LEO, Revisited
I've mentioned before that I'm a fan of using a "Liberty, Equality, Order" spectrum for analysing political idology, in contrast to a simplistic left/right divide. I ran into this spectrum early on at university, and so I found a political scientist's attempt, at The LEO Test, to create a program which could automatically analyse the ideology of political writing very interesting when I ran across it a year or so ago. This spectrum sees the political landscape as populated by (some combination of) three distinct ideologies which can be reflected in moderate or radical positions - Liberty (libertarians, anarchists), Equality (progressives, communists) and Order (Conservatives, fascists).
Cam over at the fairly new (and quite good) group blog, Polemica, recently wrote a post expressing his dissatisfaction at the traditional left/right divide and in the ensuing discussion I brought up the LEO test. One commentor there thought that the spectrum wouldn't properly capture the ideology of post-left/right divide, community-values based parties like People Power so I decided to give the program supplied by Jonathon York another try (my last attempt at doing this showed the limits of the test as applied to judicial writings, more than disproving it more comprehensively I think). So I went to their web site and collated all the relevant text I could find, from their "about" and policies documents. Just to give an overview, here's what the party has to say for themselves (in broken html on the site):
Now, just looking at that you get the strong underlying message of "society is only as strong as its weakest member" which is a paradigm Equality value. This impression is borne out in a textual analysis of the 5000-odd words I put together from the web site:
The bars represent percentages of ideological keywords (ie 55% of keywords used reflected Equality values). Now, it's a relatively small sample of words but I think the analysis has hit on the essential ideology of the party. It's strongly egalitarian with no secondary preference for either order or liberty, given that order/conservative values of "family" and "regulation" co-exist with a respect for the dynamism and freedom of small enterprise.
One of the Polemica bloggers suggested I try the analysis on the Labor and Liberal party platforms, these being the two dominant political parties in Australia. The Liberal party likes to identify itself with small-l liberalism values, but also has a strong socially/traditionally conservative element in it which is in tension with these values. The Labor party has historical roots in the union movement although moved away from those roots somewhat when it was in power in the 80s and 90s by governing in a style in some way foreshadowed the "Third Way" of Clinton/Blair.
These platform documents aren't ideal source material for two reasons. First, while the Labor platform is very large (100,000) which is good, the Liberal one is quite small (4,000). In addition, potential nuances in the party's policies in different areas such as law and order, social areas, welfare, economic regulation, industrial relations etc. aren't reflected in such a broad brush analysis. Still, given that I don't have time to do a full analysis I decided to see what the numbers brought up. I excluded one of the data source files from the analsyis because the word "Liberal" was included as an ideological key word, which is obviously inappropriate in assessing the Liberal party. This left me with three source files, and these are the averages of the results.
Some might be surprised that these aren't exactly the same, but I do think that these graphs are reasonably reflective of reality. The Liberal party, if you read its platform (I didn't, I just scanned, but you know what I mean) likes to fancy itself as a bastion of traditional liberalism and that preference is very strongly captured. The document also emphasises the "Australian" value of egalitarianim in its emphasis on "opportunity for all", which I think explains the relatively high E score. Now, the Liberal party's actions frequently belie their rhetoric, but I do think the simple test here is picking up the ideological flavour of their rhetoric quite well.
Likewise for Labor. The emphasis is much more clearly on Equality, with no strong secondary preference, although a slight leaning towards order which is not at all incosistent with the Labor party (it has only weak committment to civil liberties issues, for instance, and is unafraid to regulate private business quite heavily in some areas). I think it's also pertinant that neither party had particularly strong ideological preferences for their primary value (the highest marker was about 47% each) which is to be expected from relatively moderate mainstream parties. All in all, I think these rough analysis confirm the usefulness of the blunt textual analysis of The LEO Test in analysing the ideological content of political writing.
Cam over at the fairly new (and quite good) group blog, Polemica, recently wrote a post expressing his dissatisfaction at the traditional left/right divide and in the ensuing discussion I brought up the LEO test. One commentor there thought that the spectrum wouldn't properly capture the ideology of post-left/right divide, community-values based parties like People Power so I decided to give the program supplied by Jonathon York another try (my last attempt at doing this showed the limits of the test as applied to judicial writings, more than disproving it more comprehensively I think). So I went to their web site and collated all the relevant text I could find, from their "about" and policies documents. Just to give an overview, here's what the party has to say for themselves (in broken html on the site):
- families: the foundation of society but unrepresented by any broad, mainstream movement
- consumers: our two main parties represent employers and employees, but not consumers
- people with disabilities, chronic and mental illnesses and their families/carers: the most invisible and vulnerable Australians
- the aged: regarded as not glamorous, important or productive in our culture
- volunteers in communities: who are the glue in society but are unrepresented in any of our halls of power
- small businesses and independent owners: the backbone of our economy and employment but overlooked by governments
- individuals and communities who practice self-help: whose voices are rarely heard
Now, just looking at that you get the strong underlying message of "society is only as strong as its weakest member" which is a paradigm Equality value. This impression is borne out in a textual analysis of the 5000-odd words I put together from the web site:
The bars represent percentages of ideological keywords (ie 55% of keywords used reflected Equality values). Now, it's a relatively small sample of words but I think the analysis has hit on the essential ideology of the party. It's strongly egalitarian with no secondary preference for either order or liberty, given that order/conservative values of "family" and "regulation" co-exist with a respect for the dynamism and freedom of small enterprise.
