A 38% probability the story is about nothing

Stuff has a story from Reuters which says:

Support for Scottish independence rises

I got excited. There must have been a big movement in the polls. So I read the story.

The first opinion poll since Britain’s rulers warned Scots they would lose the pound if they voted to leave the United Kingdom showed a slight increase in support for independence ahead of a referendum on the issue scheduled for September 18.

Slight? So it wasn’t 5%. Maybe 4%? I presumed it a significant amount as Reuters has done an entire story on this.

A Survation/Scottish Daily Mail poll carried out on February 17 and 18 of 1,005 people found 37.7 percent support independence, which it said could be compared to 36.9 percent recorded in a PanelBase/Sunday Times poll carried out on January 29-February 6.

First of all comparing the results of one company to the results of another, to declare a change is bad enough. But even if the results were from the same company, is an increase of 0.8% significant?

My probability calculator tells me the two results mean that there is a 61.6% chance the second poll is actually higher than the first poll To turn that around, there is a 38% chance there has been no increase at all.

Hardly worth a story.

Personally I don’t think Scotland voting for independence would be a bad thing for England. But anyway the point is that Reuters have written a story about pretty much nothing.

Comments (24)

Login to comment or vote

Add a Comment