The story of five Danish school girls who won a prize with their school experiment that allegedly shows that the electromagnetic radiation of WiFi routers has a negative effect on the germination of garden cress, has been reported by numerous sites on the Internet. Just check the number of Google hits while searching for “wifi cress”. The girls placed twelve plates with cress seeds on cotton in front of a window, watered those regularly an watched the seeds germinate. Between six of the plates WiFi routers were placed. After 13 days the cress was cut and dried and the germinated seeds were counted. A big difference was found. From the seeds which were held under radiation far less had germinated. Evidence for negative effects of WiFi? Nope.
Nice for the girls that they won this prize, but not so great that it’s promoted as a good example of science. It’s not the girls fault, but their study is an excellent example of how pseudo- or bad science can enter the classroom.
It was first published on a Danish website (Google translation) and then picked up by Geek.com and ABC News for instance. Not very critical these reports. Norwegian science journalist Gunnar Tjomlid had a good look at the study design and the report. He found a lot of things which are questionable. I suggest you read his excellent blog yourself (Google translated if you don’t read Norwegian). I will just mention the most important things from it:
- The WiFi and control group were not just different because of the presence of the routers. On the pictures in the report it can be seen that also the laptops in the WiFi group were placed quite near to the plates. It’s very likely that this had an effect on airflow and temperature around the plates and that could have an effect on germination, which has nothing to do with the presence of EM-fields. Not properly controlled.
- It was obvious what the WiFi group was and what control group. Not blinded.
- From communication with the science teacher of the girls Tjomlid learned that their had been two experiments. The first one had the routers only sending out the SSID. A second experiment in which the laptops had been ‘pinging’ each other constantly did not show the dramatic difference in germination. Only the first experiment was used in the report (not completely clear, because the teacher gave contradicting information on this). Publication bias: not reporting negative results.
The reports on blogs illustrated the difference in germination by photographs of plates with cress, one showing a full grown, not radiated, ‘healty’ one and a plate which almost doesn’t show any sprouted seed at all, a radiated, ‘sick’ plate. If you look at the actual reported results, they do not look that shocking: on average the control group had 332 sprouted seeds versus 252 in the WiFi group. Misleading representation of the result in the press.
- (not mentioned by Tjomlid). The plates in a group were not separated in space, so we cannot regard the results of individual plates as independent observations. In fact, you could argue this is an N=2 experiment. Faulty statistical analysis.
- The girls stopped the experiment on day 13. Not because that was a predefined moment, but because on that day the control group had reached the maximum height. The problem is that due to a difference in temperature of just a few degrees, it can already take a couple of days for the cress to grow to the same height. If there was indeed a difference in temperature due to the placement of the laptops, it would be likely that the WiFi group could have germinated and grown similar to the control group if it was allowed to grow on for a couple of days. They were just looking for the result they had in their mind beforehand. Biased towards a particular result.
- So how did these young girls get biased so badly? Well, they were only fed by literature which points to studies which have shown negative effects stemming from researcher who are discredited by main stream scientists. And for a possible (dangerous) working mechanism of EM-fields, they fully rely on a single report written by Thomas Grønborg, who at his tum relies on Olle Johansson (see further). Cherry picking.
Who are the scientists who are so enthusiastic about this poor study? The article on the Danish website mentions Olle Johansson, who received the ‘Misleader of the Year‘ Award from the Swedish skeptics in 2004. He is well known for having unsubstantiated ideas of negative health effects of radiation. He is cited in the Danish article as having plans to replicate the girls’ experiment in cooperation with senior researcher Marie-Claire Cammaerts from the Université libre de Bruxelles. We shouldn’t expect anything good from this replication, because as I’ve shown in a blog some while ago, Cammaerts probably cannot be trusted with this kind of experiments (see: Ants Performing Statistical Miracle under GSM Phone Radiation?). [Update 4-1-2016: Cammaerts and Johansson published a replication, in which they managed to make even more mistakes]
Tjomlid also mentions Andrew Goldsworthy, another well known fear monger, and Dutchman Niek van ‘t Wout, who is head of green space of a Dutch city and the instigator of research into the possible deteriorating effect of WiFi on trees (so he is not a scientist himself). After a not so convincing first experiment, Wageningen University started a follow up, of which we never heard again.
It’s quite clear that based on this experiment, you can’t draw any conclusion on the non-thermic effects of WiFi routers on germination. It’s a pity that the girls had this obviously biased teacher as a supervisor and that their work is now being used by pseudo-scientists as ‘evidence’ that EM-fields are very dangerous, while there is consensus that if there is a risk at all, it’s very low. The faults made can’t be blamed on the girls and let’s hope that this experience doesn’t affect their interest in research. It could even be a very good learning experience, if they are willing to have look at what went wrong, because it has so many aspects of bad science.
Don’t forget to read Gunnar Tjomlid’s blog, it entails far more interesting stuff than my summary: http://tjomlid.com/2013/05/19/
Update 26-12-2013: Thomas Guiot made a French translation on his website