Research Report 7 - Analysis of Informality During the House of Representatives 2004 Election: Survey of informality in the 2004 House of Representatives election

Survey of informality in the 2004 House of Representatives election

Between 2001 and 2004, there was a slight increase in the percentage of informal votes for House of Representatives elections. In 2004 13 098 461 Australians were enrolled to vote, and a total of 12 354 983 votes were cast, a participation rate of 94.32%. Of the total votes cast 639 851 (5.18%) were counted as informal, an increase in 0.36 percentage points (or 7.5% of total informal votes) since 2001.

The high informality rates at the 2001 election resulted in several activities by the AEC to address and reduce informal voting. Several different initiatives were undertaken, including an enhanced public awareness program in New South Wales and Queensland to address the possible impact of optional preferential voting systems used for state elections. In addition to regular advertising, this involved having posters in all polling places to remind electors to number every square on the House of Representatives ballot paper. Issuing officers were also provided with a script, and instructed to remind all electors of this requirement when issuing ballot papers.

In New South Wales, community information sessions were conducted during August and September 2004 in Auburn, Parramatta, Liverpool, Cabramatta, Blacktown and the Canterbury–Bankstown area. These areas were selected because of the high levels of informal voting recorded at the 2001 election. Information sessions conducted in conjunction with Migrant Resource Centres were designed to educate ethnic community leaders and service providers, who in turn acted as intermediaries within their local communities to inform others about how to participate fully in the election process and make their votes count.

The 2004 informality levels suggest that there needs to be continued emphasis on activities and programs within divisions with continued high levels of informality.

Appendix: Research Report Number 7–Analysis of the Increase in Informality During the House of Representatives 2004 Election–Divisional Summaries has a full breakdown of informal votes by category for each division.

Breakdown of informality by category

In general, informal ballots can be broken into the following categories:

• Blank
• Number 1 only
• Ticks and crosses
• Langer-style voting (e.g. 1, 2, 3, 3, 3 …)
• Non-sequential (e.g. 1, 2, 400, 327 …)
• Voter identified
• Marks on ballot and no preference stated
• Slogans making preferences illegible
• Other, including other symbols
• Incomplete numbering.

There are some differences between how informal votes were categorised in 2001 and how they were categorised in 2004. In 2004, Langer-style votes were simply counted as 'Non-sequential', and separate categories were established for incomplete votes and votes that used non-numeric symbols (in 2001, any vote that listed an incomplete number of preferences was categorised as 'Other', along with those using non-numeric symbols). While electoral officials were given instructions on how to categorise informal ballots for both elections, in some cases they may have differed in their interpretation of the categories. This is largely because some informal ballots satisfy multiple criteria, and might therefore be placed in more than one category. For example, if a ballot paper has non-sequential numbering, ticks and crosses, and marks and slogans, it could be placed in any one of three categories.

To make categories used in 2001 and 2004 more nearly equivalent for statistical and comparative purposes, they have been adjusted as follows.

• 'Other 2004' includes 'Other symbols'.
• 'Langer-style' was not counted in 2004. These ballots would have been included in 'Non-sequential' in 2004.
• 'Incomplete numbering' was counted only in 2004. In 2001, ballots with incomplete numbering would have been categorised as 'Other'.

Because some ballots categorised simply as 'Other' in 2001 were subcategorised into 'Incomplete' in 2004, the 2004 'Other' category can be expected to be smaller (all other factors being equal).

Figure 1 and Table 2 show the proportions of informal votes by category for the 2001 and 2004 elections. For comparative purposes, Figure 1 combines the 2004 'Incomplete numbering' category with 'Other', since those ballots would have been categorised as 'Other' in 2001. Table 2 lists these categories separately.

Figure 1: Informality, by category, 2001 and 2004

Table 2: Informality, by category, 2001 and 2004 (% of total informal votes)

