Wednesday, December 13, 2006

Online readers = readers?

When conducting an online poll, one have to be very careful when interpreting the results, especially when generalising the results of the sample to a whole population. In this article on the news site of the Gazet van Antwerpen, de journalist is paying attention to the lack of representivity of an online poll: instead of writing that the 'Flemish citizens' (a population often used in Flemish newspapers) do not agree with the political coalition in Lier, he states that the 'readers' do not agree. But the good work is only half done ...

Besides the possible confusion thanks to the word 'reader' (all Flemish readers? Worldwide readers? Print or online readers? Oh ... only the online readers of the Gazet van Antwerpen) , this generalisation does not take into account the fact that the sample is not drawn out of all the readers of the newspaper, but instead is based on a self-selection of online readers. Thus, the title of the article should have been "online readers of the GVA do not agree with ..." or even better "online readers of the GVA who participated in the online poll do not agree with ...". But of course, this doesn't sound right.

In the article itself, the interpretation is rather good: the journalist explains that the sample is composed by self-selecting, online readers of de GVA, but nevertheless, the title on top of this article leads to some extent of confusion for readers who do not read the whole article.

More about this kind of statistical difficulties and challenges on this site.

No comments:

Add to Technorati Favorites