I was reading this article from Claire Packham about the latest release of unemployment numbers in the UK and I just wanted to share my thoughts on how statistical results are often published and interpreted.
In the case of unemployment the authorities can't ask everyone if they are working every 3 months, in fact I have never been asked. Thus, they ask a random sample of the population in order to reach statistically significant results or conclusions about the overall population.
Therefore, as stated in the article I mentioned above, the increase of the unemployment figure was estimated to be 38,000 inside a 95% confidence interval between -49,000 and 125,000. This means that if the "experiment" or survey is done 100 times, then the true value of increased unemployment will be in 95 of the intervals obtained. Therefore, we need to be careful when survey results are generalised and published by the media.
I liked the article because it allowed me to make sense of the statistic courses I have done. In fact, I wish in my undergrad we could have worked in real examples such as this one to make sense of the statistic concepts. Do you have any other examples of survey results being generalised by the media?
Posts
Hi Gonzalo!
ReplyDeleteWell.. I think you have been asked at least once!
On March 2011 you might remember that there was the UK Census. With that, you can have the complete understanding of all people unemployed!
Very nice article!
hugs
Alberto
You are right Alberto, we were asked in the Census this year, but the monthly unemployment numbers don't come from there.
ReplyDeleteIn fact I'm not sure if the results are out yet, which makes it difficult to measure unemployment dynamically from it (It is September and we would be getting March's numbers).
Thanks for your comment!