Category	National average 2001	ACT	NSW	NT	Qld	SA	Tas.	Vic.	WA
Blanks
2004	21.15	23.58	21.18	18.77	15.24	23.16	28.21	24.16	22.89
2001	21.22	30.84	20.38	20.74	15.67	24.52	27.86	24.95	23.36
Number 1 only
2004	32.83	35.57	35.65	27.65	44.57	30.91	22.37	21.81	25.25
2001	33.72	28.76	32.47	27.95	46.42	36.63	23.60	26.05	29.87
Ticks and crosses
2004	9.34	8.00	10.71	9.00	7.37	11.72	11.42	7.40	9.18
2001	12.92	8.99	12.57	10.62	11.46	14.95	15.84	12.97	9.93
Langer-style
2001 only	2.86	0.83	2.37	14.56	2.00	1.05	6.88	3.22	4.18
Non-sequential
2004	15.35	4.88	15.22	19.85	9.79	14.12	8.19	20.39	19.31
2001	17.31	7.66	22.52	15.06	10.49	13.40	13.17	14.15	21.75
Voter identified
2004	0.04	0.00	0.04	0.05	0.06	0.05	0.07	0.06	0.02
2001	0.05	0.04	0.04	0.00	0.03	0.03	0.02	0.07	0.11
Marks and scribbles
2004	14.27	20.21	9.62	15.60	15.65	13.72	24.66	20.14	15.94
2001	6.39	4.20	5.49	2.98	4.91	5.97	12.11	8.23	7.78
Slogans making numbers illegible
2004	0.83	0.58	0.73	1.13	0.97	1.37	0.62	0.78	0.60
2001	0.26	0.05	0.28	0.00	0.30	0.57	0.01	0.42	0.18
Other
2004	1.82	3.32	1.55	4.23	1.73	1.69	1.71	2.21	1.83
2001 (includes other symbols)	5.98	18.63	3.87	8.09	8.72	2.87	0.51	3.98	2.83
Incomplete numbering
2004 only	4.39	3.05	5.23	3.71	4.63	3.05	2.75	3.05	4.97
Average number of candidates: 2004 2001	7.27 6.92	5.00 6.50	7.64 7.90	6.00 6.50	7.32 6.63	6.91 5.92	5.00 5.40	7.00 6.10	8.30 7.70

Table 3: Divisions with highest/lowest informality in 2004 and comparison with 2001 informality

State	Division	Total informal % 2001	Total informal % 2004	Socioeconomic index a	Geographical classification
Division with lowest informality levels in 2004
Vic.	Higgins	2.68	2.76	High	Inner Metropolitan
NSW	New England	1.97	2.77	Lower	Rural
Vic.	Bendigo	3.72	2.87	Low	Provincial
Vic.	Indi	5.22	2.88	Lower Middle	Rural
Vic.	Kooyong	2.57	2.90	High	Inner Metropolitan
Vic.	Corangamite	3.38	3.04	Upper Middle	Rural
Vic.	Deakin	2.56	3.06	High	Outer Metropolitan
Tas.	Denison	2.91	3.10	Upper	Inner Metropolitan
Vic.	Melbourne	3.77	3.27	Lower Middle	Inner Metropolitan
Vic.	Ballarat	2.88	3.34	High	Provincial
Divisions with highest informality levels in 2004 b
NSW	Werriwa	8.51	7.98	Lower Middle	Outer Metropolitan
NSW	Kingsford Smith	6.14	8.43	High	Inner Metropolitan
NSW	Parramatta	6.21	8.53	High	Inner Metropolitan
NSW	Watson	7.52	9.10	Lower	Inner Metropolitan
NSW	Fowler	12.75	9.11	Low	Outer Metropolitan
NSW	Prospect	8.99	9.24	Lower Middle	Outer Metropolitan
NSW	Chifley	9.20	10.10	Low	Outer Metropolitan
NSW	Blaxland	9.78	10.70	Lower Middle	Inner Metropolitan
NSW	Reid	11.08	11.71	Lower Middle	Inner Metropolitan
NSW	Greenway	6.79	11.83	Upper Middle	Outer Metropolitan

1. Medew R, 2005. 'Analysis of Electoral Divisions Classification', Australian Electoral Commission. 2001 ABS Census data used.
2. These divisions had a high level of informality despite public awareness campaigns and new initiatives aimed at educating electors about how to cast a formal vote.

In 2004, the 10 divisions with the highest levels of overall informality were all in Sydney. Using the Australian Bureau of Statistics' socioeconomic index, which is based on 2001 census statistics, three out of 10 of these divisions are classified as low, lower or lower middle. In contrast, the divisions with the lowest levels of informality are in a range of socio-economic classifications, geographical locations and states.

Whilst the ranking of divisions with the highest and lowest informality levels may have changed slightly, the comparison of informality levels above reveals that informality levels did not vary significantly between 2001 and 2004.

A correlation scattergram (Figure 2) shows a very slight correlation between overall levels of informality and an area's socioeconomic index. A regression of this data reveals that this correlation is not statistically significant.

Figure 2: National level of informality, percentage against socioeconomic index

Informality and geographical location

While there has been an increase in informal voting across all division locations, informal voting levels have been consistently higher in metropolitan areas for the past two elections (Table 4).

Table 4: Increases/decreases in informality, 2004

Division location	Total informal / formal 2001	Total informal / formal 2004	Average no. of candidates 2004	Average change in candidates from 2001	% points increase/decrease in informality
Rural	4.56%	4.79%	7.49	+0.07	+0.23
Provincial	4.60%	4.81%	7.11	-0.56	+0.22
Outer Metropolitan	5.06%	5.43%	7.19	+0.51	+0.37
Inner Metropolitan	4.95%	5.52%	7.26	+0.83	+0.57

Declaration votes

Informality levels across all types of declaration votes rose in 2004 consistent with the overall increase in informality from 2001.

