Pournelle's left/right
Michael N. LeVine
how...@brazee.net wrote:
> > Has Anyone seen Jerry Pournelle's take on Left/Right?
>
> > His thesis is that Left/Right hasn't worked since the French Revolution
> > (The
> > Estates General entered from the left in order of precedence. Those with
> > the most to lose ended up on the right.) Currently, there is a total
> > mess,
> > i.e. Communists in the US are left, but in Russia they are right.
>
> > He proposed a two axis system. The horizontal axis is "What do you think
> > of
> > government?" and runs from "Absolute Evil" though "Like Fire it is
> > needed,
> > but must be kept small", "A tool, like any other, not necessarily the
> > best
> > for any particular job", "The preferred tool for any job" to "An Absolute
> > Good".
>
> The trouble with this is that every time your party gains power you move
> rapidly away from this position. Example:
> The Bush administration is trying to change Washington's assisted suicide
> law.
>
>
> > The vertical axis is "What do you think of human progress?" and run from
> > "We
> > are degenerations from a noble ideal" through "Most things where better
> > in
> > the past", "Things never really change", "Some things are better, some
> > are
> > not as good" to "We are marching to a glorious future".
>
> Again - this isn't that simple. A conservationist doesn't want to change
> things. Neither does a conservative. Or a
> preservationist (or a preservative?). A religious conservative can be a
> political radical. An environmentalist can
> come with desired huge governmental changes.
>
> > In his system, at least the Communists and Fascists don't map as
> > opposites.
>
> They both call themselves socialists, believing the state is much more
> important than individuals.
I remember seeing it in an old "Destinies" (packed away where
I can't get at it at this time). This sounds like an updated
version -- does any one know where I could find a copy???
--
Michael LeVine - mle...@redshift.com
"Thirty days hath September, April, June and November.
All the rest have thirty one except for Gypsy Rose Lee
and every one knew what she had" - Mel Blanc
John David Galt
>
>> His thesis is that Left/Right hasn't worked since the French Revolution
>> (The
>> Estates General entered from the left in order of precedence. Those with
>> the most to lose ended up on the right.) Currently, there is a total
>> mess,
>> i.e. Communists in the US are left, but in Russia they are right.
>
>> He proposed a two axis system. The horizontal axis is "What do you think
>> of
>> government?" and runs from "Absolute Evil" though "Like Fire it is
>> needed,
>> but must be kept small", "A tool, like any other, not necessarily the
>> best
>> for any particular job", "The preferred tool for any job" to "An Absolute
>> Good".
>
> The trouble with this is that every time your party gains power you move
> rapidly away from this position. Example:
> The Bush administration is trying to change Washington's assisted suicide
> law.
There's a better two-axis system, that doesn't have this problem:
http://www.self-gov.org/wspq.html
Stan
> The trouble with this is that every time your party gains power you move
> rapidly away from this position. Example:
> The Bush administration is trying to change Washington's assisted suicide
> law.
You're close...it's Oregon's law the Bush admin opposes, even though
republican orthodoxy champions "state's rights".
Stan.
Michael Ward
"Stan" <stan...@ptld.us*N0SPPAMM*west.net> wrote in message
news:IfdJ7.962$sb6.1...@news.uswest.net...
Where is the contradiction between these two points of view?
Republicans have never advocated that the States should
have authority over the Federal Government. They've simply
advocated leaving more matters in the hands of the states then
Democrats generally do. Assisted suicide is simply one
of many issues that Republicans don't want to leave up to
the states. Whether or not they are right about this issue,
they aren't being hypocritical.
Mike
Michael Ward
"John David Galt" <j...@diogenes.sacramento.ca.us> wrote in message
news:3BF55E7E...@diogenes.sacramento.ca.us...
The diagram is interesting but the quiz to determine
where you go is slanted to make most people
come across as authoritarian.
For instace, one of the questions is, "abolish all taxes"
(Yes/No/Maybe). The centris position on a question
like this would be no, since only people on the
political fringe advocate an end to all taxation.
But the quiz counts maybe as the centrist response
and a "no" will push you toward leftist/authoritarian
on the diagram.
The Same problem with several other questions.
Mike
David Cowie
>
> There's a better two-axis system, that doesn't have this problem:
> http://www.self-gov.org/wspq.html
And a similar system at www.politicalcompass.org
--
David Cowie
There is no _spam in my address.
"You had to do WHAT with your seat?"
John David Galt
>> http://www.self-gov.org/wspq.html
> The diagram is interesting but the quiz to determine
> where you go is slanted to make most people
> come across as authoritarian.
Most people you know probably ARE authoritarian. But take a look at
the site's table of results. They've gotten a huge sample size by now.
Michael Ward
A huge extremely biased sample.
On the web it's easy to find people who'll say,
"There should be no taxes, the draft is always
wrong, having a minimum wage law is bad,
government subsidies are wrong, governemets
shouldn't supply foreign aid, drugs should be
legalized, politcal borders should be open" But
in the real world these people are a harder to come
by.
Mike
Erik Max Francis
> The diagram is interesting but the quiz to determine
> where you go is slanted to make most people
> come across as authoritarian.
>
> For instace, one of the questions is, "abolish all taxes"
> (Yes/No/Maybe). The centris position on a question
> like this would be no, since only people on the
> political fringe advocate an end to all taxation.
> But the quiz counts maybe as the centrist response
> and a "no" will push you toward leftist/authoritarian
> on the diagram.
Indeed. It is also just plain ambiguous; some questions are totally
unclear, such as "People should be free to come and go across borders"
-- what borders? Intercounty, interstate, intercountry?
It's a pretty slanted quiz.
--
Erik Max Francis / m...@alcyone.com / http://www.alcyone.com/max/
__ San Jose, CA, US / 37 20 N 121 53 W / ICQ16063900 / &tSftDotIotE
/ \ Laws are silent in time of war.
\__/ Cicero
Esperanto reference / http://www.alcyone.com/max/lang/esperanto/
An Esperanto reference for English speakers.
Michael Ward
"Erik Max Francis" <m...@alcyone.com> wrote in message
news:3BF57A35...@alcyone.com...
> Michael Ward wrote:
>
> > The diagram is interesting but the quiz to determine
> > where you go is slanted to make most people
> > come across as authoritarian.
> >
> > For instace, one of the questions is, "abolish all taxes"
> > (Yes/No/Maybe). The centris position on a question
> > like this would be no, since only people on the
> > political fringe advocate an end to all taxation.
> > But the quiz counts maybe as the centrist response
> > and a "no" will push you toward leftist/authoritarian
> > on the diagram.
>
> Indeed. It is also just plain ambiguous; some questions are totally
> unclear, such as "People should be free to come and go across borders"
> -- what borders? Intercounty, interstate, intercountry?
>
borders, but it could just as easily be asking
if you can travel state to state.
"No papers?"
"No papers."
"I think I shall need two wives"
Mike
David Cowie
>
> The diagram is interesting but the quiz to determine
> where you go is slanted to make most people
> come across as authoritarian.
I disagree ... My result was Left/Liberal, which was came as a bit of a
surprise.
Geoffrey A. Landis
>
> "John David Galt" <j...@diogenes.sacramento.ca.us> wrote in message
> news:3BF572C9...@diogenes.sacramento.ca.us...
> > >> There's a better two-axis system, that doesn't have this problem:
> > >> http://www.self-gov.org/wspq.html
> >
> > > The diagram is interesting but the quiz to determine
> > > where you go is slanted to make most people
> > > come across as authoritarian.
> >
> > Most people you know probably ARE authoritarian. But take a look at
> > the site's table of results. They've gotten a huge sample size by now.
>
> A huge extremely biased sample.
Yes, I just went over and looked at the quiz. The quiz is well designed
to detect libertarian ideas (the "yes" answer of each question is the
stock libertarian position), but it is poorly phrased to distinguish or
even detect liberal or conservative ideas.
--I just went back to check "yes" on each question, to verify that these
answers give a hard-over libertarian position. People doing this will
bias their results.
(for what it's worth, this quiz put my score on the libertarian side of
the centrist/libertarian line, exactly halfway between left and right
(and by coincidence precisely in the center of the area marked
"libertarian".) My personal self-evaluation would have put me in the
centrist position, tilted somewhat toward the libertarian position.).
--
Geoffrey A. Landis
http://www.sff.net/people/geoffrey.landis
Just published: IMPACT PARAMETER (and other quantum realities)
http://www.goldengryphon.com/ip-frame.html
Erik Max Francis
> Good point. I just assumed it meant national
> borders, but it could just as easily be asking
> if you can travel state to state.
>
> "No papers?"
>
> "No papers."
"No papers. State to state," with our beloved Connery's slurring of the
_st_ into a _sht_. :-)
Erik Max Francis
> --I just went back to check "yes" on each question, to verify that
> these
> answers give a hard-over libertarian position. People doing this will
> bias their results.
No voluntary quiz available on the Web is going to be even remotely
unbiased anyway. Anyone taking stock in the validity of such quiz
results -- for _anything_ -- needs a serious reality check.
Reverend Sean O'Hara
>
> "John David Galt" <j...@diogenes.sacramento.ca.us> wrote in message
> news:3BF55E7E...@diogenes.sacramento.ca.us.
> > http://www.self-gov.org/wspq.html
>
> The diagram is interesting but the quiz to determine
> where you go is slanted to make most people
> come across as authoritarian.
>
libertarian and centrist; authoritarian ranks second to last, behind
right-conservative.
> For instace, one of the questions is, "abolish all taxes"
> (Yes/No/Maybe). The centris position on a question
> like this would be no, since only people on the
> political fringe advocate an end to all taxation.
> But the quiz counts maybe as the centrist response
> and a "no" will push you toward leftist/authoritarian
> on the diagram.
>
The questions are badly phrased (probably on purpose). For example, "Drug
laws do more harm than good. Repeal them," actually contains two questions.
I agree that drug laws do more harm than good, but I don't think they
should be universally repealed. Even with drugs like marijuana, which I
think should be legal, I believe there should still be laws to govern the
sale thereof. I had a coworker once who explained to me her plan to
legalize drugs, which would've required users to take psychological tests,
register with the police, and keep ration cards limiting how much they
could buy per week. Yet the way the question is phrased here, she would've
answered "yes" or "maybe" which inclines her to the libertarian corner
or the graph.
And I'm not even going to touch on how biased and badly phrased,
"Government should not control radio, TV, the press or the Internet" is.
--
Reverend Sean O'Hara
You too can be an ordained minister: http://www.ulc.org
Culture Editor for Expulsion: http://www.expulsion.org (new & improved)
"So what state is Wales in?" - G. W. Bush (quoted by Charlotte Church)
John David Galt
>>>> http://www.self-gov.org/wspq.html
>>> The diagram is interesting but the quiz to determine
>>> where you go is slanted to make most people
>>> come across as authoritarian.
>> Most people you know probably ARE authoritarian. But take a look at
>> the site's table of results. They've gotten a huge sample size by now.
> A huge extremely biased sample.
>
> On the web it's easy to find people who'll say,
> "There should be no taxes, the draft is always
> wrong, having a minimum wage law is bad,
> government subsidies are wrong, governemets
> shouldn't supply foreign aid, drugs should be
> legalized, politcal borders should be open" But
> in the real world these people are a harder to come
> by.
Maybe the web isn't representative of the whole country or world; but
then, neither are the traditional media. The difference is that on the
web, you can always start your own "channel" rather than be excluded.
To me, that's a very good thing. Naturally, the old-style publishers
who once got to define what's "normal for society" tend to disagree.
John David Galt
>
> Michael Ward wrote:
> >
> > "John David Galt" <j...@diogenes.sacramento.ca.us> wrote in message
> > news:3BF572C9...@diogenes.sacramento.ca.us...