One of the Polemica bloggers suggested I try the analysis on the Labor and Liberal party platforms, these being the two dominant political parties in Australia. The Liberal party likes to identify itself with small-l liberalism values, but also has a strong socially/traditionally conservative element in it which is in tension with these values. The Labor party has historical roots in the union movement although moved away from those roots somewhat when it was in power in the 80s and 90s by governing in a style in some way foreshadowed the "Third Way" of Clinton/Blair.
These platform documents aren't ideal source material for two reasons. First, while the Labor platform is very large (100,000) which is good, the Liberal one is quite small (4,000). In addition, potential nuances in the party's policies in different areas such as law and order, social areas, welfare, economic regulation, industrial relations etc. aren't reflected in such a broad brush analysis. Still, given that I don't have time to do a full analysis I decided to see what the numbers brought up. I excluded one of the data source files from the analsyis because the word "Liberal" was included as an ideological key word, which is obviously inappropriate in assessing the Liberal party. This left me with three source files, and these are the averages of the results.
Some might be surprised that these aren't exactly the same, but I do think that these graphs are reasonably reflective of reality. The Liberal party, if you read its platform (I didn't, I just scanned, but you know what I mean) likes to fancy itself as a bastion of traditional liberalism and that preference is very strongly captured. The document also emphasises the "Australian" value of egalitarianim in its emphasis on "opportunity for all", which I think explains the relatively high E score. Now, the Liberal party's actions frequently belie their rhetoric, but I do think the simple test here is picking up the ideological flavour of their rhetoric quite well.
Likewise for Labor. The emphasis is much more clearly on Equality, with no strong secondary preference, although a slight leaning towards order which is not at all incosistent with the Labor party (it has only weak committment to civil liberties issues, for instance, and is unafraid to regulate private business quite heavily in some areas). I think it's also pertinant that neither party had particularly strong ideological preferences for their primary value (the highest marker was about 47% each) which is to be expected from relatively moderate mainstream parties. All in all, I think these rough analysis confirm the usefulness of the blunt textual analysis of The LEO Test in analysing the ideological content of political writing.
posted by Jeremy at
5:19 PM
4 Comments:
Interesting stuff.
I think the point that what an organisation purports to be about in its platform or "mission statement" is not always what it is really all about is a key one here.
Although in this case the test seems to have been reasonably accurate.
By
Guy, at 5:24 PM
Cool.
Because its really late, i haven't read the whole thing. Seems to me that Labor comes out as closer to People Power than to the Liberals. In fact ot would suggest that the major difference between PP and Labor is degree of equality. Maybe this is true. I havent really been able to sound them out yet.
But it's this page which set me thinking
http://www.peoplepower.org.au/leftrightPeoplePower.htm
By
Mikey, at 1:44 AM
I think the point that what an organisation purports to be about in its platform or "mission statement" is not always what it is really all about is a key one here.
That's true, but as the guy at the LEO Test has noted (Jonathon York), the language people use will determine which segment of the voting population is more naturally attracted to them so you would expect some overlap between their language and actions (although this might be more pronounced in the US where ideological differences are more stark). If you don't walk the walk you're talking, you're likely to get disillusioned supporters.
@Mikey
Yes, that's pretty much it. PP has ~55% E cf 47% which isn't surprising. Mainstream parties have to be more moderate, and 55% is a very strong E score (Jonathon's rule of thumb was that anything above 2/3s was borderline extreme, ie communist/fascist/anarchist).
By
Jeremy, at 6:54 AM
Sorry to post so late, but your application is certainly enlightening. In the United States programmatic egalitarians are often called liberals, possibly because of the shared concern for both liberty and equality by people such as Thomas Hill Green, John Dewey and Franklin Roosevelt. In order to rebalance out the source file that employs the word "liberal" as an indicator, all you would need to do is remove the words "libertarian" and "conservative" as well as "liberal", and you'll still have a balanced table. The more I explore this topic, the more I realize the misnomers of American politics. The model itself suggests libertarianism, egalitarianism and establishmentarianism as more fundamental than the terms habitually applied, which ultimately depend on context.
I've also found, through another application of the model, that one may separate ideological preference from ideological anti-preference, treating them as separate ideological orientations. Some people identify their politics according to what they oppose rather than what they support, and as a result, ideological landscapes may be described spatially in the final analysis.
By
jonathon, at 6:14 AM
Post a Comment
<< Home