It has been hypothesised that the proportion of informal votes increases with the number of candidates on the ballot paper. In 2004, the average number of candidates per divisional ballot was 7.27, an increase of 4.9% from 6.93 candidates per ballot in 2001. Greenway (New South Wales) was the division with the highest level of informal voting (11.83%) and the highest number of candidates (14).
Table 9 compares informality, by state and nationally, with the number of candidates on the 2001 and 2004 ballot papers.
|2001 informal (%)||5.42||4.83||3.98||4.92||5.54||3.40||3.52||4.64||4.82|
|2004 informal (%)||6.12||5.16||4.10||5.32||5.56||3.59||3.44||4.45||5.18|
|Change in percentage points||+0.70||+0.33||+0.12||+0.40||+0.02||+0.19||–0.08||–0.19||+0.36|
|2001 average no. of candidates||7.90||6.63||6.10||7.70||5.92||5.40||6.50||6.50||6.92|
|2004 average no. of candidates||7.64||7.32||7.00||8.30||6.91||5.00||5.00||6.00||7.27|
|Change in percentage points||–0.26||+0.69||+0.90||+0.60||+.99||–0.40||–1.50||–0.50||+0.35|
A simple regression analysis (Table 10) to test the correlation of informality with the increase / decrease in the number of candidates on the ballot produces the following results:
Multiple R: 0.688565
R Square: 0.474121
Adjusted R Square: 0.47052
|T-stat||P-value||Lower 95%||Upper 95%|
|No. of candidates||0.003192||0.000278||11.47304||4.0E-22||0.002642||0.0037420|
Generating a line of regression produces Figure 5.
Figure 5: Percentage change in informal votes and change in total number of candidates between 2001 and 2004
The 'number of candidates' variable is a strong predictor of informality. Unsurprisingly, Greenway, the seat with the highest number of candidates (14) and the largest increase in the number of candidates from 2001 (six), had the highest percentage of informal voting.
Furthermore, a scattergram (Figure 6) shows that, as the number of candidates increases, so does the percentage of 'Incomplete' and 'Non-sequential' informal ballots.
Figure 6: Incomplete and non-sequential votes in 2004, against number of candidates
This analysis confirms the hypothesis and accords with common sense: the more candidates an elector has to give preferences to, the more likely it is that the elector will make an error in fully and sequentially numbering all the boxes on the ballot.