> > > >> There's a better two-axis system, that doesn't have this problem:
> > > >> http://www.self-gov.org/wspq.html
> > >
> > > > The diagram is interesting but the quiz to determine
> > > > where you go is slanted to make most people
> > > > come across as authoritarian.
> > >
> > > Most people you know probably ARE authoritarian. But take a look at
> > > the site's table of results. They've gotten a huge sample size by now.
> >
> > A huge extremely biased sample.
>
> Yes, I just went over and looked at the quiz. The quiz is well designed
> to detect libertarian ideas (the "yes" answer of each question is the
> stock libertarian position), but it is poorly phrased to distinguish or
> even detect liberal or conservative ideas.
>
> --I just went back to check "yes" on each question, to verify that these
> answers give a hard-over libertarian position. People doing this will
> bias their results.
>
> (for what it's worth, this quiz put my score on the libertarian side of
> the centrist/libertarian line, exactly halfway between left and right
> (and by coincidence precisely in the center of the area marked
> "libertarian".) My personal self-evaluation would have put me in the
> centrist position, tilted somewhat toward the libertarian position.).
I'm sure it would be more balanced if they'd made it 100 questions
instead of 10, but a lot fewer people would answer it. TANSTAAFL.
Old Toby
>
>
> The diagram is interesting but the quiz to determine
> where you go is slanted to make most people
> come across as authoritarian.
Ah, but the questions usually take the form of "here's
an argument for a libertarian position, do you agree
with it." or "here's what's wrong with the status quo,
don't you think we should change it?" By presenting
a pro-libertarian argument while asking the question,
they slant people toward the Libertarian position,
especially when people answer with "sounds OK" or
"nah" rather than "Let me think about all that implies".
One thing that I did notice was that there was a real
disconnect between the description of the economic
axis on my results (I'm left-liberal, BTW) and what
was really being asked in the question. According
to the results page, left-liberals prefer
"central decision-making on economics".
But the five "economic" issues involved are business
subsidies, free trade, the minimum wage, taxes, and
foreign aid.
None of these involves "central decision making", at
worst they just involve limiting the decisions that
are made by individuals, and there's a big difference
between the two.
Old Toby
Least Known Dog on the Net
Mike Van Pelt
Michael N. LeVine <mle...@redshift.com> wrote:
>I remember seeing it in an old "Destinies" (packed away where
>I can't get at it at this time). This sounds like an updated
>version -- does any one know where I could find a copy???
I recall seeing it on the Baen Books web page...
I think I bookmarked it...
Yep, here it is.
<http://www.baen.com/chapters/axes.htm>
I think this is far superior to the "Nolan Chart" that the
Libertarian Party pushes. The Nolan chart comes with ten
leading essay questions to which it demands yes/no answers, and
purports to jam you into a pigeonhole based on your answers to
those questions.
I *DETEST* "questionaires" which demand yes/no answers to
essay questions.
Also, the variables defining the two axes of the Nolan chart are
not independant. "personal freedom" and "economic freedom" are
closely interrelated.
I believe Dr. Pournelle's two-axis chart was from his Political
Science PhD dissertation. I'd love to get my hands on that...
--
Have you noticed that, when we were young, we were told | Mike Van Pelt
that "everybody else is doing it" was a really stupid | m...@calweb.com
reason to do something, but now it's the standard reason | KE6BVH
for picking a particular software package? -- Barry Gehm
Michael Ward
More question's would help, but it could also be
improved a lot with ten good questions. Many
of these questions were meaningless. For instance
the fact that someone does not oppose all taxation
tells you almost nothing about their political ideology.
Mike
Klyfix
writes:
>
>No voluntary quiz available on the Web is going to be even remotely
>unbiased anyway.
But the involuntary quizes, they're unbiased. :)
>Anyone taking stock in the validity of such quiz
>results -- for _anything_ -- needs a serious reality check.
>
Of course. The folk producing the quiz are most likely looking for
recruits; "Look, you're a Libertarian!!" or whatever.
V. S. Greene : kly...@aol.com : Boston, near Arkham...
Eckzylon: http://m1.aol.com/klyfix/eckzylon.html
RPG and SF, predictions, philosophy, and other things.
I flirt with danger; I open mail.
Klyfix
<j...@diogenes.sacramento.ca.us> writes:
Really? I get classes as a "left-liberal" on that test (for all that I've on
occasion been just about called a Commie Authoritarian by Libertarians)
and I think my friends generally would be also.
Klyfix
writes:
>>
>The questions are badly phrased (probably on purpose). For example, "Drug
>laws do more harm than good. Repeal them," actually contains two questions.
>I agree that drug laws do more harm than good, but I don't think they
>should be universally repealed. Even with drugs like marijuana, which I
>think should be legal, I believe there should still be laws to govern the
>sale thereof. I had a coworker once who explained to me her plan to
>legalize drugs, which would've required users to take psychological tests,
>register with the police, and keep ration cards limiting how much they
>could buy per week. Yet the way the question is phrased here, she would've
>answered "yes" or "maybe" which inclines her to the libertarian corner
>or the graph.
>
I had problems with that question also; while I think many recreational
drugs should be legal (for all that I think they're bad things) I'd not want
to repeal laws against minors having access to them. Question doesn't
give that option.
>And I'm not even going to touch on how biased and badly phrased,
>"Government should not control radio, TV, the press or the Internet" is.
>
Depends on the definition of "control." I suppose. Is moderate regulation
"control"? Not to me, really, but for Libertarians I suppose it is.
Jordan S. Bassior
>Indeed. It is also just plain ambiguous; some questions are totally
>unclear, such as "People should be free to come and go across borders"
>-- what borders? Intercounty, interstate, intercountry?
Interdimensional!
--
Sincerely Yours,
Jordan
--
Jordan S. Bassior
"Do you believe in subjecting the world to the terrors of nuclear annihilation,
or do you believe that America should be peaceful" vs.
"Do you believe in leaving us naked to attack by hostile nuclear-armed foreign
powers, or do you believe that America should be prepared"
when asking about strategic weapons spending.
David Johnston
>
> I had problems with that question also; while I think many recreational
> drugs should be legal (for all that I think they're bad things) I'd not want
> to repeal laws against minors having access to them. Question doesn't
> give that option.
Actually it does. If your response is more complicated than "yes/no", you
are supposed to pick "maybe". Do a lot of that because you realise that these
questions are more complex than a doctrinaire ideological stance would have
it and you'll come out as a centrist.
Johnny1A
The trouble with this poll is that it is not complicated enough (I
hate to say that about anything to do with politics.) Changing just
one answer can shift you all over their map. And the 'maybe' answers
cover huge swaths of territory.
Shermanlee
Bill Snyder
>In article <3BF58982...@alcyone.com>, Erik Max Francis <m...@alcyone.com>
>writes:
>
>>
>>No voluntary quiz available on the Web is going to be even remotely
>>unbiased anyway.
>
>But the involuntary quizes, they're unbiased. :)
An involuntary quiz would seem likely to be free of _pro-Libertarian_
bias.
--
Bill Snyder [This space unintentionally left blank.]
Arnold Bailey
> > Has Anyone seen Jerry Pournelle's take on Left/Right?
>
> > His thesis is that Left/Right hasn't worked since the French Revolution
> > (The
> > Estates General entered from the left in order of precedence. Those
with
> > the most to lose ended up on the right.) Currently, there is a total
> > mess,
> > i.e. Communists in the US are left, but in Russia they are right.
>
> > He proposed a two axis system. The horizontal axis is "What do you
think
> > of
> > government?" and runs from "Absolute Evil" though "Like Fire it is
> > needed,
> > but must be kept small", "A tool, like any other, not necessarily the
> > best
> > for any particular job", "The preferred tool for any job" to "An
Absolute
> > Good".
You can read the article Jerry wrote at
http://www.baen.com/chapters/axes.htm
with the graph
John David Galt
> You can read the article Jerry wrote at
>
> http://www.baen.com/chapters/axes.htm
>
> with the graph
It begins with a major historical error. The "left/right axis" did
not start in the French National Assembly, it goes back at least to
the Roman Senate (plebes on the left, patricians on the right).
Margaret Young
Every standard American Political Text lists seating in the French
National Assembly as the reason for the modern-day designation of
right and left to the different poles of political ideology. (Rather
than rely on my memory I spent the last 15 minutes yanking all the
polisci texts I have [dozens] from the shelves and checking them.
While it may be true that the seating of other assemblies was known to
some there is general agreement that it was the seating in the French
National Assembly that directly influenced the modern day usage of the
terms left and right.
By the way, when in the history of the Roman Senate are you
referencing? For quite a period of time _no_ plebians were
allowed/named to the Senate and then those who were were first raised
to patrician rank _then_ named to the senate. For quite a long time
there was not enough plebians in the Senate to justify any seating
pattern at all. None of the references I could find quickly cite your
explanation of the seating in the Roman Senate.
Margaret
-----------------------------------------------------------------------------
Come the apocalypse there will be cockroaches, Keith Richards and the
faint smell of cat pee.
-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
Check out our new Unlimited Server. No Download or Time Limits!
-----== Over 80,000 Newsgroups - 19 Different Servers! ==-----
Geoffrey A. Landis
>
> More question's would help, but it could also be
> improved a lot with ten good questions. Many
> of these questions were meaningless. For instance
> the fact that someone does not oppose all taxation
> tells you almost nothing about their political ideology.
It doesn't tell you anything about the left-right axis, but it does well
in diagnosing if you're a hard-over libertarian.
If I were actually interested, I could try the test with only that
question answered "no", and see the result.
Matthew DeBell
>> There's a better two-axis system, that doesn't have this problem:
>> http://www.self-gov.org/wspq.html
>
>The trouble with this poll is that it is not complicated enough (I
>hate to say that about anything to do with politics.) Changing just
>one answer can shift you all over their map. And the 'maybe' answers
>cover huge swaths of territory.
Exactly. Two axes are better than one, but two axes are still insufficient
to accurately characterize the political orientations of most people. This
is because most people's attitudes on specific issues are not very strongly
constrained by their general attitudes regarding concepts like active
government or the possibility of social progress. IOW, knowing where
someone falls on a two-axis system like this is only of modest value if you
want to predict the candidates for whom they will vote or whether they
support specific policies like a higher tax on tobacco, or privatizing
social security, or the death penalty.
--
Matthew DeBell
Paul Austin
"Old Toby" wrote
> Michael Ward wrote:
> >
> >
> > The diagram is interesting but the quiz to determine
> > where you go is slanted to make most people
> > come across as authoritarian.
>
> to the results page, left-liberals prefer
> "central decision-making on economics".
>
> But the five "economic" issues involved are business
> subsidies, free trade, the minimum wage, taxes, and
> foreign aid.
>
> None of these involves "central decision making", at
> worst they just involve limiting the decisions that
> are made by individuals, and there's a big difference
> between the two.
Not really. Subsidy, tariff, tax, wage and price controls are the classic
methods by which centralized rather than market decisions are enforced. To
limit the taxonomic class "central decision making" to states that employ
truncheon and bullet freely misses the bulk of what central planning _does_
go on.
--
"Whatever happens will be for the worse and
therefore it is in our interest that
as little should happen as possible."
Paul F Austin
pau...@digital.net
Matt McIrvin
"Matthew DeBell" <m...@attglobal.net> wrote:
> Exactly. Two axes are better than one, but two axes are still insufficient
> to accurately characterize the political orientations of most people. This
> is because most people's attitudes on specific issues are not very strongly
> constrained by their general attitudes regarding concepts like active
> government or the possibility of social progress.
