Stats: Are the media ignorant or willfully misleading people?

If newspapers are going to quote statisitics then either the writer or the editor should have at least some knowledge of statistics. Anyone who has done the slightest bit of statistics knows making an inference based on such a small sample is ludicrous. I can't believe such a **** article made the front page of any newspaper.

IRELAND has an alarmingly high rate of family murder suicide, with the incidence of familicide here seven times that of the UK and six times higher than the US, the Sunday Tribune can reveal.

Despite a population of just over four million people, there were four cases of familicide in Ireland in 2007 - one case for every one million people. In the UK, familicide occurs significantly less often, with only 1 case per 7.5 million people.

The rates in the US are one case per six million.

Already this year there has been one family murder-suicide in Ireland - last weekend's tragedy in Clonroche, Co Wex fo rd .

..........

http://www.tribune.ie/article.tvt?_scope=Tribune/News/Home%20News&id=87388&SUBCAT=Tribune/News/Home%20News

jdivision

It's not a small sample, it's based on the entire population. Having said that it's mad that it's that rare and made the front page

mike65

Extrapolation of such a small number of incidents would be nonsense. 1 in a million and 1 in 7.5 million in a given 12 month period is a difference best not examined too closely.

Mike.

jmcc

illegalheadbutt wrote: »

If newspapers are going to quote statisitics then either the writer or the editor should have at least some knowledge of statistics.

Most journalists in Ireland seem to be from an Arts background. Things like Science, Mathematics and Technology are way beyond them. But then accurate stats would not make for such a sensationalist headline for a newspaper long past its best.

Regards...jmcc

chekov

I figure that they generally just don't care - the only important thing is "can i possibly cherry pick a sensationalist headline from these numbers".

In this case, the important sample size is 1 - they only looked at the numbers from a single year. I wonder could anybody be dim witted enough to think that a single year's figures of an enormously rare event is enough to reach any conclusion at all. If the figures in Ireland were the same as they were in the UK, you would expect to see clusters in single years and none in other years. My back of the envolope calculations are that the statistical significance can be summed up as "snort, snort, snigger".

The Sindos in-house polls are another good example of the famous techniques described in the huff's famous pamphlet: http://en.wikipedia.org/wiki/How_to_Lie_with_Statistics

lostexpectation

calculating over the past year is no proper. your right i saw that story i thought it was junk

BurnsCarpenter

jmcc wrote: »

Most journalists in Ireland seem to be from an Arts background. Things like Science, Mathematics and Technology are way beyond them.

Yeah, basic statistics are a stretch for us poor Arts graduates. :rolleyes:

J.S. Pill

lostexpectation wrote: »

calculating over the past year is no proper. your right i saw that story i thought it was junk

Story = junk

Paper = junk