I've been reading up for my thesis, on voting behavior. Anyway, I'm currently reading an article that deals with the concept of party identification. It brings data from two panel surveys in Germany and Britain, that deals with which parties respondents identify with over time.

At one point, they discuss people who named only one of the large parties, and did so at least once during the survey years (9 in the UK, 14 in Germany). They say that there are two types of people - those who vote for that party always (or almost always), and those that vote so "haphazardly", or at random. In their own words: "Of those who do not always choose their party, the mean rate of selection is four in eight years, indicating a pattern that resembles a coin toss between “yes” and “no” for each year." Seems reasonable. Only that merely stating the mean number of times they chose that party doesn't really indicate anything of the sort. It could, but it doesn't have to. It doesn't take a mathematical genius to know that there would have to be a "normal" curve if that was indeed the case - if people chose the party at random, most people would chose it 4 times, and the least would choose it 1 or 7 times.

Let's look at the graph, shall we?





Hrm, the results are exactly opposite to what we would be looking for! Not only is there no peak at 4, there is, in fact, a low-point just there. If anything, this data shows that people don't choose "haphazardly", and only a few are so unstable as to be as likely to choose the party as not.

This sort of stupid academic dishonesty really annoys me. I mean, the graph is right there in front of me - I'm not stupid. You're only going to fool the poor saps who don't know shit about statistics. That hardly seems fair or smart. It's abuse. It also, as it so happens, topples their basic hypothesis in this article (from the abstract: "Most take a definitively
negative stance towards one of the parties and a positive stance towards the other major party. Of these, about half display behavior that reflects a psychological commitment and about half are as likely as not to pick that party when asked").

Fa.