Ultimately these things are a lot like personality typing schemes (four
humors, Myers-Briggs, Enneagram...) only somewhat more focused. They're
fun for characterizing what we already know about people, but to call
them "scientific" as Jim Baen did is a little odd.
I do think that Pournelle's two-axis scheme is more interesting than
the Nolan one, in which the traditional American left-right spectrum
runs across the diagonal like the main sequence in a
Hertzsprung-Russell diagram, and most Americans who are not Libertarian
Party or fascists are just somewhere in there. Pournelle's does a
particularly good job of illustrating the split between cultural
conservatives and laissez-faire conservatives, and between anti-WTO
kids and mainstream Democrats. I'm not sure that it really has much
predictive power, though.
> IOW, knowing where
> someone falls on a two-axis system like this is only of modest value if you
> want to predict the candidates for whom they will vote or whether they
> support specific policies like a higher tax on tobacco, or privatizing
> social security, or the death penalty.
Except that very doctrinaire people and politicians seeking party
support will actively fit themselves into party molds-- but in that
case a traditional left-right scale is as good as anything, because
it's probably the mold they're thinking of.
--
Matt McIrvin
Alan McIntire
> On Friday 16 November 2001 18:44, John David Galt wrote:
> >
> > There's a better two-axis system, that doesn't have this problem:
> > http://www.self-gov.org/wspq.html
>
The self-government.org site is quite biased. The questions seem to be
constructed to elicit libertarian responses, no doubt because the site
is constructed by a libertarian group anxious to promulgate their
ideas.
The www.politicalcompass.org site is much better, though I felt like
checking
"neither agree nor disagree" a number of times, when the option was
not available. Still, the site placed my political philosophy about
where I would
rate myself, somewhat right-libertarian.
Michael Ward
"Matthew DeBell" <m...@attglobal.net> wrote in message
news:3bf72...@news1.prserv.net...
> It was written ...
>
> >> There's a better two-axis system, that doesn't have this problem:
> >> http://www.self-gov.org/wspq.html
> >
> >The trouble with this poll is that it is not complicated enough (I
> >hate to say that about anything to do with politics.) Changing just
> >one answer can shift you all over their map. And the 'maybe' answers
> >cover huge swaths of territory.
>
>
> Exactly. Two axes are better than one, but two axes are still
insufficient
> to accurately characterize the political orientations of most people.
Two axes are better than one if each axis measures
completely different things. An additional problem
here is than there isn't always a clear cut distinction
between what is an economic issue and what is a
social issue.
I still like this method of categorizing political idealogies
but ut is extremely flawed.
Mike
Matthew DeBell
>
>Exactly. Two axes are better than one, but two axes are still insufficient
>to accurately characterize the political orientations of most people.
Actually the weakness is at least as much a matter of precision as accuracy.
--
MD
pervect
"Alan McIntire" <mcin...@earthlink.net> wrote in message
news:48658b64.01111...@posting.google.com...
> The www.politicalcompass.org site is much better, though I felt like
> checking
> "neither agree nor disagree" a number of times, when the option was
> not available. Still, the site placed my political philosophy about
> where I would
> rate myself, somewhat right-libertarian.
Unfortunatley, they forgot the political orientation for people who are
concerned about privacy and won't answer questionaires on the internet. But
it was somewhat interesting clicking through the questions (without filling
them out) to eventually get to their classification scheme.
Craig S. Richardson
<geoffre...@sff.net> wrote:
>Michael Ward wrote:
>> A huge extremely biased sample.
>
>Yes, I just went over and looked at the quiz. The quiz is well designed
>to detect libertarian ideas (the "yes" answer of each question is the
>stock libertarian position), but it is poorly phrased to distinguish or
>even detect liberal or conservative ideas.
It does very well at detecting black-white along one axis. The other
axis, and any shades of gray, somewhat less well.
>--I just went back to check "yes" on each question, to verify that these
>answers give a hard-over libertarian position. People doing this will
>bias their results.
Heh. <half-serious> I think the lifeguard should come out right now
and remove anyone who legitimately answered an unequivocal "yes" _or_
"no" to all questions from the gene pool. </half-serious>
>(for what it's worth, this quiz put my score on the libertarian side of
>the centrist/libertarian line, exactly halfway between left and right
>(and by coincidence precisely in the center of the area marked
>"libertarian".) My personal self-evaluation would have put me in the
>centrist position, tilted somewhat toward the libertarian position.).
I was similar, slightly more "left" and less "center". Which will
amuse actual leftists no end, and surprise and alarm anyone who knew
me before I was 25.
--Craig
--
David Collins from Burnley: 70K pounds
Luke Weaver from Spurs: 500K pounds
Matthew Etherington from Grasshoppers-Zurich: 1.2M pounds
Leyton Orient 1-0 St. Mirren in the 2003 UEFA Cup Final: Priceless
Nyrath the nearly wise
that hoary old game Dungeons & Dragons.
You can be classifed "lawful" or "chaotic" while at the same time
you can be classified "good" or "evil".
For the first, answer the question: Does the good of the many
outweight the good of the few (or the one)?
Yes = lawful, no = chaotic.
For the second, answer the question: Do the ends justify the means?
Yes = evil, no = good.
Chris Byler
Well, there are both precision and accuracy problems in the testing,
but the main problem with a two-axis classification is that it smashes
information. _Actual_ political beliefs are somewhere in a multi-axis
manifold (I wouldn't even try to estimate the number of dimensions
offhand), and projecting that into a two-dimensional classification
will necessarily screw with the nearness metric.
--
Chris Byler cby...@vt.edu
Kubera: "It occurred to me that Sam would be the number one suspect,
except for the fact that he was dead."
Sam: "I had assumed that to be sufficient defense against detection."
-- Roger Zelazny, _Lord of Light_
Jordan S. Bassior
>Well, there are both precision and accuracy problems in the testing,
>but the main problem with a two-axis classification is that it smashes
>information. _Actual_ political beliefs are somewhere in a multi-axis
>manifold (I wouldn't even try to estimate the number of dimensions
>offhand), and projecting that into a two-dimensional classification
>will necessarily screw with the nearness metric.
Jerry Pournelle was, however, offering this as an improvement upon the
_one_-axis system currently in wide use, which smashes even _more_ information.
James S. Battista
> Well, there are both precision and accuracy problems in the testing,
> but the main problem with a two-axis classification is that it smashes
> information. _Actual_ political beliefs are somewhere in a multi-axis
> manifold (I wouldn't even try to estimate the number of dimensions
> offhand), and projecting that into a two-dimensional classification
> will necessarily screw with the nearness metric.
People have done this for Congress, sort of. Poole and Rosenthal's NOMINATE
procedure (voteview.uh.edu) looks at roll-call votes and finds the
constellations of ideal points and alternatives that are most likely given
the actual votes, in however many dimensions you care to look for (me, I've
never run it up past 5). It can't tell you what the dimensions mean, but
you can usually get a sense of that by looking at the ordering and seeing
who's on each end.
It turns out that you can correctly classify ~85% of votes with one
dimension (which turns out to look basically like a classic L-R one).
The second dimension (whose content seems to vary over time) adds
another 2-4 percentage points, and all the ones after that don't add
more than another point or two collectively (until you run out of
parameter space). This should be taken with a bit of salt since
NOMINATE isn't trying to maximize correct classifications or minimize
errors; it's a maximum-likelihood procedure. Also, the votes are
constrained -- NOMINATE can't see the votes that never happened --
which probably reduces the apparent dimensionality.
It wouldn't make much sense to apply this to regular schmoes. Elites,
like congresscritters are, are vastly more likely to have constrained,
cohesive preferences across the vast sea of issues, and the procedures
of Congress almost certainly act so as to limit the dimensionality of
their decision (hyper)space.
--
James S. Coleman Battista
--A noble spirit embiggens the smallest man
(J. Springfield)
Geoffrey A. Landis
>
> Well, there are both precision and accuracy problems in the testing,
> but the main problem with a two-axis classification is that it smashes
> information. _Actual_ political beliefs are somewhere in a multi-axis
> manifold (I wouldn't even try to estimate the number of dimensions
> offhand),
As it turns out, one is sufficient for most applications.
I can't think of any good reason why this is, but it turns out
experimentally that political beliefs are very strongly correlated onto
a one-dimensional axis, and that the remaining variation is random, and
not easily mapped into a second dimension.
(I find this annoying, since my own politics are *not* well defined on
that axis, but it's clear I'm an outlier).
> and projecting that into a two-dimensional classification
> will necessarily screw with the nearness metric.
This is actually in some senses useful, since if political positions are
correlated to a single axis of measurement, there exists a well-defined
median. Our political system doesn't *seek* that median very
effectively, but it does at least exist. In two or more dimensions, you
can get situations where proposal A beats proposal B, proposal B beats
proposal C, and proposal C beats proposal A (a "Concorcet cycle")
--
Geoffrey A. Landis
http://www.sff.net/people/geoffrey.landis
Just published: IMPACT PARAMETER (and other quantum realities)
http://www.goldengryphon.com/ip-frame.html
Christopher M. Jones
> correlated to a single axis of measurement, there exists a well-defined
> median. Our political system doesn't *seek* that median very
> effectively, but it does at least exist. In two or more dimensions, you
> can get situations where proposal A beats proposal B, proposal B beats
> proposal C, and proposal C beats proposal A (a "Concorcet cycle")
Well, not exactly. There is no ordering for R^n (n>1) or the
complex numbers. You simply can't take one complex number (or
a pair of real numbers) and be able to divide, meaningfully and
systematically, all the other complex numbers into "less than"
and "greater than" categories (i.e. such that the only number
not less than and not greater than the number in question is
equal to it). The best you can do is to try to kludge the
ordering of the reals onto the complex numbers through some
sort of map (for example, x > (complex) y = f(x) > f(y)
(reals), such as ignoring the imaginary part, or using the
imaginary part alone, or taking the absolute value (the
distance from the origin in the complex plane)) which
invariably results in an infinite set of numbers that are
neither greater than nor less than each number, but which
are not equal (except in the case of zero), or (as
mentioned above, although 'Condorcet' is misspelled) an
ordering which is not consistant.
--
Move 'em on, head 'em up,
Head 'em up, move 'em on,
Move 'em on, head 'em up,
Rawhide!
Head 'em out, ride 'em in,
Ride 'em in, let 'em out,
Cut 'em out, ride 'em in,
Rawhide!
RAWHIDE!!!
Robert Shaw
"Christopher M. Jones" <christ...@spicedham.qwest.net> wrote
> "Geoffrey A. Landis" <geoffre...@sff.net> wrote:
> > This is actually in some senses useful, since if political positions are
> > correlated to a single axis of measurement, there exists a well-defined
> > median. Our political system doesn't *seek* that median very
> > effectively, but it does at least exist. In two or more dimensions, you
> > can get situations where proposal A beats proposal B, proposal B beats
> > proposal C, and proposal C beats proposal A (a "Concorcet cycle")
>
> Well, not exactly. There is no ordering for R^n (n>1) or the
> complex numbers
Which is just what Landis said.
The reals have a natural order, which makes left
and right useful concepts.
While R^n (n>1) may be well-orderable (by the axiom of
choice) there is no natural way of doing so that respects
the topology of the space.
Our preferences could create a partial ordering in the space
of political possibilities (policy A is better than policy B)
but, most likely, our preferences will not even qualify
as a partial ordering because of Concorcet cycles.
That means that, if the political space is multidimensional,
sorting out preferences is significantly harder.
