|Alexander Stubb, Leader of conservative party and Prime Minister of Finland.|
fredag 13 mars 2015
Is there a collaps of the finnish Kokoomus party?
Dear statistic-friends (Rakas tilastoystävät),
I am sending a blogpost from Helsinki in Finland where I have been working all week.
One of the students showed me an interesting clip from the finnish TV channel YLE. Later this year there will be a parliamentary election in Finland and the comments on the polls become more and more frequent. The president and the prime ministers are both representing the Conservative parti Kokoomus (Kok).
According to a recent survey presented in YLE Kok is now just third biggest party. Following the Centre party (Kesk) and Socialdemocrats (SDP) but before the True finns (PS) and the Green party (Vihr). The result was presented as a “collaps” of the government party . They are now only third, and their main competitor, SDP, is bigger!
But this is not all true. As we can see from the table and the figure, Kok is 3rd, but the difference between them and SDP and PS is small.
Infact, if we test all pairs of proportions we find an interesting pattern: There are clear and distinctive groups, or clusters, of parties in this poll. Centre party (kesk) form one group. Socialdemocrats (SDP), Consevative (Kok) and True finns (PS) are basically equivalent. And there is no significant differences in the third group, not even between the biggest and smallest party (Green (vihr) versus Other parties).
However, the Conservative is significantly smaller than the Centre party. On the other hand, so is the Socialdemocrats.
In order to get a significant difference between SDP and Kok we would need a poll with almost 200000 respondents!
The lesson to be learned from this story is that ranking data can cause misinterpretation. In sports it is important to be the best. There is only one winner, one silver and one bronze medal and there is always a separation between. The differences between SDP, Kok and PS are exaggerated when they are replaced by ranks, while the difference between PS and Vihr is actually looking smaller.
And always be careful when reading statistics processed by politicians and journalists.
I thank Oskari Tinkanen for showing me the data!