An analysis of all declaration votes by state (Table 5) suggests that people who cast postal and pre-poll votes have lower incidences of informality. Postal voters often have long periods of time at home to read through the ballot and the instructions, and are not hurried by the normal business of election-day voting, which might contribute to informality levels.

The informality level for absent votes is 5.13 percent, consistent with the informality level for all votes cast – 5.18 percent. Informality levels among provisional voters have typically been higher – in 2001, 6.73%, in 2004 6.82%.

Table 5: Declaration votes 2004

State	Absent votes	Informal absent	Postal votes	Informal postal	Pre-poll votes	Informal pre-poll	Provisional votes	Informal provisional
ACT	4 368	168	7 567	104	33 289	744	2 643	165
Total informal		3.85%		1.37%		2.24%		6.24%
NSW	245 680	14 283	189 256	4 743	259 022	8 986	28 544	2 290
Total informal		5.81%		2.51%		3.47%		8.02%
NT	2 070	93	3 107	43	10 050	268	1 183	61
Total informal		4.49%		1.38%		2.67%		5.16%
Qld	132 752	5 941	136 977	3 018	116 870	3 393	14 667	1 033
Total informal		4.48%		2.20%		2.90%		7.04%
SA	61 997	3 326	44 662	839	40 807	1 224	4 843	391
Total informal		5.37%		1.89%		3.00%		8.07%
Tas.	15 038	421	15 837	249	14 465	275	3 353	184
Total informal		2.80%		1.57%		1.90%		5.49%
Vic.	180 961	8 692	176 645	3 071	187 468	5 128	22 201	1 139
Total informal		4.80%		1.74%		2.74%		5.13%
WA	102 246	5 307	39 226	840	56 078	1 550	13 078	906
Total informal		5.19%		2.14%		2.76%		6.93%
Total	745 112	38 231	613 277	12 907	718 049	21 568	90 512	6 169
Total informal '04		5.13%		2.10%		3.00%		6.82%
Total informal '01		4.89%		1.69%		2.81%		6.73%

Blanks

The proportion of informal ballots declared informal because they were blank did not change significantly between the 2001 and 2004 elections. Table 11 in the appendix of this paper lists the 10 divisions with the highest proportion of blank ballots in the 2004 election, with figures for those divisions in 2001. Figure 8 in the appendix charts blanks against the socioeconomic index.

Across all states and territories, percentage levels of informal blank ballots were at similar levels in 2001 and 2004.

Marks

The only category with significant changes between 2001 and 2004 was 'Marks and scribbles'. Ballot papers categorised under this heading are perhaps the most indicative of political protest. Most of these papers have been marked with slogans and words of protest against the political and electoral system, without the elector expressing a sufficient number of preferences. Such papers most clearly indicate the voter's intention to cast an informal ballot. This could perhaps be attributed to because of apathy, dissociation from the electoral process or dissatisfaction with the choice of candidates.

Had the proportion of ballots informal because of marks and scribbles stayed constant from 2001 to 2004, there would have been approximately 49 973 fewer informal ballots in 2004. This would have reduced the overall informality rate to 4.78%, close to the rate of 4.8% in 2001. That is, if the level of informality due to marks had stayed the same, total informality would also have been virtually unchanged.

The 10 divisions with the lowest level of informal ballots due to scribbles and marks were in inner or outer metropolitan areas in New South Wales (Table 6).

Table 6: Divisions with highest and lowest percentages of informality due to marks and scribbles [8]

State	Division	Informal from scribbles/marks 2004 (%)	Informal from scribbles/marks 2001 (%)	Socioeconomic index	Geographical classification
Divisions with highest percentages of informal ballots in marks/scribbles category
Vic.	Maribyrnong	29.11	0.64	Lower Middle	Inner Metropolitan
Tas.	Lyons	28.27	15.70	Low	Rural
Vic.	Indi	27.18	4.58	Lower Middle	Rural
Tas.	Denison	26.50	13.76	Upper Middle	Inner Metropolitan
Vic.	Scullin	26.40	11.16	Lower Middle	Outer Metropolitan
Vic.	Corangamite	26.15	3.75	Upper Middle	Rural
Vic.	Gippsland	25.70	8.03	Low	Rural
Vic.	Batman	25.57	0.17	Upper Middle	Inner Metropolitan
Tas.	Bass	25.10	4.22	Low	Provincial
Vic.	Calwell	24.73	1.36	Lower Middle	Outer Metropolitan
NSW	Cook	6.41	4.67	High	Outer Metropolitan
Divisions with lowest percentages of informal ballots in marks/scribbles category
NSW	Watson	6.06	4.19	Lower Middle	Inner Metropolitan
NSW	Barton	5.86	4.95	Upper Middle	Inner Metropolitan
NSW	Parramatta	5.79	4.69	High	Inner Metropolitan
NSW	Greenway	5.66	3.66	Upper Middle	Outer Metropolitan
NSW	Kingsford Smith	5.64	5.09	High	Inner Metropolitan
NSW	Blaxland	5.10	6.05	Lower Middle	Inner Metropolitan
NSW	Bennelong	4.91	5.83	High	Inner Metropolitan
NSW	Wentworth	4.62	7.52	High	Inner Metropolitan
NSW	Cunningham	3.63	0.72	Upper Middle	Provincial