--
'It is a wise crow that knows which way the camel points' - Pratchett
Robert Shaw
Geoffrey A. Landis
"Christopher M. Jones" wrote:
>
> "Geoffrey A. Landis" <geoffre...@sff.net> wrote:
> > This is actually in some senses useful, since if political positions are
> > correlated to a single axis of measurement, there exists a well-defined
> > median. Our political system doesn't *seek* that median very
> > effectively, but it does at least exist. In two or more dimensions, you
> > can get situations where proposal A beats proposal B, proposal B beats
> > proposal C, and proposal C beats proposal A (a "Concorcet cycle")
>
> Well, not exactly.
Exactly. In one dimension, Condorcet cycles cannot exist.
In two or more dimensions, Condorcet cycles can exist.
>...
Your commentary on complex numbers is true, but is there an obvious
applicability of this result to voting theory? I don't see where
absolute ordering of all the points is required.
> ...mentioned above, although 'Condorcet' is misspelled) an
> ordering which is not consistant.
sorry for the typo.
James S. Battista
>
>
> "Christopher M. Jones" wrote:
>>
>> "Geoffrey A. Landis" <geoffre...@sff.net> wrote:
>> > This is actually in some senses useful, since if political positions are
>> > correlated to a single axis of measurement, there exists a well-defined
>> > median. Our political system doesn't *seek* that median very
>> > effectively, but it does at least exist. In two or more dimensions, you
>> > can get situations where proposal A beats proposal B, proposal B beats
>> > proposal C, and proposal C beats proposal A (a "Concorcet cycle")
>>
>> Well, not exactly.
>
> Exactly. In one dimension, Condorcet cycles cannot exist.
Only if preferences are single-peaked. If preferences are non-single-peaked,
you can still get a cycle. The canonical example whose first preference was
to get us out of 'Nam, whose second preference was, if you're going to fight,
go all-out, and whose third-place preference was the current limited war.
Charles R Martin
> Chris Byler wrote:
> >
> > Well, there are both precision and accuracy problems in the testing,
> > but the main problem with a two-axis classification is that it smashes
> > information. _Actual_ political beliefs are somewhere in a multi-axis
> > manifold (I wouldn't even try to estimate the number of dimensions
> > offhand),
>
> As it turns out, one is sufficient for most applications.
>
> I can't think of any good reason why this is, but it turns out
> experimentally that political beliefs are very strongly correlated onto
> a one-dimensional axis, and that the remaining variation is random, and
> not easily mapped into a second dimension.
I wonder, though, how much that is simply because we tend to model
using the one-dimensional model? Given that three are two parties,
and while we don't have as strong an idea of "party discipline" as the
Brit's do, we do still have legislative rules and social norms that
push many votes to match the Party line.
--
Chrysanthemum growers -- you are the slaves of your chrysanthemums! -- Buso
______________________________________________________________________________
Charles R (Charlie) Martin Broomfield, CO 40N 105W
John David Galt
>> I remember seeing it in an old "Destinies" (packed away where
>> I can't get at it at this time). This sounds like an updated
>> version -- does any one know where I could find a copy???
Mike Van Pelt wrote:
> <http://www.baen.com/chapters/axes.htm>
>
> I think this is far superior to the "Nolan Chart" that the
> Libertarian Party pushes. The Nolan chart comes with ten
> leading essay questions to which it demands yes/no answers, and
> purports to jam you into a pigeonhole based on your answers to
> those questions.
>
> I *DETEST* "questionaires" which demand yes/no answers to
> essay questions.
>
> Also, the variables defining the two axes of the Nolan chart are
> not independant. "personal freedom" and "economic freedom" are
> closely interrelated.
>
> I believe Dr. Pournelle's two-axis chart was from his Political
> Science PhD dissertation. I'd love to get my hands on that...
Pournelle's horizontal axis ("how much do you trust government?")
makes a lot of sense. However, his other axis, "rationalism", just
begs the question of whom you consider to be rational. Each reader
will have his own opinions about where other people and groups fall
on that second axis. (With himself near the top, of course.)
For example, how "rational" is the environmental movement? If your
notion of which sources of environmental "facts" to trust is close
to Pournelle's own (and mine), they're not very rational at all.
OTOH, if you're convinced that the Earth really is in trouble, then
you would rank them much higher. Each side in this debate is
rigorous in following the "facts" it believes, while accusing its
opponents of making up theirs.
If you call the second axis "attitude toward planned social progress"
rather than rationalism, then there's no reason to distinguish the
second axis from the first. For example, take the War on Drugs.
Its supporters clearly see it as "planned social progress": the
elimination of vices that have nasty consequences, including some
side-effects on non-users. OTOH, those who are working to end the
WoD consider it a cultural war against human rights, and see most
of those nasty consequences as consequences of the War itself.
_Either_ agenda is "planned social progress", if you see it as
progress. Similarly the "war" of right-wing Christians vs. gays.
What's missing, I think, is David Friedman's concept of "consumer
sovereignty": the attitude that each person's good really is what
he or she says it is (so that all voluntary trades do produce
gains, even if it's somebody buying a product or service you don't
think is good for him). To the extent one accepts _that_, one is
a libertarian. To the extent one doesn't, one is an opponent of
liberty IMAO, regardless of whether they'd want restraint to be
carried out by the state or by someone else (family, community
pressure, etc). _That_ would make a good second axis.
Matthew DeBell
>Chris Byler wrote:
>>
>> Well, there are both precision and accuracy problems in the testing,
>> but the main problem with a two-axis classification is that it smashes
>> information. _Actual_ political beliefs are somewhere in a multi-axis
>> manifold (I wouldn't even try to estimate the number of dimensions
>> offhand),
>
>As it turns out, one is sufficient for most applications.
>
>I can't think of any good reason why this is, but it turns out
>experimentally that political beliefs are very strongly correlated onto
>a one-dimensional axis,
Like a liberal-conservative axis? If that's what you're saying, it's not
true. The liberal-conservative "ideology" construct is useful, but the
highest correlations you'll typically find between "ideology" and any
measures of political attitudes are in the ballpark of .35, which I would
call a moderate correlation. For many political attitudes the correlation
with liberal-conservative ideology is effectively zero.
>and that the remaining variation is random, and
>not easily mapped into a second dimension.
All sophisticated models of political attitudes rely on more than one
independent variable, and it is quite common for something other than
liberal-conservative ideology to be the most predictive of the IVs. It's a
useful construct, but it's not _that_ powerful.
--
Matthew DeBell
Mark Atwood
>
> There is some excellent work in this kind of thing
> from the Public/Social Choice theorists. One gentleman
> patiently explained why, in a two-party system, one
> sees slow but steady convergence on the median voter.
> Basically, it goes like this.
>
> Pick an issue or idea with political currency. Create
> an axis which ranges from "very much for" to "very much
> against". Line up all the voters end to end on it.
> Pick the median voter: if you take his line, then by
> the definition of the median you will get half or better
> of the votes available.
>
> Pick another issue, and then another, and so forth and
> so on. Basically it is far more profitable to pick
> the median voter on each issue and sell yourself to
> them and only them, because you can give everyone else
> the presidential finger once in office. Since the
> identification of median voters on certain issues can
> be more quickly and directly identified than ever
> before, gut-feeling or conviction politics has a real
> hard time beating off the numbers men.
And what happens is that every election that can afford "the numbers
men" will converge into a 50/50 split between the two main candidates.
Great. The "pin balanced on it's point" that we in the US enjoyed last
November is not an abberation, it's just the first of many.
We need to take another cue from the ancient Athenians, and decide
such cases by coin flip.
--
Mark Atwood | I'm wearing black only until I find something darker.
m...@pobox.com | http://www.pobox.com/~mra
James S. Battista
> Jacques Chester wrote in message <9thng1$i43$1...@spacebar.ucc.usyd.edu.au>...
>
>>There is some excellent work in this kind of thing
>>from the Public/Social Choice theorists. One gentleman
>>patiently explained why, in a two-party system, one
>>sees slow but steady convergence on the median voter.
>
> is a bimodal distribution in which most voters are extremists, the parties
> will not seek the median.
As long as everyone's preferences are single-peaked, the distribution
of preferences doesn't matter; it's still the median that matters. What
a bimodal distribution *would* do is admit the possibility of instant
and large change from relatively small changes in the electorate/committee.
Jacques Chester
<how...@brazee.net> wrote in message
news:uS7J7.2388$uB.2...@bin3.nnrp.aus1.giganews.com...
> > Has Anyone seen Jerry Pournelle's take on Left/Right?
>
[...]
>
> > He proposed a two axis system.
[...]
Others here have already pointed some of you to the
"Smallest Political Test in the World", in its
various libertarian guises.
There is some excellent work in this kind of thing
from the Public/Social Choice theorists. One gentleman
patiently explained why, in a two-party system, one
sees slow but steady convergence on the median voter.
Basically, it goes like this.
Pick an issue or idea with political currency. Create
an axis which ranges from "very much for" to "very much
against". Line up all the voters end to end on it.
Pick the median voter: if you take his line, then by
the definition of the median you will get half or better
of the votes available.
Pick another issue, and then another, and so forth and
so on. Basically it is far more profitable to pick
the median voter on each issue and sell yourself to
them and only them, because you can give everyone else
the presidential finger once in office. Since the
identification of median voters on certain issues can
be more quickly and directly identified than ever
before, gut-feeling or conviction politics has a real
hard time beating off the numbers men.
The other finding of this theory - and it's an interesting
one - is that somewhere out there in policy space (the
n-dimensional set of all combinations of policies) are
combinations that can break the local minima, and force
both parties to migrate to a new policy location. But
then one sees "policy cycling", because typically, once
the new minima is entrenched, the old minima is suited
to unseating it.
There are many books that may enlighten you on these
theories. There are other books that - seizing on the
economists who founded the field - declare that it is
all bunk, sociology super omnia and so on.
YMMV.
JC.
Matthew DeBell
>There is some excellent work in this kind of thing
>from the Public/Social Choice theorists. One gentleman
>patiently explained why, in a two-party system, one
>sees slow but steady convergence on the median voter.
...provided the distribution of voters is unimodal. If, for example, there
is a bimodal distribution in which most voters are extremists, the parties
will not seek the median.
--
Matthew DeBell
John Schilling
Oh, yes they will. They are guaranteed the votes of their local peak in
the distribution, and denied those in the opposite peak. Where else can
the losing party go to pick up more votes next time, than towards the
center? However sparsely populated, that *is* where both parties end
up in time. Or possibly on the inward slope of one of the peaks, in
a sufficiently asymmetric distribution.
This assumes an absolute two-party system. If you have weak third parties,
they can pick up defectors as the two major parties abandon their ideological
core and potentially grab for major-party status in their own right. At
which point things go unstable for a while, and settle down with two parties
sitting on their locak peaks and ready to start inching towards the center
again.
If you can have multiple strong parties, you wind up with one centrist and
two sitting on the peaks, plus a couple minor parties out on the wings.
This can be stable if the distribution is only weakly bimodal; if the
center is nearly depopulated I don't think it works, but I don't think
you really want to incorporate such a population under a single government
in the first place.
--
*John Schilling * "Anything worth doing, *
*Member:AIAA,NRA,ACLU,SAS,LP * is worth doing for money" *
*Chief Scientist & General Partner * -13th Rule of Acquisition *
*White Elephant Research, LLC * "There is no substitute *
*schi...@spock.usc.edu * for success" *
*661-951-9107 or 661-275-6795 * -58th Rule of Acquisition *
Samuel Kleiner
>
> And what happens is that every election that can afford "the numbers
> men" will converge into a 50/50 split between the two main candidates.
>
> Great. The "pin balanced on it's point" that we in the US enjoyed last
> November is not an abberation, it's just the first of many.
>
> We need to take another cue from the ancient Athenians, and decide
> such cases by coin flip.
>
/S
--
Torque-bomb Ekaterin!
