This is a well-written piece, but I generally found the author to lean too much in to the 'we can't be absolutely sure, so we know nothing for certain' type of argument. He points this out himself, saying that the presence of conflicting data suggest that we need to "reason under uncertainty".
I think his approach is characterised well by the following section:
"1. Yes, okay, that line is pointing very slightly down, and apparently this is statistically significant.
2. But also, the data are very noisy. Some studies from 2005 show higher sperm counts than most studies from the 1970s. The biggest pre-1980 study shows sperm counts very similar to today’s."
The style of language "yes, okay" and "apparently" soften the position of statistical significance and the counterpoint is vague and dated. Again, there seems to be a bias towards claiming "we just don't know", whereas the data show that there is a statistical trend.
I am not saying that I take any single paper as gospel, but the general conclusion he comes to: 'let's wait 20 years and see what happens", feels like a call to inaction, which can be risky.
Also, for a meta-analysis, the article is not very well-referenced and doesn't seem all that broad. Having done a couple of these myself, I would have expected some kind of summary of the number and size of studies pointing to each/no conclusion. This shortcoming makes it easier to come to his stated conclusion that 'we just don't know'.
I think his approach is characterised well by the following section: "1. Yes, okay, that line is pointing very slightly down, and apparently this is statistically significant.
2. But also, the data are very noisy. Some studies from 2005 show higher sperm counts than most studies from the 1970s. The biggest pre-1980 study shows sperm counts very similar to today’s."
The style of language "yes, okay" and "apparently" soften the position of statistical significance and the counterpoint is vague and dated. Again, there seems to be a bias towards claiming "we just don't know", whereas the data show that there is a statistical trend.
I am not saying that I take any single paper as gospel, but the general conclusion he comes to: 'let's wait 20 years and see what happens", feels like a call to inaction, which can be risky.
Also, for a meta-analysis, the article is not very well-referenced and doesn't seem all that broad. Having done a couple of these myself, I would have expected some kind of summary of the number and size of studies pointing to each/no conclusion. This shortcoming makes it easier to come to his stated conclusion that 'we just don't know'.