Example 4
Declining semen quality (chapter 22)
Back to front page                                                                                                              To list of six examples

Summary. In order to assert that there is no general decline in human semen quality, Lomborg tells us that earlier measurements were mostly from New York, where the semen quality is high, whereas later measurements are mostly from elsewhere, where the semen quality is lower. So, he claims, the purported decline is just an artifact due to geographical variation. He substantiates this assertion by referring to the latest New York study which still shows a high semen quality - although Lomborg has known from the beginning that this late study is suspect. He postulates that if data from New York are excluded, the reamining data no longer show a significant decline - although  he knows that this claim is also wrong. In fact the meta-analyses available - also to Lomborg - show that there has been a significant decline in semen quality in North America and in Europe. The most recent papers, which Lomborg could have found if he had searched for them, reinforce the impression of a continuing decline.

   On pages 238 - 241, Lomborg deals with the subject of declining semen quality.  This subject is important because if the levels of poisonous chemicals to which we are exposed are causing a reduction in the production of human sperm, then we must take the threat from these chemicals very seriously. Lomborg, however, claims that there is no danger from synthetic chemicals. One of the things he must do to maintain this claim is to demonstrate that there is actually no decline in semen quality.
    Concern about semen quality emerged in 1992, when a group of four scientists based in Denmark published an overview (a meta-analysis) of previously reported sperm counts from 1938 to 1990. Lomborg presents their data in his Figure 129. The conclusion that is evident from the figure is that the average sperm count decreases during the 20th century. So, Lomborg sets out to pull this conclusion apart bit by bit until he feels able to completely dismantle it.
    The dismantling process starts when Lomborg writes that the first four large-scale studies (for 1938-51) are from a single city, New York, and postulates that sperm counts in New York - due to unknown causes - are higher than in most other places, namely around 130 million/ml. This means that there is an overweight of high figures at the start of the time series, and the subsequent decline (he maintains) is therefore an artifact. In fact the average sperm count for these four studies from New York is 112, not 130 as he postulates.
    Let us then look at the trend over time in New York. There were seven studies, which, in chronological order, produced the following sperm counts: 121 - 134 - 101 - 107 - 110 - 79 - 132. Lomborg´s whole argumentation hinges on the last study (covering the period 1972 to 1994) that yielded 132. This study is suspect of having been influenced by a certain chemical company; if it is disregarded, the time series for New York shows a decline, like elsewhere; but by stressing just this result - knowing that it is suspect - Lomborg is able to maintain that there is no decline here.
    Next he postulates that if the data from New York are excluded from the data material, then there is no longer a declining trend among the rest of the samples. This is simply a lie. The original meta-analysis by the Danish group has been reanalyzed in a paper by Shanna Swan et al. from 1997, which Lomborg has read. Their conclusion is that even if the data are split into different regions, a significant declining trend remains for USA as well as for Europe.
    Lomborg has even phoned Shanna Swan to get additional information. It is therefore strange that he is not aware that Swan and co-workers published a new meta-analysis in 2000, in which they included more samples and especially considered new data from the period 1990-2000. Most of these new samples have given relatively low sperm counts, which means that the declining trend is even more firmly established in 2000 than it was in 1990. For instance, the Danish group published a new paper in 2000 in which they demonstrate that in recent years there has been an additional large decline in sperm quality in Denmark. Many studies from elsewhere corroborate the decline, but when Lomborg wrote the English version of his book in 2001, he did not cite any of these new studies. Does that mean that he simply did not care to update his book on this point, and thereby missed the new evidence ? No. He did make a sort of update. His note 1860 contains references up to the year 2000. But the references that he cites are those that support one of his own assertions. He could very easily have searched the database Medline and found many papers that document the further decline in sperm quality, but he did not.
    According to Lomborg´s own professed standards, one should not focus on data from single localities, because local trends do not tell us what is the global situation. One should consider all the evidence, world-wide. This is what Shanna Swan and co-workers have done, and they find that the postulated decline is real, significant, and continuing. Although Lomborg has read her paper from 1997, and talked with her in the telephone, he does not refer to her results. The global data are there - especially in Swan´s next paper from 2000 - but Lomborg does not cite them, obviously because they do not support his postulates. Instead he focuses on the local variation, which, according to his own professed principles, is precisely what he should not do.
    As to Swan´s paper, he does not cite its conclusions on the general trends, but cites it only as evidence that shorter abstinence time between ejaculations now than previously could explain much of the decline. However, the papers cited by Swan on the subject of abstinence time do not support Lomborg´s assertion that this factor is an important confounder. Thus the only use that Lomborg makes of this paper is a misuse.
    On page 240-241, Lomborg goes on to dismantle the conclusion that organic farmers have higher sperm counts than greenhouse gardeners that use pesticides. Just as on the preceding pages, this dismantling is completely unwarranted. The conclusion that organic farmers have relatively high sperm counts is actually robust.
    We happen to know that Lomborg´s information sources on the subject of declining semen quality were not biased. So the bias in Lomborg´s text is not due to what was available to him. The bias is deliberately produced by Lomborg himself.