> Studies also indicate a continuous decline in testosterone (in males) over 20th century. I'd welcome someone correct my biological knowledge, but I would guess that testosterone influences production of sperm, not the other way round. So they're probably studying the wrong thing.

Yup. And our society also happens to encourage activities which lower testosterone levels (various things like inactivity, weight gain, poor sleep and stress all sap testosterone levels).

>And our society also happens to encourage activities which lower testosterone levels

also important that natural selection in our society isn't for higher testosterone anymore. Peaceful interaction and coexistence lead to more success than open aggression ("success" here is in natural selection sense, i.e. producing next generation and bringing them into "successful" (in the same sense) adulthood)

Any study looking for long term health effects of specific behavior, no matter how good the methodology, is going to wind up looking stupid in the end. There are simply too many variables to account for.

With so many variables to consider, of course the scientists will be able to wrangle some explanation from the data that fits their hypothesis.

This kind of study makes a mockery of the scientific method by following it in name only.

I'm sure you could come to the same conclusion for dozens of factors other than cell phone radiation. How about hormones in the food supply? Work, sleeping habits? Changing life timelines? Social media?

You could attribute literally any new phenomenon from the past century to a correlated observation.

This seems like such poor science it makes me angry my tax dollars are paying for this so-called "research." What a waste of time.

Relevant: http://kill-or-cure.herokuapp.com/

Also isn't clear at all if they're accounting for multiple comparisons. Given they're looking at (glancing through they paper) at least seven or eight cell phone related factors, it'd be pretty crucial here.

The authors note that it is a contributing factor, which means it is not the only factor.

I'm not really sure why people referred for semen analysis would be a biasing factor in this when it is present in both test groups. It seems like it would be the only reasonable way to get statistical power in a study like this since if the effect is small and most people don't suffer from it there would be a much larger sample size required.

> I'm not really sure why people referred for semen analysis would be a biasing factor in this when it is present in both test groups. It seems like it would be the only reasonable way to get statistical power in a study like this since if the effect is small and most people don't suffer from it there would be a much larger sample size required.

You're right, everything else being equal, that probably wouldn't be the biggest issue... Although in this specific case, it seems like a way to amplify the effect - if this study would be done on the general population, where (1) almost everybody uses mobile phones extensively, and (2) presumably, men are less likely to have poor sperm than in this focus group, the resulting effect would be diluted (as in most cases, mobile phone usage would not result in poorer sperm quality).