Tuesday, November 22, 2011

Respect my analysis!

Here's something quite amusing: a blog post (in Polish) written by an active politician and political science PhD, with some quantitative analysis of the situation in Sejm (a Polish equivalent of the House of Representatives) after recent elections. Or something the author thinks is analysis anyway.

Because he's all analytic and quantitative, the author kicks things off with a graph:


which he explains as follows: "The above graph depicts the distribution of power in Sejm. The percentages shown on the graph are percentages of seats belonging to each party."

So far, so good. Nothing inherently wrong with the graph itself (except maybe for labels; do we really need precision of two decimal places when talking about percentages of parliament seats?) It's the conclusions he draws from this graph that decidedly move his post from merely trivial to utterly ridiculous. He writes:
What are the implications of the above graph? 
First, the overwhelming advantage of PO. Even when compared to the largest opposition party (PiS), PO seems extremely powerful.
All this may very well be true, but it definitely is not a consequence of "the above graph". To make these kinds of conclusions we'd need to know many things that the graph doesn't tell us, like what kind of majority is needed to do what (do you just need plurality? Simple majority? Qualified majority? Maybe it depends on what you're voting on?), or what the likelihood of many possible coalitions is. For example, what if you need more than 50% of seats to do anything at all, and all the other parties hate PO and will never vote with them? Where's the overwhelming advantage?

But that's not all.
Second, the central position of PO. This party not only has the most seats, but is also situated in the most comfortable spot--the very center.
Oh god, where do I start. What is he talking about here, what does "central position" mean? In the center of what? The way he explained the graph, it is a visualization of one variable only: percent of parliament seats. Now he's talking as though the X axis measured another variable, the values of which are used to establish a linear order among parties. What is that variable? Why is is it not explained and why is the axis not labeled? Because according to the explanation, there is no such variable, so if I decide to order parties alphabetically by name, I'll be equally justified in claiming that PO has very little room for maneuver because it's second from the left.

No comments:

Post a Comment