Log in

View Full Version : Help Design New Political Test



NovelGentry
4th May 2005, 01:59
After responding to a thread about the "loop theory" of political left and right and mentioning that I viewed political orientation more in a sphere form, I decided that maybe a new test, similar to politicial compass, but more in depth, was in order. I came up with a drawing of a "sphere" which would have three dimensional factors (although certainly more could be added with time (although it would get more and more confusing to look at). The way I see it there are three primary issues which decide politics. Political compass has two in order, economic and authoritative positions, however, neither of these play into the vast majority of other political issues such as environmentalism or social liberties, etc. So I have added the third axis (a z-axis if you will) to represent social liberties (environment will have to stay out for now).

The way the test would be done is that you would be posed a question from one of the categories. You would either disagree strongly, disagree, stay neutral, agree, or agree strongly. Each would factor in a step depending on how many questions we have (the more questions the more fine grained it becomes). It would then plot three points. The three points are easy enough to understand and display, however, from these three points we would have to come up with a final single point comprised of all three in the three dimensional space.

We could use a single straight line to determine the distance one is from another political figure or from one to another, however, this would be fairly inaccurate since it is placed inside a sphere. So what would be far more appropriate would be an arch length given two three dimensional point, the curve matching that of a perfect sphere. So here is what I need from anyone looking to help. A) a bunch of questions and a category that they would fall in, after we get maybe 20 questions from each category we can determine the factor which they will push one forwards or backwards so that they don't actually break out of the sphere if they were too extreme. B) Someone who knows how to calculate the vector length, preferably in a broken down formulat that could be represented in simple computer functions (many math functions are available, so things like roots, powers, etc are doable, and obviously simple math functions like subtraction and multiplication. If anyone would like to help provide any of these (especially the formulat for calculating the arch distance, that would be grand.

For example someone with the points 0 (x), 100 (y), 50 (z) would be located at center left and right, as high as they can go before hititng the edge of the sphere up and down, and depth within the sphere half way. So looking at the sphere from the front the point would not be visible, as it would lie half way up the center of the back portion of the sphere. Hope that all made sense.

Here's a graphic I drew up for simplified representation of points.

http://www.dotink.org/~gent/content/Images/DigitalArt/sphere.png

Commie Rat
4th May 2005, 09:11
would be interesting to see what you come up with
good luck

lvialviaquez
4th May 2005, 22:47
I really want to see what you come up with, but I don't think I can offer much help. Sorry.

lvialviaquez
4th May 2005, 22:47
I really want to see what you come up with, but I don't think I can offer much help. Sorry.

apathy maybe
6th May 2005, 05:43
I've had similar thoughts, though because I envisaged a number of axis I was just going to have numbers, and not plot them on anything. But a sphere seems cool.

The way I thought it should be is to have a number of areas, environmentalism, human rights, personal rights, economics, etc. When you start the test you are zero on all. The questions are divided amongst the areas. As you answer a question, the score goes up or down depending on the answer. Because each question is in a separate area, you don't have to worry about complex maths.

Some example questions from personal rights,

Do you think that people have the right to use drugs that may harm them? (Yes -1, no +1)
Do you think that people have the right to kill themselves? (Yes -1, no +1)



Though it must be said that scales such as on political compass and what you are proposing are simplistic. They show to a limited extent the differences between peoples political opinions. Apparently political compass was created because the old left-right scale is simplistic. But by only have two scales, political compass ignores a range of other issues.

As soon as you start adding more scales (environmentalism, etc) you start becoming too complex. And you risk putting people in little boxes.