A correlation scattergram (Figure 3) confirms the examination of the top 10 divisions and indicates a slight negative correlation between the percentage of informality due to marks and scribbles and the socioeconomic index.

Figure 3: Informality due to marks and scribbles, 2004, against socioeconomic index

Number 1 only

Full preferential voting for the House of Representatives requires every square to be numbered in order, but not all elections in Australia require this. Of the 50 divisions with the highest percentage of informal votes (ranging from 5.46% to 11.83%), 35 are in either Queensland or New South Wales, the two states where preferential voting is optional at the state level.

The state with the largest proportion of 'Number 1 only' informal votes in 2004 was Queensland with 44.57% of informal votes, a slight falling from 46.42% in 2001). In New South Wales, the proportion of 'Number 1 only' votes rose by more than three percentage points, from 32.47% to 35.65% (see Table 7; Figure 7 in the appendix to this paper charts these votes against the socioeconomic index).

Table 7: Divisions with the highest proportions of 'Number 1 only' informal votes

State	Division	Total informal 2001 (%)	Total informal 2004 (%)	Socioeconomic index	Geographical classification
Qld	Ryan	55.62	38.10	High	Outer Metropolitan
Qld	Lilley	55.16	45.05	Upper Middle	Inner Metropolitan
NSW	Bradfield	54.88	49.32	High	Inner Metropolitan
Qld	Dickson	54.13	46.65	Low	Outer Metropolitan
Qld	Longman	51.33	47.82	Low	Outer Metropolitan
Qld	Leichhardt	49.64	54.01	Upper Middle	Rural
Qld	Herbert	49.40	48.43	Upper Middle	Provincial
Qld	Petrie	49.25	50.93	Upper Middle	Outer Metropolitan
NSW	Cook	49.17	44.75	High	Outer Metropolitan
Qld	Moreton	49.00	49.25	High	Inner Metropolitan

Ticks and crosses

The use of ticks and crosses renders a House of Representatives ballot informal for Australian federal elections. The national average percentage of papers counted as informal for this reason fell from 2001 to 2004.

Eight of the 10 divisions with the highest proportion of informality due to ticks and crosses were in New South Wales. However, this proportion fell in New South Wales from 12.57% in 2001 to 10.71% in 2004. Table 12 in the appendix to this paper lists the 10 divisions with the highest proportion of informal ballots declared informal in 2004 because of ticks and crosses.

In Queensland and New South Wales, around 50% of informality is caused by either 'Ticks and crosses' or 'Number 1 only'. These are potentially avoidable types of voting informality, in which people are attempting to vote correctly but are perhaps confused by differences between state and federal voting systems.

Furthermore, it is apparent that there is a positive correlation between the percentage of informality due to 'Ticks and crosses' and the statistical census variable, 'Not fluent in English' (see Figure 4). The AEC's 2003 analysis of the 2001 election showed that the 'Not fluent In English' variable is a predictor of informality and is highly significant statistically. A regression analysis shows that this variable is a major predictor for 'Ticks and crosses' informality. Figure 9 in the appendix charts 'Ticks and crosses' against the socioeconomic index.

Figure 4: 2004 Ticks and crosses as percentage of overall informality, and variable 'Not fluent in English'.

Possible targeted actions to reduce informality

Since it is reasonable to assume that informal ballots in the 'Marks' and 'Blanks' categories indicate intentionally informal or 'protest' votes, the 'Number 1 only', 'Ticks and crosses', 'Non-sequential' and 'Incomplete' categories might best be targeted to lower the overall national informality rate. Table 8 shows the predicted effects on overall rates from reductions in all categories.

Table 8: Hypothetical changes in overall informality based on reductions within categories

Highest categories of informality	Raw total in 2004 (% of overall informality)	National informality if category reduced by 25%	National informality if category reduced by 50%	National informality if category reduced by 75%
Blanks	21.15%	4.84%	4.57%	4.30%
'One' only	32.83%	4.69%	4.27%	3.85%
Ticks and crosses	9.34%	4.99%	4.87%	4.75%
Marks	14.27%	4.92%	4.74%	4.56%
Non-sequential and incomplete	19.74%	4.85%	4.60%	4.35%