Posts by BenWilson

Last ←Newer Page 1 2 3 4 5 Older→ First

  • Speaker: Meet the middle, in reply to Brent Jackson,

    Can we categorise people in some other way ? For example :

    There's a very large number* of possibilities. Rob Salmond mentioned upthread what his method was. It's not crazy, but it does use National and Labour as the reference point and is thus, as Sacha comments, FPP flavoured. He mentions his method for centrists, and I presume left and right are the leftovers in each direction, although he doesn't actually say so.

    We've also got the respondent's own self rating of which party they are closest to. I think that might be useful for finding "Labourites" and "Nationalites", possibly more useful than what you suggest. This self rating is in two separate questions, just for more multi-dimensional confusion :-). One is "which are you close to". The other is "which are you closest to". I wonder how much trouble respondants who were not highly competent in English might have had seeing the difference here.

    Auckland • Since Nov 2006 • 9307 posts Report Reply

  • Speaker: Meet the middle, in reply to Brent Jackson,

    So they are presumably Labour supporters, and got the axes wrong (ie left and right held no meaning for them).

    That was one hypothesis I mulled. I decided I'd leave it unmediated by my perceptions, and see if you got the same idea.

    Auckland • Since Nov 2006 • 9307 posts Report Reply

  • Speaker: Meet the middle,

    Wicked, thanks for that David. I didn't spot any labels that got mixed up, but good heads up. I'll keep an eye out for it. So far the variable names seem to be mnemonics for the labels I've got.

    Auckland • Since Nov 2006 • 9307 posts Report Reply

  • Speaker: Meet the middle,

    Actually, I revise my opinion on the lack of conditioning. It does look like the closer to the center value a person self rated, the less distinction they see between Labour and National. The center of the clouds seems to start in the bottom right for self=left, moves toward the middle for self=center, and then it moves back to the bottom right for self=right.

    ETA: Fixed erroneous left/right confusion in above.

    Auckland • Since Nov 2006 • 9307 posts Report Reply

  • Speaker: Meet the middle,

    The only thing that seems interesting to me in the conditioning plot is that the little clusters in the top left of each graph representing the people saying Labour is extreme right and National is extreme left. It’s biggest in people rating themselves in the exact center (but probably because this is the biggest group in total), following by the people rating themselve extreme right, followed by people rating themselves extreme left.

    But all of these clusters are small. No matter how they positioned themselves, most people rate Labour as to the left of National.

    ETA: Also interesting, just noticed. People who rate themselves as 1, so very left but not extremely, have the least variation by far. Only 1 single person took the weird point on the top left, all the rest are around the average position in the bottom right.

    Auckland • Since Nov 2006 • 9307 posts Report Reply

  • Speaker: Meet the middle,

    Attachment

    Last one for Brent…conditioning plot. One plot for each level of self rating. The order of these is annoying and hard to label. Counting the self positioning from 0 to ten, you find the correct plot by reading from the bottom left to the top right.

    So bottom left is for self-position=0. Bottom right is self-position=4. Left middle is self-position=5. Top right-most plot is for self-position=10.

    I love these plots, only just learned about them. So cool to be able to see three dimensional structure in 2 dimensions.

    I don’t see any particular conditioning effect – it seems that the same kinds of proportions of the answer combinations occur no matter where the respondant positioned themselves.

    ETA: This is on jittered points. Impossible to see anything interesting otherwise. Each plot is how the respondent rated Labour's position against how they rated National's.

    Auckland • Since Nov 2006 • 9307 posts Report Reply

  • Speaker: Meet the middle,

    Attachment Attachment

    In case it didn’t (the scatters have a lot of data points so blobs lose structure), here’s binned versions. The pretty one is jittered, so it’s shape looks a bit more random than it actually is. The one using relative sizes is centered so you can read off the actual count ranges for the particular values.

    ETA: I think the centroid one shows a little more clearly how insignificant that upper left corner cluster really is. It only stand out in the scatterplot because its isolated.

    Auckland • Since Nov 2006 • 9307 posts Report Reply

  • Speaker: Meet the middle,

    On second thoughts, it makes sense. I was puzzling about the apparent negative correlation in the big crowd to the lower right of the Nat vs Lab scatter. But I guess that does make sense – that's the trendline of people who think there’s a real difference, through to those who think there’s no difference (reading right to left).

    You’ll note that there’s a cluster in the top left who think Labour’s extreme right and National is extreme left. Did that answer your question?

    Auckland • Since Nov 2006 • 9307 posts Report Reply

  • Speaker: Meet the middle,

    Attachment

    And same plot but only with National and Labour, for a blow up. The Nat vs Lab is the hardest to interpret. If it really says what it looks like it says, then that's highly counter-intuitive.

    Auckland • Since Nov 2006 • 9307 posts Report Reply

  • Speaker: Meet the middle, in reply to Brent Jackson,

    Attachment

    Actually I had a thought how to answer your question, Brent. I had to jitter the points, otherwise the graphs all just looked like a grid because the variables are categorical.

    Let me know if you don't understand how to interpret this graph...it's kind of overkill on your question :-)

    Auckland • Since Nov 2006 • 9307 posts Report Reply

Last ←Newer Page 1 2 3 4 5 931 Older→ First