Klyfix
<geoffre...@sff.net> writes:
>
>I can't think of any good reason why this is, but it turns out
>experimentally that political beliefs are very strongly correlated onto
>a one-dimensional axis, and that the remaining variation is random, and
>not easily mapped into a second dimension.
>
I'd think that's in part because we're pretty much taught that's how
political beliefs should be so we generally try and fit our ideas
into such a spectrum, force fitting if necessary.
V. S. Greene : kly...@aol.com : Boston, near Arkham...
Eckzylon: http://m1.aol.com/klyfix/eckzylon.html
"To commemorate a past event you kill and eat an animal.
A ritual sacrifice... with pie."- Anya, BtVS "Pangs"
Geoffrey A. Landis
Interesting example, but note that this position cannot be mapped as a
point on a single axis, with one side ("left") "get us out of the war"
and the other side ("right") being "escalate the war".
I would thus say that this is *not* an example of a one-dimensional preference.
--
Geoffrey A. Landis
http://www.sff.net/people/geoffrey.landis
Geoffrey A. Landis
> >experimentally that political beliefs are very strongly correlated onto
> >a one-dimensional axis,
>
> Like a liberal-conservative axis?
No, not precisely. What you do is ask N questions that allow a
well-defined range. For example, views on gun control and views on
welfare. You then plot these results against each other, and compute
the correlation coefficient. After doing this, if you wish, you can
then label one end of the graph "liberal" and the other "conservative",
but that's not precisely the same as correlating the position with the
liberal-conservative axis, particularly if your way of measuring
liberal-conservative position is self-identification (most people put
themselves close to the middle.)
> If that's what you're saying, it's not
> true. The liberal-conservative "ideology" construct is useful, but the
> highest correlations you'll typically find between "ideology" and any
> measures of political attitudes are in the ballpark of .35,
Can you give a reference to this number? I've not seen a correlation
coefficient this low; numbers I've seen hve been closer with 0.9
I expect it may depend on what particular views are asked.
> ...
> All sophisticated models of political attitudes rely on more than one
> independent variable, and it is quite common for something other than
> liberal-conservative ideology to be the most predictive of the IVs. It's a
> useful construct, but it's not _that_ powerful.
Sounds like you have some interesting references, which are most likely
more recent than my information (which is probably 20 years out of date
by now); I'd be quite interested to know your references.
James S. Battista
>
>
> "James S. Battista" wrote:
>>
>> In rec.arts.sf.written Geoffrey A. Landis <geoffre...@sff.net> wrote:
>> >
>> > Exactly. In one dimension, Condorcet cycles cannot exist.
>>
>> Only if preferences are single-peaked. If preferences are non-single-peaked,
>> you can still get a cycle. The canonical example whose first preference was
>> to get us out of 'Nam, whose second preference was, if you're going to fight,
>> go all-out, and whose third-place preference was the current limited war.
>
> Interesting example, but note that this position cannot be mapped as a
> point on a single axis, with one side ("left") "get us out of the war"
> and the other side ("right") being "escalate the war".
Sure it can. The utility function looks like this
U|
|
|\
| \
| \
| \ /
| \ /
| \ /
| \ /
| \/
_______________________
Intensity
Which is a perfectly good function, with an ideal point and everything.
> I would thus say that this is *not* an example of a one-dimensional
> preference.
You'd be wrong, as it happens. Even some of the old stuff like Black
or Downs goes into single-peakedness.
Geoffrey A. Landis
> ....
> There is some excellent work in this kind of thing
> from the Public/Social Choice theorists. One gentleman
> patiently explained why, in a two-party system, one
> sees slow but steady convergence on the median voter.
> Basically, it goes like this.
I'd be interested in the reference to this. My information, I think, is
rather out of date, but the old data I had seen did *not* show
convergence to the median; in fact, it showed chaotic flopping between
positions exactly 25% "left" and "right" of median in the one-D model
(which, in a different post, Matthew DeBell claims is no longer believed
to be realistic.)
> Pick an issue or idea with political currency. Create
> an axis which ranges from "very much for" to "very much
> against". Line up all the voters end to end on it.
> Pick the median voter: if you take his line, then by
> the definition of the median you will get half or better
> of the votes available.
>
> Pick another issue, and then another, and so forth and
> so on. Basically it is far more profitable to pick
> the median voter on each issue and sell yourself to
> them and only them, because you can give everyone else
> the presidential finger once in office. Since the
> identification of median voters on certain issues can
> be more quickly and directly identified than ever
> before, gut-feeling or conviction politics has a real
> hard time beating off the numbers men.
This is incorrect for the U.S., since we have a two-party system.
To win your parties' nomination, your best strategy is to approach the
median of YOUR PARTY (not the median of the entire country.)
Candidates who do not understand that they first have to win their
party's nomination do not ever enter the main presidential race. (and
it is instructive to note how very often polls often show that primary
candidate X has the best chance of winning the main election, but the
party picks candidate Y.)
James S. Battista
>> ....
>> There is some excellent work in this kind of thing
>> from the Public/Social Choice theorists. One gentleman
>> patiently explained why, in a two-party system, one
>> sees slow but steady convergence on the median voter.
>> Basically, it goes like this.
>
> I'd be interested in the reference to this. My information, I think, is
> rather out of date, but the old data I had seen did *not* show
> convergence to the median; in fact, it showed chaotic flopping between
> positions exactly 25% "left" and "right" of median in the one-D model
> (which, in a different post, Matthew DeBell claims is no longer believed
> to be realistic.)
Non. In a 1-space with single-peaked preferences, the ideal point
of the median voter is a Condorcet winner (unbeatable).
You want to look at Downs, _An Economic Theory of Democracy_, or Black,
_The Theory of Committees and Elections_. Downs will be easier to find,
and for cheap.
phil hunt
>
>To win your parties' nomination, your best strategy is to approach the
>median of YOUR PARTY (not the median of the entire country.)
Yes, and when you've won your party's nomination, you change your
rhetoric to appeal to the general public.
Always remembering to speak in vague generalities, that way you
upset as few people as possible. "a thousand points of light",
anyone?
--
*** Philip Hunt *** ph...@comuno.freeserve.co.uk ***
Geoffrey A. Landis
I was thinking of a geometric model, where each person's preference is
mapped onto a point (not a function), and the choice is determined by
the how close the option discussed is to that position. Now that you
point it out, however, it's clear that this in fact is not an applicable
model for all possible positions, and it's not even necessarily a good
definition of a one-axis system. (it's a great model in that you can
use all sorts of math developed for other purposes to analyze it, though.)
Geoffrey A. Landis
>
> Non. In a 1-space with single-peaked preferences, the ideal point
> of the median voter is a Condorcet winner (unbeatable).
The primary+main election system used in America doesn't pick the
Condorcet winner.
phil hunt
>
>(in other words the
>"liberal/conservative axis" is a strong predictor, but there is no
>other axis that is a similarly strong predictor, the rest of the
>variance is due to a lot of smaller factors). Also, this holds very
>reliably cross-culturally (and evolutionary psychologists point out
>that there are good adaptive explanations for the phenomenon).
Which are?
Jordan S. Bassior
>Well you could read "Social Dominance" by Sidanius and Pratto which
>provides a better explanation than I'm going to. To put it briefly,
>the cross-cultural psychological components of "liberalism vs.
>conservatism" (as opposed to the specific policies these personality
>traits are associated with in a specific culture), basically have to
>do with favoring hierarchical/dominant versus equitable social
>organization. The "conservative" direction, which is assessed as high
>Right-Wing Authoritarianism on Altemeyer's measure and high Social
>Dominance on Sidanius' and Pratto's measure, is associated with
>increased discrimination against outgroups in general, increased
>belief that some groups are simply better than others and that some
>groups deserve to do better, and so on. When discrimination against
>outgroups occurs, its severity is proportional to the social status of
>the outgroup, and anti-outgroup discrimination is directed primarily
>against males (so is pro-ingroup discrimination).
Neither Libertarianism nor Leninism fit well into this model.
--
Sincerely Yours,
Jordan
--
David M. Palmer
<geoffre...@sff.net> wrote:
> This is incorrect for the U.S., since we have a two-party system.
>
> To win your parties' nomination, your best strategy is to approach the
> median of YOUR PARTY (not the median of the entire country.)
No, because there is more than one candidate in the primaries.
Assume for simplicity a single-axis, 0-1 uniformly distributed
population within a party: There are equal numbers of people in the
range 0-0.1 as there are in 0.45-0.55 as there are in 0.9-1
If one candidate is at 0.6 and another at 0.45 on this axis, then a
new candiate will be best off chosing a position of 0.44 . After a few
primaries, Mr. 0.45 will find his numbers low enough that he will quit,
leaving 0.53 as the dividing line which gives the newcomer victory.
Of course, candidates don't always go by number of voters, but by
dollars. The purchasers each place their money on one or more of the
most-likely victors among the acceptable candidates.
Shifting positions after the early primaries is impeded by the news
media, who call it flip-flopping, and by the purchasers, who call it
breach of contract, reducing the amount of fluff and money the
candidates get.
--
David M. Palmer dmpa...@email.com (formerly @clark.net, @ematic.com)
Jordan S. Bassior
>Of course, candidates don't always go by number of voters, but by
>dollars. The purchasers each place their money on one or more of the
>most-likely victors among the acceptable candidates.
This incidentally complicates matters _further_, as in some cases the goals
"get the maximum amount of contributions" and "get the maximum number of votes"
conflict, forcing the candidate to choose some median between the two. If he
gets lots of money but alienates the voters he will lose with a big war chest;
if he has positions that would theoretically woo the voters but he lacks
campaign funds, he may never get to realize the potential of his platform.
Gene Wirchenko
>This vaguely reminds me of the classification in
>that hoary old game Dungeons & Dragons.
>
>You can be classifed "lawful" or "chaotic" while at the same time
>you can be classified "good" or "evil".
>
>For the first, answer the question: Does the good of the many
>outweight the good of the few (or the one)?
>Yes = lawful, no = chaotic.
>
>For the second, answer the question: Do the ends justify the means?
>Yes = evil, no = good.
It isn't that easy.
If I kill someone in self-defense, does the end justify the
means?
Maybe.
If I couldn't reasonably see any other way to effectively protect
myself, then the answer would be yes. Would my act be evil? No.
Sincerely,
Gene Wirchenko
Computerese Irregular Verb Conjugation:
I have preferences.
You have biases.
He/She has prejudices.
Grinch
<i...@ianmontgomerie.com> wrote:
>On Sat, 24 Nov 2001 02:39:11 +0000, ph...@comuno.freeserve.co.uk (phil
>hunt) wrote:
>
>>On Fri, 23 Nov 2001 15:24:11 -0800, Ian Montgomerie <i...@ianmontgomerie.com> wrote:
>>>
>>>(in other words the
>>>"liberal/conservative axis" is a strong predictor, but there is no
>>>other axis that is a similarly strong predictor, the rest of the
>>>variance is due to a lot of smaller factors). Also, this holds very
>>>reliably cross-culturally (and evolutionary psychologists point out
>>>that there are good adaptive explanations for the phenomenon).
>>
>>Which are?
>
>Well you could read "Social Dominance" by Sidanius and Pratto which
>provides a better explanation than I'm going to. To put it briefly,
>the cross-cultural psychological components of "liberalism vs.
>conservatism" (as opposed to the specific policies these personality
>traits are associated with in a specific culture), basically have to
>do with favoring hierarchical/dominant versus equitable social
>organization. The "conservative" direction, which is assessed as high
>Right-Wing Authoritarianism on Altemeyer's measure and high Social
>Dominance on Sidanius' and Pratto's measure, is associated with
>increased discrimination against outgroups in general, increased
>belief that some groups are simply better than others and that some
>groups deserve to do better, and so on. When discrimination against
>outgroups occurs, its severity is proportional to the social status of
>the outgroup, and anti-outgroup discrimination is directed primarily
>against males (so is pro-ingroup discrimination).