If you simply have a number of 'tests' for each area, a person then doesn't have to answer every single test, they can do the ones that they want to.

encephalon
6th May 2005, 05:57
NovelGentry: I've a book of algorithms for perl on a lot of calculus functions, including vector length, I believe. It should be relatively easy, I would imagine, to convert it to whatever language you'll be using.. php? Although they may use perl-specific libraries in the process.. still, it shouldn't be difficult to convert. Before I try to go find it, however, let me know if you want me to do so.. I tend to be a bit disorganized with my books, especially with the ever growing number I seem to amass. I'd make a horrendous librarian, I'm positive :P

Regardless, I do think the third axis is a decent idea.. I'm not sure everything can be put in simply three dims, though, let alone two.. before I started messing with the red test (which is far from perfect.. or accurate half the time), I was thinking about doing something similar to political compass as well.

I would suggest.. unlike my very basic test.. that to increase accuracy, one could possibly have graduated questions.. that is, then next 5-10 questions are based on your answers to the first 5-10. It would be infinitely more complex than the basic test format, but it would also allow better differentiation between different ideologies/etc. that have more and more in common.. eg. it could go from very basic, holistic lines of questioning to very particular ones towards the end.

Anyhow, let me know if you want the algorithm(s). The book is filled with them.. I've only ever used a few of the things in it.

NovelGentry
6th May 2005, 07:54
I would suggest.. unlike my very basic test.. that to increase accuracy, one could possibly have graduated questions.. that is, then next 5-10 questions are based on your answers to the first 5-10. It would be infinitely more complex than the basic test format, but it would also allow better differentiation between different ideologies/etc. that have more and more in common.. eg. it could go from very basic, holistic lines of questioning to very particular ones towards the end.

This would probably work if I was trying to focus people into an ideology, but that is not so much the case. Anyway, I've revised the means a bit.

Anyone who does not fall on 0 for all, that is 100% centrist will have their location mapped to the outer edge of the sphere. What will happen is that the three points will form a triangle, and lines will be drawn to midpoints to find the "center" where the distance is equal from each question. This way, answering completely liberally on one question and conservative on another will "tug" your overall political spectrum. One could add more axis's and simply increase the numbers of lines to find the center point over time. Then a line is drawn from center through the middle point and seen where it hits on the outer sphere. Distance around the sphere can then be determined for other people you'd like to compare yourself to.

In this sense someone who is economically liberal but conservative in social issues and authoritarian could be very close to someone who is more conservative and "right" in all those issues. Then how radical you are, for whatever your ideology would be determined by the overall depth of that center point. If you had a postivie 1 on all scales, you'd have a perfect triangle and would be smack dab in the center of one of the 8ths of the sphere, but you'd also be ranged as non-radical. If you had +20 on all, you'd also be smack dab in the center of the same section, but would be considered extremely radical.

Is this making any sense?

Only thing I should need to get this working is a formula to find the distance between 3 dimensional points.

Sukhe-Bator
6th May 2005, 08:06
If you see left-wing as leaning towards the working class and right-wing as leaning towards the bourgeoisie in the great class contradiction of modern society, I do not see how the opposite extremes loop around and meet; masters and slaves have unreconcilable interests.

NovelGentry
6th May 2005, 08:10
If you see left-wing as leaning towards the working class and right-wing as leaning towards the bourgeoisie in the great class contradiction of modern society, I do not see how the opposite extremes loop around and meet; masters and slaves have unreconcilable interests.

Because ideologies as a whole are far more complex than this. It is why a once socialist can create fascist ideology. It's where "third positionism" rises from.

The minute certain lines are crossed, they can skew a person's entire perspective.

encephalon
6th May 2005, 08:17
calculating distance in 3D space goes by this formula:

d = ((X2-X1)^2 + (Y2-Y1)^2 + (Z2-Z1)^2)^0.5


This would probably work if I was trying to focus people into an ideology, but that is not so much the case. Anyway, I've revised the means a bit.

True, but it could also refine the coordinates as well.. someone might answer "strong agree" to "is pollution bad?" but strongly disagree with "should we save the rainforests?" They might be confused if they answer that way.. but I've heard worse :P

Anyhow, I hope the formula helps.

EDIT: It might actually be faster, depending on whether the language you use has an in-built root function, to do the square root of it all instead of ^0.5.. it's the same thing, but they usually try to optimize functions like that.