Reader Comments

Post a new comment on this article

a tainted sample gives tainted results

Posted by frontierscientist on 06 Feb 2009 at 14:17 GMT

There are some obvious errors in this study, which are the same erros I have commented upon in previous studies from the Danish Cancer society on mobile phone and power line radiation and cancer. etc.
The sample contains more that one kind of mobile telephony: in 1982 no GSM service was available in Denmark, so these older subjects had NMT handsets, which broadcast on a different frequencies...this sample was mixed with a newer sample, which includes GSM and GPRS handsets, but not 3G handsetw...therefore this sample is tainted...

As I have already pointed out for previous studies, this sample only SUBCRIBED to a mobile telephone service, no one actually know is all 420.000 people actually USED their phoes, how often, how close to the head, etc. so this is also a tainted sample!

I will have a closer look at the statistics on this study, to see if the significance level is affected by the use of a large, inhoogeneous, and mutiply-tainted sample....

RE: a tainted sample gives tainted results

BobOH replied to frontierscientist on 08 Feb 2009 at 15:02 GMT

Frontierscientist - you might want to read the paper before commenting on it.

If there were a problem, then the mixed telephony methods would lead to a mixture of different effect sizes, hence if only GSM (for example) was problematic, then effect would be stronger than the one estimated here.

Your comment about use not being the same as subscription is correct, and is discussed in the paper. The conclusion, again, is that the effect would be underestimated.