>
>This is exactly what one would expect from evolutionary adaptations
>for social groups and alliance forming. Males were in reproductive
>competition to a much greater extent than females, and thus the most
>extreme alliance-forming and favoritism is males banding together into
>ingroups to compete against outgroup males (outgroup females aren't so
>much competitors, as potential reproductive partners). Humans are,
>however, the most egalitarian primate and thus pure competition was
>not always the winning strategy - cooperative prosperity was also a
>viable strategy. Evolutionarily speaking, intergroup (and
>inter-individual) competition was more of a benefit for those who
>happened to find themselves in a good competitive position, and thus
>intergroup discrimination is directed opportunistically against low
>status groups, is strongest in high-status groups, is correlated with
>measures of phenotypic dominance potential such as testosterone
>levels, and so on.
>
>If this sounds to you like an analysis that is very unpopular with
>conservatives, you would be right...
In fact, it sounds like a bit of "outgroup discrimination" against
political conservatives of the sort that might be found among an "in"
group that is politically non-conservative -- such as, oh, among
contemporary academics. A veritable claim of the genetic inferiority
of political conservatives. ;-)
Now, I think it's not the above analysis of the evolutionary roots of
intergroup discrimination that would be unpopular with conservatives,
but the assumption that it is a typical trait of *political*
conservatives as opposed to political "nonconservatives". And they
would have a fair point.
I just lived through a NYC Mayoral election campaign, and during it
one couldn't turn on the news or look at a newspaper without seeing
some liberal being a bigot or playing a race card.
If discrimination of this sort has been evolutionary adaptive, then we
can assume it is in all our genes, until the day comes that the Genome
Project shows that liberals and conservatives really do have different
DNA (and politicians are genetically less self-interested than
business people, etc.)
Certainly highly partisan ethnic "outgroup" politics has played a huge
role in the great liberal history of my home town.
Of course, if by "conservative" we mean conservative on some social
interaction scale that *doesn't* map over the political scale -- so
that over in the politics ngs the liberals who band together to make
high-testosterone denouncements of conservatives *as a class* as being
greedy/stupid/bigoted/fascistic/etc., are in fact taking the adaptive
"conservative" social tact against a competing social group -- then
that's another thing.
But then we must be careful because this thread was discussing
*political* left/right, authoritarian/libertarian, etc/etc. scales,
IIRC.
Coridon Henshaw
dmpa...@email.com:
> Shifting positions after the early primaries is impeded by the news
> media, who call it flip-flopping, and by the purchasers, who call it
> breach of contract, reducing the amount of fluff and money the
> candidates get.
Rather, the media cry of 'flip-flopping' is one of the ways the purchasers
use to control their sock puppets. The media, of course, being either one
in the same as the purchasers or at least under their indirect control.
--
49 Americans exposed to anthrax: U.S. considers invalidating Cipro patent
to reduce costs. 36 million people with AIDS: U.S. sues governments which
invalidate drug patents not to save money but so their people can live.
With policies like these, is it any wonder why Osama bin Laden has so much
support? -- Coridon Henshaw / http://www3.sympatico.ca/gcircle/csbh
Grinch
<dmpa...@email.com> wrote:
>In article <3BFE8305...@sff.net>, Geoffrey A. Landis
><geoffre...@sff.net> wrote:
>
>> This is incorrect for the U.S., since we have a two-party system.
>>
>> To win your parties' nomination, your best strategy is to approach the
>> median of YOUR PARTY (not the median of the entire country.)
The median of the *likely voters* in the primary of your party, who
are not representative of all the registered party members as a group,
and who are likely to be centered towards the political wing of the
party rather than the median of all party members. (As activists tend
to be more extreme in their views than the larger numbers of modestly
engaged). After which...
>No, because there is more than one candidate in the primaries.
>
>Assume for simplicity a single-axis, 0-1 uniformly distributed
>population within a party: There are equal numbers of people in the
>range 0-0.1 as there are in 0.45-0.55 as there are in 0.9-1
>
>If one candidate is at 0.6 and another at 0.45 on this axis, then a
>new candiate will be best off chosing a position of 0.44 . After a few
>primaries, Mr. 0.45 will find his numbers low enough that he will quit,
>leaving 0.53 as the dividing line which gives the newcomer victory.
>
>Of course, candidates don't always go by number of voters, but by
>dollars. The purchasers each place their money on one or more of the
>most-likely victors among the acceptable candidates.
>
>Shifting positions after the early primaries is impeded by the news
>media, who call it flip-flopping, and by the purchasers, who call it
>breach of contract, reducing the amount of fluff and money the
>candidates get.
Until the nomination is won and the general election appears on the
horizon, after which the candidates flip, flap, flop, back to the
center where the bulk of votes now really count. And the
activists/lobbyists who got their candidate the nomination have to put
up with it because while they may now come to view him as a sell-out
double-dealer, at least he is *their* sell-out double-dealer.
For example, George Stepahopolis on Larry King Live praising
Bill Clinton by saying: "He's kept the promises he meant to keep."
;-)
David Johnston
>
> >This vaguely reminds me of the classification in
> >that hoary old game Dungeons & Dragons.
> >
> >You can be classifed "lawful" or "chaotic" while at the same time
> >you can be classified "good" or "evil".
Actually there are three categories in each axis. Neutrality is
allowed.
> >
> >For the first, answer the question: Does the good of the many
> >outweight the good of the few (or the one)?
> >Yes = lawful, no = chaotic.
> >
> >For the second, answer the question: Do the ends justify the means?
> >Yes = evil, no = good.
Feh. Nonsense. Many evil people have no noble intentions. They just
want to take what they want. And many good people in the D&D sense will
do things like killing people and justify it because it is intended to
produce good results.
Ron Bean
Ian Montgomerie <i...@ianmontgomerie.com> writes:
>(in other words the
>"liberal/conservative axis" is a strong predictor, but there is no
>other axis that is a similarly strong predictor, the rest of the
>variance is due to a lot of smaller factors).
How strong a predictor is it? That is, how much of the variance
is predicted by the liberal-conservative axis, and how much by
the "smaller factors"?
Benjamin Adams
news:20011124000943...@mb-cr.aol.com...
Sounds to me like Leninism (in practice) fits the model perfectly
...it's "Right-Wing Authoritarian" all the way. The Party is the
ingroup, and low-status groups (as the Party defines it) get
stomped on hard.
As for Libertarianism, I'd say it depends on the assumptions of
the libertarian in question. They have different reasons for
wanting to curtail the state.
-Ben Adams
Matthew DeBell
>"Matthew DeBell" <m...@attglobal.net> writes:
>
>>Jacques Chester wrote in message
<9thng1$i43$1...@spacebar.ucc.usyd.edu.au>...
>
>>>There is some excellent work in this kind of thing
>>>from the Public/Social Choice theorists. One gentleman
>>>patiently explained why, in a two-party system, one
>>>sees slow but steady convergence on the median voter.
>
>>...provided the distribution of voters is unimodal. If, for example,
there
>>is a bimodal distribution in which most voters are extremists, the parties
>>will not seek the median.
>
>
>Oh, yes they will. They are guaranteed the votes of their local peak in
>the distribution, and denied those in the opposite peak. Where else can
>the losing party go to pick up more votes next time, than towards the
>center? However sparsely populated, that *is* where both parties end
>up in time. Or possibly on the inward slope of one of the peaks, in
>a sufficiently asymmetric distribution.
They are only "guaranteed the votes of their local peak" if voters are
narrowly rational in the sense of always voting for the party "closest" to
them.
However, if voters are rational in a different (and equally legitimate) way,
the parties won't converge on the median. Voters who to look to the future
by punishing their party today in order to get what they want tomorrow will
prevent the convergence by punishing the party that moves too far to the
center. So will voters who merely insist on their party remaining "close"
to them in an absolute sense instead of just being slightly closer than the
other side.
In the real world (pardon the expression) people often abstain from voting
(or defect to third parties, if available) if the party moves too far from
their position. If a party in the sharply bimodal distribution moves very
far toward the center to capture the median voters, that party risks losing
far more votes from alienated extremists than it can gain from the sparsely
populated center. Thus, more likely than not the two parties will remain
ideologically divided when the electorate consists primarily of extremists
on the two ends of the spectrum.
--
Matthew DeBell
Matthew DeBell
Depends on your dependent variable. Models of public opinion usually
"explain" perhaps 20% to 40% of the variance in the DV. In some models the
liberal-conservative variable is not significant at all. In some it
accounts for half or more of the overall variance that is explained. So,
AFAIK it's very rare that more than a quarter of overall variance is
explained by the liberal-conservative variable.
--
Matthew DeBell
Matthew DeBell
>of preferences doesn't matter; it's still the median that matters. What
>a bimodal distribution *would* do is admit the possibility of instant
>and large change from relatively small changes in the electorate/committee.
It all depends on your criteria for voter "rationality." The distribution
does matter in models reflecting realistic voter behavior; see my reply to
John Schilling.
--
Matthew DeBell
Matthew DeBell
>Matthew DeBell wrote:
>> ...
>> All sophisticated models of political attitudes rely on more than one
>> independent variable, and it is quite common for something other than
>> liberal-conservative ideology to be the most predictive of the IVs. It's
a
>> useful construct, but it's not _that_ powerful.
>
>Sounds like you have some interesting references, which are most likely
>more recent than my information (which is probably 20 years out of date
>by now); I'd be quite interested to know your references.
Well, regarding the general idea that constructs other than the
liberal-conservative axis have significant predictive value in studies of
political attitudes and behavior:
The work of Tom Tyler (e.g. _Why People Obey the Law_) has shown that legal
compliance and the endorsement of institutions are highly contingent on the
perception that the institutions' procedures are fair.
Traditional political science theories of trust in government attribute
trust to the system's performance in the provision of benefits or to
evaluations of leaders (see Arthur Miller, 1974 article in American
Political Science Review; compare reply by Jack Citrin that same year). For
more recent take on trust, see Pew Research Center for The People & The
Press, _Deconstructing Distrust_ (1998).
In an important study of participatory behavior, liberalism-conservatism
doesn't even figure in the models; participation is primarily a function of
education and civic skills. Sidney Verba, Kay Lehman Schlozman, and Henry
E. Brady, Voice and Equality: Civic Voluntarism in American Politics
(Cambridge, MA: Harvard University Press, 1995).
William G. Jacoby ("Issue Framing and Public Opinion on Government
Spending," Am. J. of Political Sci., vol. 44, pp. 750-767 (2000)) notes that
issue framing (i.e. the way an issue is presented) affects the determinants
of opinion (i.e. which variables "explain" the opinion); he identifies
ideological self-placement as one among many significant predictors of
expressed opinions regarding government spending, including evaluations of
national economy, race, age, sex, and employment status.
In explaining voting behavior, liberal-conservative outlook is very
important, but party identification is the single most useful predictor.
See Campbell, Converse, Miller, and Stokes, The American Voter (1956).
I would never argue that the liberal-conservative construct is not
conceptually or empirically useful or important in political science. I'm
just reacting to what I saw in this thread as an exaggeration of its powers.
--
Matthew DeBell
Matthew DeBell
>Matthew DeBell wrote:
>>
>> Geoffrey A. Landis wrote in message <3BFBB943...@sff.net>...
>>
>> >I can't think of any good reason why this is, but it turns out
>> >experimentally that political beliefs are very strongly correlated onto
>> >a one-dimensional axis,
>>
>> Like a liberal-conservative axis?
>
>No, not precisely. What you do is ask N questions that allow a
>well-defined range. For example, views on gun control and views on
>welfare. You then plot these results against each other, and compute
>the correlation coefficient. After doing this, if you wish, you can
>then label one end of the graph "liberal" and the other "conservative",
>but that's not precisely the same as correlating the position with the
>liberal-conservative axis, particularly if your way of measuring
>liberal-conservative position is self-identification (most people put
>themselves close to the middle.)
I'm not sure if I understand you correctly. I would imagine you'd want to
construct a scale by taking the mean of individuals' scores on the various
measures of views (gun control, welfare, etc.), which might be useful.
However it presents the hazard that the scale components may not be
unidimensional.
>
>> If that's what you're saying, it's not
>> true. The liberal-conservative "ideology" construct is useful, but the
>> highest correlations you'll typically find between "ideology" and any
>> measures of political attitudes are in the ballpark of .35,
>
>Can you give a reference to this number? I've not seen a correlation
>coefficient this low; numbers I've seen hve been closer with 0.9
I don't have a reference that catalogues the correlations, but I can compute
them for you from relevant data. (And btw r=.9 is virtually unheard of in
public opinion and political psychology, except between variables that
purport to measure exactly the same thing.)
Aside from various measures of authoritarianism, the most frequently used
empirical measure of liberalism-conservatism is a simple pair of survey
questions that place people on a 7 point scale ranging from strong liberal
to strong conservative. (The first question is approximately, "Do you
consider yourself liberal, moderate, or consertative?" The second, depending
on the answer to the first one, is either "Are you a strong liberal
(conservative) or not?" or, if they say they're moderate, "Do you lean more
toward liberal or conservative?") When you ask these questions of
Americans, you get results like this:
1.6% extremely liberal
9.5 liberal
14.3 slightly liberal
30.5 moderate
19.7 slightly conservative
20.8 conservative
3.6 extremely conservative
(Source: U.S. National Election Study, 1996; N = 1350)
Many general political attitudes are quite poorly predicted by (i.e. very
weakly correlated with) the liberal-conservative scale. For example, a
sense of political efficacy is not correlated with liberal-conservative
views at all, and trust in government is only very weakly correlated (r=.06,
p<.05).
Here are some other correlations between survey respondents' placement on
the liberal-conservative scale and opinion on various matters. (Source: my
own calculations using data from the 1996 National Election Study; see
http://www.umich.edu/~nes/) (All correlations are significant at p<.01
unless noted "NS" for "nonsignificant.")
Correlation with Liberal-Conservative Position:
support for Congressional term limits: r = .05 (NS)
willingness to use military force: .13
support death penalty: .21
support handgun control: -.22
favor environmental protection over job protection: -.31
support abortion rights: -.32
Agree that...
...one should help the less fortunate: .02 (NS)
...we have gone too far pushing equal rights: .40
...there would be fewer problems if people were treated more equally: -.28
...spending on AIDS research should be decreased: .29
...spending on foreign aid should be decreased: .09
...spending on welfare programs should be decreased: .30
...spending on immigration control should be decreased: -.16
Positive coefficients indicate conservative support/agreement and liberal
opposition/disagreement. Negative coefficients indicate the reverse.
Now, these numbers show that liberal-conservative postion is correlated with
a whole slew of opinions in fairly predictable ways. Although you can't
tell from the stats I've posted, the lib-con variable is unusual in this
respect (though party identification is similarly pervasive). Its
significance holds up when statistical controls are in place for other
variables (like age, income, sex, race, education, party, etc.), so the
correlations are not spurious. So it's a useful construct.
However, none of these correlations can be considered terribly strong in a
substantive sense. Consider that
squaring a correlation yields the % of one variable's variance that is
"explained" by the other, and if you square the highest correlation in the
list (.4) you find 16% of the variance being explained. That's interesting,
but it still leaves most of the story untold.
Other variables do matter. Age, income, levels of interpersonal trust,
perceptions of justice... It's hard to be specific without reference to a
specific dependent variable, but sophisticated accounts of the variance in
political beliefs don't stop with one independent variable. It's not hard
to develop statistical models of public opinion that explain 20% or more of
the variance in the dep. variable, but it's relatively unusual that more
than about 10 or 15% of the variance is explained by a single dimension.
For example, I just threw together a quick regression model of abortion
attitudes. Permissiveness toward abortion is the dependent variable, and
the independent variables are liberal-conservative self-placement, sex,
party identification, level of education, and the frequency of religious
service attendance. All together, these ind. variables explain about 19% of
the variance in abortion attitudes. Church attendance explains a bit more
of the
variance than liberal-conservative ideology.
The liberal-conservative dimension is very useful, and for many purposes it
is the single most useful unidimensional summary of political orientation,
but it is important to note its limitations. It is often inaccurate,
imprecise, or inapplicable, and sometimes fails to account for much of the
variance in attitudes or behavior. These failures become increasingly
common as one moves away from the list of policy questions that are
ideologically polarized in public debate. It is actually not infrequently
left out of the statistical models developed in public opinion and political
psychology these days, because combinations of variables measuring views on
equality, limited government, and the like are equally predictive and
theoretically richer or more appropriate.
--
Matthew DeBell
Chris Byler
<geoffre...@sff.net> wrote:
>
>
>"Christopher M. Jones" wrote:
>>
>> "Geoffrey A. Landis" <geoffre...@sff.net> wrote:
>> > This is actually in some senses useful, since if political positions are
>> > correlated to a single axis of measurement, there exists a well-defined
>> > median. Our political system doesn't *seek* that median very
>> > effectively, but it does at least exist. In two or more dimensions, you
>> > can get situations where proposal A beats proposal B, proposal B beats
>> > proposal C, and proposal C beats proposal A (a "Concorcet cycle")
>>
>> Well, not exactly.
>
>Exactly. In one dimension, Condorcet cycles cannot exist.
>
>In two or more dimensions, Condorcet cycles can exist.
Dimension has nothing to do with it. It's the non-transitivity of the
ordering relation that's at fault. You can have transitive partial
orders in any dimensional space (for example, "further from the center
of the Earth" is a transitive partial order on points in space; some
distinct points tie, which is why it's a partial order, but Condorcet
cycles are impossible). On the other hand, you can certainly have
non-transitive relations on a one-dimensional space; they just won't
be as simple as "to the left of".
--
Chris Byler cby...@vt.edu
Kubera: "It occurred to me that Sam would be the number one suspect,
except for the fact that he was dead."
Sam: "I had assumed that to be sufficient defense against detection."
-- Roger Zelazny, _Lord of Light_
Chris Byler
<christ...@spicedham.qwest.net> wrote:
>"Geoffrey A. Landis" <geoffre...@sff.net> wrote:
>> This is actually in some senses useful, since if political positions are
>> correlated to a single axis of measurement, there exists a well-defined
>> median. Our political system doesn't *seek* that median very
>> effectively, but it does at least exist. In two or more dimensions, you
>> can get situations where proposal A beats proposal B, proposal B beats
>> proposal C, and proposal C beats proposal A (a "Concorcet cycle")
>
>Well, not exactly. There is no ordering for R^n (n>1) or the
>complex numbers. You simply can't take one complex number (or
>a pair of real numbers) and be able to divide, meaningfully and
>systematically, all the other complex numbers into "less than"
>and "greater than" categories (i.e. such that the only number
>not less than and not greater than the number in question is
>equal to it). The best you can do is to try to kludge the
>ordering of the reals onto the complex numbers through some
>sort of map (for example, x > (complex) y = f(x) > f(y)
>(reals), such as ignoring the imaginary part, or using the
>imaginary part alone, or taking the absolute value (the
>distance from the origin in the complex plane)) which
>invariably results in an infinite set of numbers that are
>neither greater than nor less than each number, but which
>are not equal (except in the case of zero), or (as
>mentioned above, although 'Condorcet' is misspelled) an
>ordering which is not consistant.
Hold it. Isn't the cardinality of R equal to the cardinality of R^k
for *ANY* finite k? In which case, there exists a *one-to-one* map,
which will necessarily result in an ordering (not just a partial
ordering) by using the procedure you defined above.
It isn't a very intuitive or useful ordering, because the kind of maps
you have to use to map R onto R^2 one-to-one can only politely be
described as pathological. But it exists. (If, that is, I'm
correctly recalling that crucial first point - that R and R^k can be
put into 1-1 correspondence for any finite k.)
Regardless, in order for a median (or other measure of central
tendency) to be *useful*, what you really need is a metric space,
where you can say something about how *close* the other data points
are to the median. "90% of all voters are within X distance of this
viewpoint" is more useful than "By some ordering I pulled out of my
ass, 50% of voters are on this side of this viewpoint, and 50% are on
the other side" when you're establishing the viewpoint in question as
"mainstream" or some such.
Is the distribution of political views in the USAn general public
bimodal? How about in the pool of USAn politicians? In general, do
N-party systems polarize the electorate into N camps, or do they only
polarize the political class?
Chris Byler
wrote:
Sure they do - if you get over the problem that you're incongruously
labeling Leninism "right wing authoritarianism", it's a dead ringer
for fascism on the issues being measured here (primarily suppression
of dissent and xenophobia, both of which Leninism is notorious for).
For Libertarians you have the opposite problem - some people have the
preconceived notion that Libertarianism is right wing, but according
to *this* axis it is in fact *left* wing, being (by self-definition at
least) the opposite of authoritarianism.
David Johnston
Of course I've seen self proclaimed libertarians who were in fact damned
authoritarian. They were were just more into the "not giving social
assistence" part and the "I should be able to do hard drugs if I want
to". But their attitudes toward crime control, collectivism, and
freedom of religion were something else altogether.
James S. Battista
> James S. Battista wrote in message <9tibuj$mr4$1...@hermes.acs.unt.edu>...
>>
>>of preferences doesn't matter; it's still the median that matters. What
>>a bimodal distribution *would* do is admit the possibility of instant
>>and large change from relatively small changes in the electorate/committee.
>
> It all depends on your criteria for voter "rationality." The distribution
> does matter in models reflecting realistic voter behavior; see my reply to
> John Schilling.
Well, sure, if you want to go and make it complicated. I thought
people were just talking about toy spatial models. In simple
lineland, the distribution is irrelevant; the median voter is the
median voter and is a condorcet winner. And of course work on
multiparty elections is trickier.
Tony McGann has been doing work on optimal party positions given
different kinds of distributions of preferences... I don't think
it's been published yet (but IIRC it's coming out in AJPS 2/02),
but I'm sure he'd pass along a copy. He's at UC-Riverside or
UC-Irvine.
Aaron Denney
> Hold it. Isn't the cardinality of R equal to the cardinality of R^k
> for *ANY* finite k? In which case, there exists a *one-to-one* map,
> which will necessarily result in an ordering (not just a partial
> ordering) by using the procedure you defined above.
>
> It isn't a very intuitive or useful ordering, because the kind of maps
> you have to use to map R onto R^2 one-to-one can only politely be
> described as pathological. But it exists. (If, that is, I'm
> correctly recalling that crucial first point - that R and R^k can be
> put into 1-1 correspondence for any finite k.)
Yes. The typical one is "interleave the digits (or bits, or ...) of
the k numbers". (This just maps [0,1)^k to [0,1), but you can map and
unmap R to [0,1) trivially). Other nice mappings are Peano or Hilbert
curves and higher-dimensional variations of space-filling curves, which
tend to be *slightly* less pathological.
--
Aaron Denney
-><-
Matthew DeBell
>
>Is the distribution of political views in the USAn general public
>bimodal?
If you force it on to one left-right dimension, it's unimodal. If you use
self-reported liberal-conservative position, it's slightly skewed to the
right but resembles a normal curve.
