Before scientists start trying to explain the ‘unexplained
difference in sex ratios’ found by V. Grech et al. (vol. 324, p.
1010-1), it might be wise to take a closer look at their data and
methods. Several objections to the publication can be raised.
To get their results, the authors banded European countries by
latitude (<_40 degrees="degrees" _40-55="_40-55" and="and"/>55 degrees). These bands
are quite wide. Newborns in northerly Gdansk (54 degrees) will be
surprised to learn that they count for Central-Europe just like
those born in Avignon (44 degrees). By the samen token, Copenhagen
(56 degrees) is Northern Europe, Edinburgh at the same latitude is
Central. Milan (45 degrees) is Southern Europe, but Bordeaux at the
same latitude is Central again. Bulgaria is considered to be
Southern Europe, even though the country extends from 42 to 44
degrees.
If this does not make the whole exercise rather futile, the
statistics are no less absurd. Obviously, large data groups give
better statistics than smaller groups, but this is going too far.
Comparing the southern countries with Central Europe, the authors
find a chi-square of 57, which they say is significant. If the table
is correct, this chi-square should be 177, which is even more
significant. But is it relevant?
Thanks to the unheard-of precision, the investigators find a
difference in the fourth decimal place: 0.5148 against 0.5140. In
other words, for every 10,000 baby’s born, there were 8 boys more
in ‘Southern Europe’ than in Central Europe. For Northern Europe,
where we would expect still less boys to be born, we find a sex
ratio of 0.5140. Still, the chi-square for trend is significant,
according to Grech et al., and BMJ duly publishes ‘significant’
findings.
In North America the trend is just the other way around, say the
authors. Since Canada has been omitted from the table, this is hard
to check, but the reversal can only be ascribed to the very low sex
ratio in Mexico: 0.5087. This seems to be the only interesting
finding from the study: such a low sex ratio is remarkable, if not
suspicious. Do they kill young girls in Mexico? Is there selective
abortion? Can we trust the data?
Hans van Maanen
Science editor
‘Het Parool’,
Postbus 433,
1000 AK Amsterdam,
The Netherlands
Rapid Response:
Sex ratios & silly significance
Before scientists start trying to explain the ‘unexplained
difference in sex ratios’ found by V. Grech et al. (vol. 324, p.
1010-1), it might be wise to take a closer look at their data and
methods. Several objections to the publication can be raised.
To get their results, the authors banded European countries by
latitude (<_40 degrees="degrees" _40-55="_40-55" and="and"/>55 degrees). These bands
are quite wide. Newborns in northerly Gdansk (54 degrees) will be
surprised to learn that they count for Central-Europe just like
those born in Avignon (44 degrees). By the samen token, Copenhagen
(56 degrees) is Northern Europe, Edinburgh at the same latitude is
Central. Milan (45 degrees) is Southern Europe, but Bordeaux at the
same latitude is Central again. Bulgaria is considered to be
Southern Europe, even though the country extends from 42 to 44
degrees.
If this does not make the whole exercise rather futile, the
statistics are no less absurd. Obviously, large data groups give
better statistics than smaller groups, but this is going too far.
Comparing the southern countries with Central Europe, the authors
find a chi-square of 57, which they say is significant. If the table
is correct, this chi-square should be 177, which is even more
significant. But is it relevant?
Thanks to the unheard-of precision, the investigators find a
difference in the fourth decimal place: 0.5148 against 0.5140. In
other words, for every 10,000 baby’s born, there were 8 boys more
in ‘Southern Europe’ than in Central Europe. For Northern Europe,
where we would expect still less boys to be born, we find a sex
ratio of 0.5140. Still, the chi-square for trend is significant,
according to Grech et al., and BMJ duly publishes ‘significant’
findings.
In North America the trend is just the other way around, say the
authors. Since Canada has been omitted from the table, this is hard
to check, but the reversal can only be ascribed to the very low sex
ratio in Mexico: 0.5087. This seems to be the only interesting
finding from the study: such a low sex ratio is remarkable, if not
suspicious. Do they kill young girls in Mexico? Is there selective
abortion? Can we trust the data?
Hans van Maanen
Science editor
‘Het Parool’,
Postbus 433,
1000 AK Amsterdam,
The Netherlands
Competing interests: No competing interests