Tuesday, 19 May 2009

We're from the internet, we're here to help (part 3)

Was there a point in me doing the table in the previous post? I don't recall, it took me ages and I lost track of my intentions.

Right, Mark Reckons, he's got all the MP's expense totals ranked and all the majorities ranked and he's done some quartile graphs that show a correlation, the more secure the MP, the higher their majority, then the more likely they are to claim huge expenses.

I've run pretty much the same data and here's a scatter graph with majority rank along the bottom and expense rank is along the side.

It ain't a straight line as MarkWadsworth predicted in a comment the other day. But MarkReckons found a correlation, and I'm going to find it too. You just can't see it from a noisy scatter graph thing.

So, my next graph shows the correlation in deciles, I guess they're like the quartiles of Mark Reckon's chart, but there's ten intervals instead of his four.

You can see there's a bit of a correlation, one end's lower than the other, but in the middle its all a bit wibbly.

So I guess, validating Mark Reckon's conclusions, the MPs with the biggest majorities are most likely to claim the most expenses, and the MPs with the smallest majorities are likely to claim the lowest expenses. And all the MPs in the middle, they're all wibbly.

1 comment:

  1. Even from this one it looks possible to identify various clusters, though any conclusions that can be drawn from them depends on deeper analysis.

    So how does the scattergraph look if you try totals instead of rank?