>How about in the pool of USAn politicians?
I could believe that distribution is bimodal or unimodal, but I've never
seen anything to indicate what it is. I'd be pretty surprised if there
hasn't been a study, though.
>In general, do
>N-party systems polarize the electorate into N camps, or do they only
>polarize the political class?
The extent to which the party system polarizes the electorate probably
depends on how entrenched the party system is. If the system will readily
accomodate any number of parties from 2 to 20, then the system probably
won't push the electorate very hard. If the system makes it very difficult
for third parties to succeed (as ours does) then most of the voting public
will get into the habit of voting for one of the two parties, so there would
be a tendency for N parties to encourage the development of N camps.
However, this is not the same thing as producing an N-modal distribution of
political views, because partisanship is not really a political view; it's
an identity and/or a behavioral habit.
--Matthew DeBell
mark
Byler) wrote:
<snip>
I cheerfuly admit all the math in this thread went right by me, but
this bit didn't.
>Is the distribution of political views in the USAn general public
>bimodal? How about in the pool of USAn politicians? In general, do
>N-party systems polarize the electorate into N camps, or do they only
>polarize the political class?
Well, why do "third" parties pop up every time you take a deep
breath? Because the platforms (coordinate clusters on your
hypothetical axis/axes) of the two "official" parties just don't
adequately model the views of many real people.
Do you want to keep what you earn? You're a Conservative. Do you
think abortion should be a matter of unrestricted choice? You're a
Liberal. What if you hold both "positions" at the same time? You're a
Mugwump.
Parties try to model a sufficient number of people's views to
generate a majority of votes, and politicians express their views to
fit within a party.
The whole idea of axial modeling of politics looks backwards to me.
First try to generate a spatial model of _views_, then decompose that
to a coordinate system that will allow "parties" to actually represent
people, instead of the other way around.
Yeah. Like that'll happen.
OTOH why has nobody mentioned that Pournelle considers himself a
"closet royalist" who wishes he were at least a Duke?
Mark L. Fergerson
Matthew DeBell
>Jacques Chester wrote:
>> ....
>> There is some excellent work in this kind of thing
>> from the Public/Social Choice theorists. One gentleman
>> patiently explained why, in a two-party system, one
>> sees slow but steady convergence on the median voter.
>
>I'd be interested in the reference to this.
I'll echo JSC Battista's recommendation of Downs, An Economic Theory of
Democracy.
Also see an article by Morris P. Fiorina (who's a prominent political
scientist):
http://www.igs.berkeley.edu:8880/research_programs/ppt_papers/MIT.pdf
--
Matthew DeBell
phil hunt
>much competitors, as potential reproductive partners). Humans are,
>however, the most egalitarian primate and thus pure competition was
>not always the winning strategy - cooperative prosperity was also a
>viable strategy. Evolutionarily speaking, intergroup (and
>inter-individual) competition was more of a benefit for those who
>happened to find themselves in a good competitive position, and thus
>intergroup discrimination is directed opportunistically against low
>status groups, is strongest in high-status groups, is correlated with
>measures of phenotypic dominance potential such as testosterone
>levels, and so on.
>
>If this sounds to you like an analysis that is very unpopular with
>conservatives, you would be right...
It's a hypothesis that's testable experimentally. Does anyone know
of any studies attempting to correlate conservastism with testosterone
levels.
(High testosterone can lead to baldness, and the current and previous
leaders of the UK Conservative party are both bald, BTW) :-)
phil hunt
>
>In fact, it sounds like a bit of "outgroup discrimination" against
>political conservatives of the sort that might be found among an "in"
>group that is politically non-conservative -- such as, oh, among
>contemporary academics. A veritable claim of the genetic inferiority
>of political conservatives. ;-)
I don't think Ian was saying that hierarchialism/egalitarianism is
genetically determined. It seems to me that it quite probably isn't.
I bet it correlates quite well with birth order, however.
Ron Bean
"Matthew DeBell" <m...@attglobal.net> writes:
>The work of Tom Tyler (e.g. _Why People Obey the Law_) has shown that legal
>compliance and the endorsement of institutions are highly contingent on the
>perception that the institutions' procedures are fair.
This would apply to any organization, not just governments.
Everyone knows that "life is not fair", but we expect
institutions to nudge things *toward* fairness rather than away
from it. It's a major factor in things like job satisfaction.
Robert Wright also touches on this in his book "Non Zero: The
Logic of Human Destiny". He points out that leaders who don't let
the good things "trickle down" far enough tend to get overthrown
by the "have-nots".
Jacques Chester
"Geoffrey A. Landis" <geoffre...@sff.net> wrote in message
news:3BFE8305...@sff.net...
> Jacques Chester wrote:
> > ....
> > There is some excellent work in this kind of thing
> > from the Public/Social Choice theorists. One gentleman
> > patiently explained why, in a two-party system, one
> > sees slow but steady convergence on the median voter.
> > Basically, it goes like this.
>
> rather out of date, but the old data I had seen did *not* show
> convergence to the median; in fact, it showed chaotic flopping between
> positions exactly 25% "left" and "right" of median in the one-D model
> (which, in a different post, Matthew DeBell claims is no longer believed
> to be realistic.)
I was working from a description of Downs's
basic theory. I am no economist, merely an
interested layman. A little knowledge ...
There will be many volumes at any decent University's
library.
JC.
Geoffrey A. Landis
> Geoffrey A. Landis wrote in message <3BFE804E...@sff.net>...
> >Matthew DeBell wrote:
> >>
> >> Geoffrey A. Landis wrote in message <3BFBB943...@sff.net>...
> >>
> >> >I can't think of any good reason why this is, but it turns out
> >> >experimentally that political beliefs are very strongly correlated onto
> >> >a one-dimensional axis,
> >>
> >> Like a liberal-conservative axis?
> >
> >No, not precisely. What you do is ask N questions that allow a
> >well-defined range. For example, views on gun control and views on
> >welfare. You then plot these results against each other, and compute
> >the correlation coefficient. After doing this, if you wish, you can
> >then label one end of the graph "liberal" and the other "conservative",
> >but that's not precisely the same as correlating the position with the
> >liberal-conservative axis, particularly if your way of measuring
> >liberal-conservative position is self-identification (most people put
> >themselves close to the middle.)
>
> I'm not sure if I understand you correctly. I would imagine you'd want to
> construct a scale by taking the mean of individuals' scores on the various
> measures of views (gun control, welfare, etc.), which might be useful.
> However it presents the hazard that the scale components may not be
> unidimensional.
I'm not sure what you're saying.
What you want to do is to construct a multi-variable linear regression
in the N-dimensional space.
You don't have to take the mean of individual scores to do that.
> ...
> Aside from various measures of authoritarianism, the most frequently used
> empirical measure of liberalism-conservatism is a simple pair of survey
> questions that place people on a 7 point scale ranging from strong liberal
> to strong conservative. (The first question is approximately, "Do you
> consider yourself liberal, moderate, or consertative?" The second, depending
> on the answer to the first one, is either "Are you a strong liberal
> (conservative) or not?" or, if they say they're moderate, "Do you lean more
> toward liberal or conservative?")
We seem to be answering different questions.
My assertion was that you can graph a person's opinions on one subject
with that person's opinions on another subject, and that these positions
can be mapped onto a linear function, with low variance across the
entire population. That linear scale can then, if you wish, be labelled
"left" at one pole, and "right" at the other. You could also label it
"high" at one end and "low" at the other; it's just a lable.
Your statement is that you can start with a liberal-conservative axis,
ask people to self identify along that axis, and then correlate their
position on other issues with that self-identification.
Two very different things!
--
Geoffrey A. Landis
http://www.sff.net/people/geoffrey.landis
Just published: IMPACT PARAMETER (and other quantum realities)
http://www.goldengryphon.com/ip-frame.html
Matthew DeBell
I'm talking about computing a "scale" to measure liberalism-conservatism.
You measure lib-con by asking say 5 or 6 survey questions that should
measure the concept, and then taking the mean of those scores. This makes
for a more reliable measure than a single question, if those 5 or 6
questions are all measuring the same thing. That's a big "if," since the
left-right dimension contains more than one underlying factor.
>
>What you want to do is to construct a multi-variable linear regression
>in the N-dimensional space.
>
>You don't have to take the mean of individual scores to do that.
True, but it's a way of constructing the dependent variable that measures a
person's liberal-conservative position. By no means the only way, of
course.
>> ...
>> Aside from various measures of authoritarianism, the most frequently used
>> empirical measure of liberalism-conservatism is a simple pair of survey
>> questions that place people on a 7 point scale ranging from strong
liberal
>> to strong conservative. (The first question is approximately, "Do you
>> consider yourself liberal, moderate, or consertative?" The second,
depending
>> on the answer to the first one, is either "Are you a strong liberal
>> (conservative) or not?" or, if they say they're moderate, "Do you lean
more
>> toward liberal or conservative?")
>
>We seem to be answering different questions.
>
>My assertion was that you can graph a person's opinions on one subject
>with that person's opinions on another subject,
I'm with you so far.
>and that these positions
>can be mapped onto a linear function, with low variance across the
>entire population. That linear scale can then, if you wish, be labelled
>"left" at one pole, and "right" at the other. You could also label it
>"high" at one end and "low" at the other; it's just a lable.
So we would come up with a linear regression equation that fits the graph,
and you expect a pretty good fit?
If you're correct, opinions on different subjects (e.g. abortion and welfare
spending, etc.) should be strongly correlated.
One of the key findings of public opinion research dating from the '60s is
that ideological consistency in the mass public is not very high. Issues
with a close conceptual connection to each other can be an exception to this
rule. For instance, views on abortion are related to views about feminism,
though not terribly strongly (Spearman correlation=.24); views on a gov't
guaranteed standard of living are strongly related to views on gov't aid to
blacks (Spearman = .52). For other issues that lack a conceptual link the
correlations are low or nil, e.g. views on abortion are unrelated to views
on aid to blacks and views on the role of women are very weakly related to
views on welfare spending (.08).
(Statistics computed from 1996 National Election Study data.)
--
Matthew DeBell
Matthew DeBell
>You basically seem to be talking at complete cross-purposes with what
>Geoffrey and I were talking about.
[....]
>I never said that the "liberal/conservative axis" was a
>great predictor of every single thing relevant to politics. It is a
>good predictor of peoples' political attitudes, over large data sets
>(i.e. ones where variations in presentation of specific situations and
>so on, the formation of attitude highly influenced by short-term
>circumstance, have basically faded to noise).
Indeed, Geoffrey and I did seem to be at cross-purposes. Perhaps my
definition of a liberal-conservative axis is rather narrow. For instance, I
would not call the SDO a liberal-conservative axis; I would call it an
intriguing predictor of liberal and conservative sentiments, but I would not
equate it with those sentiments. I don't think the F-scale or the RWA scale
are properly considered liberal-conservative axes either, because
conservatism is not the same thing as authoritarianism.
--
Matthew DeBell
Matthew DeBell
>
>"liberal-conservative variable". I am talking about models which have
>inferred explanatory axes from the data (or in the case of SDO from
>theory and data) based on regressions and analysis from data points on
>individual opinion. I am _not_ talking about whether people
>self-identify as "liberal" or "conservative", which indeed may not
>always be significant. The good models correlate well with
>self-identification but not at all perfectly.
Ok. SDO isn't used all that much in public opinion research (which is
probably too bad; it seems quite interesting), probably because measuring it
is a lot more expensive than taking a simple liberal-conservative
self-report, and because it's conceptually more complicated. I know it has
personality correlates and policy correlates, but I don't know how strong
they are, or precisely how well it fits in regression models that explain
variance in policy preferences.
--
Matthew DeBell