Wednesday, 10 November 2010

Antony Green analysis exposed

ABC Electoral Analyst, Antony Green, has been exposed publishing false and misleading information, In doing a comparison analysis between the 2010 Senate and the Victorian upper-house statistics Antony green excluded from his calculation pre-polling and postal votes. This omission has the effect of inflating the ALP and the Green preferences and underestimating the true value of the Liberal Party vote.

UPDATE: Antony green has made some partial corrections to his initial data analysis but he still is a few percentage point overall out. The ALP and Greens are short 0.5% each and the LNP 1-2% points undervalued - Still missing is the breakdown of the method used in proportioning out the absentee. pre-poll and postal votes which could explain further the discrepancy in his data set. The other missing aspect of his analysis is he has not undertaken a distribution, allocation of the below the line vote. The Sex Party did much better then he has suggested. Bias or just careless? The same issues appeared in his failure to properly examine the Queensland 2007 Senate vote. Addition, It need to be noted that Antony Green was prepared to speculate on the ALP changing its preference allocation in Western Victoria but shied away form considering the Greens issuing a split ticket and the effect that that would have on the outcome of the election. I wonder why he has attributed one option but discounted the other. - come sunday all will be revealed.

In responding to our question Antony Green stated:

GREEN COMMENT: I stand by figures. I have allocated all polling places by which region they lie in, including splitting electorates that lie in more than one region. I didn't include the pre-poll and postal votes because it rarely makes a massive difference.

Well it does make a difference and the margin between the ALP winning a second seat and the Liberal party winning three seats is much closer as a result.  If the Greens, as has been suggested here and on numerous other web sites, issue a split ticket they will be giving the Liberal party a heads up  The analysis provided by Anthony Green of ALP 2, LNO 2. Gen 1) is wrong.

GREEN COMMENT: Why should I allocate a split Green preference ticket? You assert the Greens will do this. If the Greens do issue a split ticket, we will know on Sunday and it will be the biggest story of the election campaign. I see no reason to start off analysis by assuming a flow of preferences that has never happened.

Well Sunday is a bit late the public need to be properly informed.  Antony Green in not publishing in more detail the margin for a change in outcome has mislead the public.  On comparing the Southern Metro to Senate vote the margin is within 1-2% and a split Green ticket would favour the Liberal Party. Anthony Green should know this to be true but he committed to mention this fact. Why?  He makes all kinds of predictions after all is not the swing chart a prediction based on statistical data? To publish an incomplete data set is another issue that only compounds the omission.

Further there is concern that Antony Green has politically censored comments published on his web site. Edited out are the following comments on Victoria's Upper house regions. (Published here in full)


The ALP and Green data has been inflated and the LNP under valued. If you run a simulation count based on the 2010 preference distribution, including the below the line vote , for Southern Metro and then add in a split ticket or have the Greens Preference the Liberal Party in their above-the-line group voting ticket then the results most certainly do change.

The Greens can not direct preferences for the lower house BUT they can direct preferences for the upper-house, Most above the line voters will not know where the preferences are allocated and how they will play out in the count. Only 3% of all voters vote below-the-line. Add to that the distortion on the proportionality of the count arising from the flawed non weighted calculation of the Surplus Transfer value and the method and order of distributing preference data from excluded candidates and the election results are very much on a knifes edge.

The Greens won Western Metro after a recount.  500 votes went missing and unaccounted for between Count A and Count B.  Either the VEC double counted in count A or votes were removed.  the total number of votes between count A and count B should never change. The Greens won on the second recount by less than 150 votes. 

When the  Parliament requested copies of the Count A preference data files for comparison the VEC claimed that the data had been deleted and overwritten.  No backup copies made.  This is hard to believe for a professional organisation were this information costs millions of dollars to collate.  No audit trail of count A in comparison to count B exists. 

The AEC maintain copies of both count A and count B data comparisons. what benefit was there in the VEC duplicating the development of software to count the vote when the AEC already had a better version which  could have been used free of cost?  Millions of dollars wasted in duplicated software resources.

The VEC due to a lack of due diligence resulted in doggy data being feed into their computerised count. There was no check or verification to ensure that the number of votes recorded reconciled with the number of ballot papers issued prior to the solution of the election results.  we are told that the VEC will produce a reconciliation report prior to the count but has yet to provide a sample copy of what this report will look like and what information it will contain. In 2006 the VEC failed to provided copies of the preference data files to scrutineers.,  This information was only made available following an FOI application and even then they only provided information related to the final count not the preliminary counts.  Data files had been overwritten without backup copies being made.  Even though copies of the information had been requested prior to the commencement of the count

The elections costs Victoria over 50 million dollars and hopefully we will not see a repeat of the mistakes that were made in 2006.

No comments: