Oz Politics Blog

I have upgraded my blog to Wordpress 2.1. With one exception, the process was fairly painless. The exception was my politics test plugin. The new version of Wordpress was inserting line breaks into the questions. I had to get my fingers dirty and find out what was happening. It turned out that the auto-formatting function [wpautop() in the file ~/wp-includes/formatting.php] had been rewritten so that input XHTML tags were now treated as blocks. The fix was easy: I deleted the input keyword from the list of blocks. I am buggered if I know why they did it.

As far as I can tell, all of my other plugins are now working fine (after the odd random fix). Please let me know if you find any problems with the upgraded site.

On a different matter, when I went on holidays, I introduced a registration requirement before you can post a comment. The registration requirement significantly reduced the daily spam I receive. Consequently, I have decided to keep the requirement.

To help with the next redevelopment of the Australian Politics Test, I have added a page with an analysis of the answers to individual questions by self-reported party preference. Because some of the questions have changed over time, the analysis only considers the 1000 most recent responses from people living in Australia of voting age.

Take a look. There are some unexpected surprises.

• Question 1 — All third world debt should be cancelled — while the left splits roughly 80-20 in favour as I expected them to, the right only splits roughly 60-40 against
• Question 2 — There is too much nudity and explicit sexual material on television — another question where the left split as I expected them to, but the right less so
• Question 4 — Government should increase taxes so that it can do more for its citizens — Here the left had a higher disagreement (roughly 60-40) than I expected
• Question 5 — Abortion is murder — The Liberal party split against (66-34) was higher than I had expected
• Question 6 — People who enter Australia illegally and who claim to be a refugee should be detained for a time while their claims are assessed — The Greens only managed a disagreement of 65-35. I expected higher
• Question 7 — Governments should do more to help the small business sector — The 20-80 support split pretty well across the board, including from the Greens was unexpected — this question may need to go as it does not help discriminate political preference
• Question 8 — The government should pay people to care for their children and elderly parents — the low level of disagreement (60-40) from the Liberals surprised me — it looks like middle class welfare is a vote winner
• Question 10 — Governments should identify and regulate things that might be harmful to people — Another question where the splits are roughly the same across the party spectrum (40-60) — this question may need to go as it does not help discriminate political preference
• Question 11 — The government should do more to protect the rights of workers — Another surprise with a majority of Liberals supporting the statement (and it appears contradictory to the results for Question 3, ‘Employers need a lot of flexibility when it comes to hiring and firing staff’, which got strong support from the Liberals)
• Question 17 — For a political party, refining its policy position on the basis of opinion polls and talk-back-radio is a good thing — splits ranged from 60-40 against to 50-50 — this question may need to go as it does not help discriminate political preference
• Question 24 — The government should set quotas and fine large companies that do not employ a sufficient proportion of people with a disability or women in senior management positions — did not get the level of support from the left I was anticipating (only the Greens split for at 49-51)
• Question 25 — As a general principle, school teachers should be paid bonuses and promoted on the basis of their students’ performance — with the way the Teachers’ Union has carried on I had expected higher disagreement from the left (only 60-40) — I also expected higher agreement from the right (rather than 50-50) — this question may need to go as it does not help discriminate political preference
• Question 26 — Protecting the environment is more important than providing social welfare services to people in need — I did not expect Green voters to disagree with this proposition (roughly 60-40) — this question may need to go as it does not help discriminate political preference
• Question 32 — The Medicare scheme should be restricted to people on low incomes — With all parties opposed this question may need to go as it does not help discriminate political preference
• Question 35 — The government should establish a national filtering scheme to protect children from pornography on the Internet — I did not expect Labor to split in favour 40-60
• Question 37 — Banks should be required to provide customers with basic fee-free accounts – I did not expect the Liberals to split in favour 40-60
• Question 39 — Voting at elections should be compulsory — with broad cross party support this question may need to go as it does not help discriminate political preference

Update 25 April 2006: The questions have been tweaked and re-scored. I will review the test again at 12,000 hits.

We are approaching a milestone. The Australian Politics Test has been taken almost 10,000 times. To commemorate this occasion, Ray Hentzschel sent me some fascinating graphs of his analysis of these statistics.

Ray has looked at the left-versus-right leanings of the people who have taken the test according to their self-reported party preference. He has plotted the results, and a polynomial regression line for each of the economic, social and political values dimension of the test. However, before looking at the graphs, I will quickly recap the three dimensions.

The economic policy dimension reports the extent to which respondents think the state should be regulating the economic aspects of our lives. A negative score means you believe the state should, on more issues than not, intervene in the economic lives of its citizens. A positive score means you believe the state should be less interventionist.

The social policy dimension reports the extent to which respondents think the state should be supporting its citizens as well as protecting its citizens from making decisions that could be harmful (in social policy areas other than those covered by the traditional values dimension below). A negative score means you believe the state should, on more issues than not, intervene in the social lives of its citizens as a force for good. A positive score means you believe we are responsible enough to live and run our own lives free from excessive government intervention.

The traditional values dimension reports the extent to which you think the state should act to maintain conservative moral standards (for example in respect of abortion, divorce and drug use). A negative score means you believe the state should not overly intervene in the moral lives of its citizens. A positive score indicates you believe the state should intervene on more of these issues than not. A strongly positive score is consistent with the position adopted by the ‘Christian Right’ in Australia.

A number of points of interest emerge from these graphs.

• Most surprisingly for me, the self-reported One Nation voters were not all that far to the right in their answers to the questions - especially on economic policy where they were typically left of centre
• Next most surprising, Liberal Party voters tend to be to the left of centre when it comes to traditional values
• Family First voters tend to be middle of the road on social and economic policy, and clearly to the left of the Liberal and National Parties on these issues
• Family First and National voters tend to have similar right of centre views on traditional values
• Liberal voters tend to be slightly more right wing on economic policy issues, whereas National voters tend to be slightly more right wing on social policy issues
• There is not a lot of difference in the profiles of Democrats and Labor voters
• Green voters are much more likely than Democrat or Labor voters to hold far-left views on economic and social policy, and traditional values

I am working on a second beta test which I will release shortly. These graphs will help me refine the test.

In case you are wondering, the 100 most recent test results can be seen here.

Thank you everyone for testing the Australian Politics Test. I have progressively shaped the questions in response to the suggestions and issues you raised.

One issue particularly intrigued me. Matthew Hill made the following observation.

My reading of your [individual] results seems to show a lot of people with very high scores for the Greens. Assuming that your readership is not completely skewed, doesn’t that indicate that the test is slightly inaccurate?

It was a good question. So I added to the Australian Politics Test a page to analyse the results to date. I was expecting a left wing bias. The ANU 2004 Federal Election Online Poll exhibited a left wing bias among Internet users in the order of 5.6 per cent. But the extent of bias in the cumulative results surprised me. Of the first 2146 completed tests, 75% percent of the respondents prefer the Greens, Democrats or Labor. 66% describe their politics as far-left, left or centre-left.

Preferred party of respondents (self perception)

Political inclination of respondents (self perception)

Compared to the self-perceptions, only the economic dimension exhibits a similar distribution. The other test results are skewed to the left. The more difficult question is to determine whether the test or the self-perceptions are to blame for this left-wing bias. I will need to think about this some more.

Preferred party of respondents (test results)

Political inclination of respondents (test results)

Economic policy inclination of respondents (test results)

Social policy inclination of respondents (test results)

Traditional values inclination of respondents (test results)

A problem with classifying party preferences is handling the broad-church parties (Labor and Liberal). In order to win government, the Labor and Liberal parties seek to accommodate voters with a wide diversity of opinion. Some of their voters have views that are highly aligned with the space occupied by the Greens, the Australian Democrats or Family First. It is difficult finding questions that tell whether people prefer the minor parties or the majors. I will think about this one some more.

The graphical analysis also reveals some interesting facts. First, the next three graphs show a strong correlation between the social policy and economic economic dimensions.

Self identified Greens voters - Social by Economic

Self identified Labor Party voters - Social by Economic

Self identified Liberal Party voters - Social by Economic

Update: Opps. I ran of memory producing the graphs, so they have been dropped.

The second interesting fact is that the same correlation does not hold with traditional values. For Green voters, opposition to traditional values is strong. However, traditional values are (surprisingly) irrelevant to Liberal voters. Rather than the graphical analysis I will use a couple of tables (the graphs are available on the cumulative results page).

Traditional values inclination of 566 self reported Greens voters

Traditional values inclination of 415 self reported Liberal Party voters

I have re-written my politics test in PHP and put it on the blog. The new test will identify which of the major Australian political parties most aligns with your views. It will also tell you where your views lie on the left-right political spectrum.

I am looking for people to take the test and then post comments here on whether they agree or disagree with the findings. You may want to post your comments anonymously, especially if you are concerned about making a public statement about your political position.

If you feel courageous enough, you can also post the permalink to your results (which can be found at the bottom of the results page). Pasting the link from the top of your browser will not work. You must copy and paste the permalink from the bottom of the results page.

The most recent results can be viewed here.

Cumulative results and accuracy analyses are here

Update 23 October 2005 — I have further tweaked the questions to address the perceived left wing bias in the results.