Research Report 8 - Analysis of Informality in Werriwa During the March 2005 By-Election: Analysis of Werriwa

Updated: 30 May 2013

Analysis of Werriwa

The AEC maintains ballot papers from each election by Polling Place. Informal ballots from the 2005 Werriwa by-election were sorted by informality and categorised by the same informality types used in the analysis of the 2004 Election.

At the 2004 federal election, the informal vote in Werriwa was 7.98 percent, with seven candidates listed on the ballot. The national informality level for 2004 was 5.18 percent. In an attempt to limit the level of informality at the by-election, (especially given an exceptionally high number of candidates – 16), the Australian Electoral Commission conducted a series of public information campaigns to remind electors that they must number all squares and not use ticks and crosses for their ballots to be counted as valid. Statistics [5] show that many of the electors living in the division of Werriwa are from culturally and linguistically diverse backgrounds. The 2003 analysis on Informality revealed a correlation between informality levels and electors from Non-English Speaking Backgrounds (NESB).

Table 2: Breakdown of languages spoken in Werriwa as a percentage of the total population
Language Percent of population in Werriwa National percentage
English 58.32% 79.36%
Arabic 4.26% 1.11%
Australian indigineous languages 0.00% 0.07%
Cantonese 1.32% 1.19%
Mandarin 0.75% 0.73%
Other – Chinese 0.36% 0.20%
Croatian 0.91% 0.37%
French 0.47% 0.21%
German 0.28% 0.40%
Greek 1.17% 1.40%
Hindi 3.21% 0.25%
Hungarian 0.14% 0.13%
Indonesian 0.22% 0.20%
Italian 3.07% 1.87%
Japanese 0.02% 0.15%
Khmer 0.56% 0.12%
Korean 0.06% 0.21%
Macedonian 0.80% 0.38%
Maltese 0.43% 0.22%
Netherlandic 0.09% 0.21%
Persian 0.13% 0.13%
Polish 0.71% 0.31%
Portuguese 0.32% 0.13%
Russian 0.16% 0.18%
Samoan 1.00% 0.12%
Serbian 0.96% 0.26%
Sinhalese 0.15% 0.11%
South Slavic 0.19% 0.08%
Spanish 3.00% 0.49%
Tagalog 2.25% 0.42%
Tamil 0.13% 0.13%
Turkish 0.50% 0.27%
Vietnamese 2.17% 0.92%
Other languages 6.02% 1.91%
Not stated 5.46% 4.71%

Following are the informality results from the 2005 by-election with comparisons to the 2001 and 2004 elections:

Table 3: Summary of Werriwa Informality
  2001 2004 2005 by-election
Total number of votes cast: 78138 84306 77291
Total number of informal votes: 6651 6724 10162
Percentage of Informality: 8.51% 7.98% 13.15 %
Number of Candidates 8 7 16
Category of Informality 2001 Ballots 2001 % of total informality 2004 Ballots 2004 % of total informality 2005 by-election ballots 2005 by-election % of total informality
Blanks 1170 17.59% 1404 20.88% 1873 18.43 %
Number '1' Only 2299 34.57% 2482 36.91% 927 9.12 %
Ticks and Crosses 949 14.27% 983 14.62% 489 4.81 %
Langer Style * 266 4.00% N/A* N/A* N/A* N/A*
Non-sequential 691 10.39% 793 11.79% 3659 36.01%
Voter Identified 1 0.02% 1 0.01% 1 0.01%
Marks 1266 19.03% 699 10.40% 2209 21.74%
Slogans making numbers illegible 9 0.14% 0 0.00% 268 2.64%
Other *
(includes other symbols)
0 0.00% 48 0.71% 122 1.20%
Incomplete Numbering (grouped with 'Other' in 2001) N/A* N/A* 314 4.67% 614 0.12%

Overall informality from the 2004 Federal Election to the 2005 by-election increased by 5.17 percentage points.

The NSW State election was held in March 2003. At the 2004 federal election for the House of Representatives, informal ballots due to the elector marking 'Number 1 only' or 'non-sequential' made up almost 50 percent of overall informality in NSW. This category decreased significantly from 36.91 percent (Werriwa) at the 2004 federal election to only 9.12 percent at the by-election. This could suggest that more time between a federal and state election for states with optional preferential voting may decrease informality levels in this category.

Important also is the absence of the Senate election at by-elections. There is no advertising by political parties to place a '1' only in the Senate square and no combined House of Representatives and Senate 'how to vote' cards. There is consequently only one ballot paper handed to the elector on polling day.

Previous research undertaken by the AEC has argued that the difference in the voting system between the House of Representatives and the Senate may impact on the informal vote [6]. The difference between the two systems may also be compounded by 'How to Vote Cards' and the order the voter completes their papers (Senate and House of Representatives). At by-elections there is no Senate election and consequently no advertising by political parties to place a '1' only in the Senate square and no combined 'how to vote' cards. There is consequently only one ballot paper handed to the elector on polling day. The absence of the Senate ballot paper at the Werriwa by-election may have made voting easier for the elector and may have contributed to the substantial decline in 'number 1' only.

As evidenced in the analysis from the 2004 election, as the number of candidates increases, so does the overall level of informality. As the number of candidates more than doubled, the informality rate would be expected to increase. Indeed, the category of informality "Non-sequential" increased to 36.01 percent from 11.79 percent at the by-election.

The analysis of 2004 election produced the following regression:

Table 4 Regression Analysis of Candidates/Informality
  Coefficient Standard error T-stat P-value
Intercept 0.002591 0.000604 4.278721 3.28351E-05
No. Of candidates 0.003193 0.000279 11.42212 5.94221E-22

This regression indicates that as the number of candidates increases, the informality rate will increase by 0.3 percent. Applying the coefficient and intercept from this regression to the by-election, we would have the following equation:

Ŷ = b0 + b1 χ

Projected informality by-election = 7.98 + 0.3(9) + 0.2 = 10.88%

Although the actual informality rate was higher at 13.15 percent, this projection only factors in changes in the number of candidates. In the Australian context, factors which can influence informality levels include differences in the voting systems between the States and the Commonwealth, differences in the voting arrangements between the House of Representative and the Senate, and sociological factors. Given the significant increase in "Scribbles/Marks" ballots, their analysis could provide further insight into why the overall informality level increased.

Given the extremely high number of candidates, an analysis of the informality data collected provides further information on how many electors began to number their ballots, but stopped at some point, rendering the ballot informal with an insufficient number of preferences expressed.


  1. This data has been extracted from the Australian Bureau of Statistics CDATA2001 database.
  2. A discussion of the differences between the Senate and the House of Representatives voting system and the impact on informal voting is found in 'Research Report 1 – Informal Vote Survey House of Representatives 2001 Election'