been close to the point where below the line votes mattered, I would have included the ungrouped candidates in additional commentary. But the idea of commentating on below the line votes at early stages of the count was not justified, and nor was it worth including them in the calculator.

If you think that a calculator that incorporates analysis of ungrouped votes and below the line preferences is of value, I would encourage you to produce such a tool. In my view the current ABC automated calculator, augmented by blog posts where I express my professional judgment on the calculator's output, is sufficient for analysing the result.

democracyATwork said...


Contrary to your denial of bias, it is a fact that you have selectively edited statements that are pertinent to the discussion on how our electoral system works and removed comments that are critical of your assessment most certainly indicates a bias. Whether you acknowledge and agree it or not. Not to mention your lack of professionalism and beaches of ABC guidelines.

As any scientist will tell you if you apply one assumption to a sample you must apply the same to all samples if your are to do a comparative analysis without distorting or establishing a bias in your statistics.

At the stage in the count when the comment in relation to Whiteheads preferences was made Whitehead's preferences most certainly did play significant role in the projected outcome of the election. If you’re going to treat below-the-line preference as if they are ticket votes, in your analysis (And I have no problem with that) then you should also apply the same logic to the independent candidate who had also registered a how to vote card. Same rules should apply to all players.

One of the aspects of your calculator I dislike is that it masked the calculations and assumptions you have applied. most viewers do not realise the ho the surplus transfer value is calculated and the effect that has on the overall result. The same applies to the method of segmentation. (Again if you count the 2007 Queensland Senate count and exclude all candidates except the last seven standing- ALP 3, LNP 3, and Grn 3 - Greens Candidate Larissa Waters should have been elected. Waters, the reason she was not elected was due to the distortion in the proportionality of the count arising from the method of segmentation. segmentation is akin to dealing from the bottom of the deck and increased the power of below the line and minor party tickets that are distributed at a later stage in the count - The principle should be that if a candidate is excluded from the count then their ballot paper should be allocated as if the candidate excluded that had not stood. IE go to my next choice. If that choice is a candidate who has already been elected then the vote should form part of the candidates surplus and distributed according with the surplus. Not jump the queue.

Segmentation was implemented to facilitate a manual count and to limit the other distortions that arise from the flawed method of calculating a surplus transfer value (An issue that you have previously addressed - but you have not examined or even considered the issue of segmentation which is just as equally flawed as demonstrated by the 2007 Queensland Senate result.)

Yes these are issues of a technical nature and most onlookers would not understand, but your response is to not even allow any reference to the fact that the system as it stands is not proportional - Semi proportional at best.

democracyATwork said...

It is not too much to ask that the counting of the ballot be open, transparent and mathematically correct.

We fight over single member electorates that are won or lost on the back of a single vote but ignore the fact that the system as implemented delivers a distortion equivalent to 1,000's of votes.

In fact you at one stage even denied that the distortion existed, Which was false. Try Counting the 2007 Queensland senate vote as suggested above then comment on the basis of being informed.

Your calculator would be of great use if the VEC and the AEC provided real time or snapshots of the below the line data files as and when the data is being recorded. It is not that difficult and could readily be undertaken, if the Electoral Commissions maintained an open and transparent counting system.

The other aspect that you have not touched on and your calculator provides misleading data on is the percentage of the vote. In fact it is not the percentage of the vote but the percentage of the vote counted in relation to the enrolled vote that your are reporting.

If the electoral commission maintain proper and accurate data on the number of ballot papers issued and received back then the data published on your calculator would be more useful.

It would also help if you included the date-stamp of the data you had processed. As I have noticed that there is a time lag between the Commission publishing the data and your calculator picking up the change. (which was what was happening when the Whitehead commenst were made, which you denied effected the result)

The role of the electoral commission in providing information is not just to inform the general public but to also assist in the scrutiny of the ballot. Its not about the theatre of both the ABC and Commission opening an envelope and saying "And the winner is..."

I raise all these issues as I believe they have some significance and are important, even if you do not think they have any relevance.

I for one think if we elect the wrong person that is important to discuss and know ho and why this is the case?

I certainly do not support people trying to prevent information being discussed which is what you are doing by seeking to censor the discussion.

Post a Comment