jackman.stanford.edu/blog
• 114th U.S. Senate
• ideal point estimates svg pdf csv 12/29/15
• scatterplot against 2012 Obama vote share svg pdf
• roll call object: RData
• 114th U.S. House
• ideal point estimates svg pdf csv 5/25/16
• scatterplot vs Obama vote share svg pdf
• roll call object: RData
• Bayesian Analysis for the Social Sciences Wiley; Amazon; errata as of 3/6/15

## Tuesday December 18, 2007

Filed under: Australian Politics — jackman @ 9:43 pm

The usual suspects appear to be drivers of informality in House of Reps voting this time around: (a) proportion of an electorate’s residents that are from non-English speaking households (NESH), (b) ballot length, plus (c) whether we’re in NSW or Qld (states with optional preferential voting in their state legislative assembly elections).

You put these three factors together in a regression (with a quadratic term for NESH) and you explain about 2/3 of the variation in informality across electoral divisions. A summary of the regression fit appears at the bottom of the post (over the fold, as they say).

See the graphs below (thumbnails). I wrote on this re 2004 in a short note here.

This does give us yet another take on the McKew effort in Bennelong. She drew 13 on the ballot paper, which was not only the bottom of the ballot paper in Bennelong, but the lowest position on any House of Reps ballot in the country (i.e., Bennelong had the largest number of candidates, and McKew drew slot 13). Bennelong also scores quite high on NESH (36.3%, 16th highest in the country). With the long ballot and the confusion caused by NSW having optional preferential, we see a 6.22% rate of informality, compared with average informality rate of 3.95%, and a 4.96% average informality rate in NSW seats. The model described above predicts an informality rate of 6.57% for Bennelong.

Many of those informal ballots were likely votes intended for McKew (i.e., NESH votes skew Labor), perhaps shaving a percentage point off the vote total she might have got with a shorter ballot?

R output:

> summary(model)

Call:
lm(formula = InformalPercent ~ poly(nes, 2) + BallotLength +
OpPref, data = foo)

Residuals:
Min       1Q   Median       3Q      Max
-2.03104 -0.47986 -0.03003  0.47632  2.34372

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept)    1.93790    0.30113   6.435 1.69e-09 ***
poly(nes, 2)1  9.77897    0.79282  12.334  < 2e-16 ***
poly(nes, 2)2  4.87446    0.81225   6.001 1.50e-08 ***
BallotLength   0.22586    0.04194   5.386 2.85e-07 ***
OpPrefTRUE     0.81249    0.13289   6.114 8.56e-09 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Residual standard error: 0.7841 on 145 degrees of freedom
Multiple R-squared: 0.6407,	Adjusted R-squared: 0.6308
F-statistic: 64.65 on 4 and 145 DF,  p-value: < 2.2e-16