Following this previous article, a friend got in touch to thank me for disproving some astrological ‘nonsense’. I replied that I had not disproved anything – I had just failed to find supporting evidence – but it did get me wondering about the nature of the conclusions that can be drawn from this sort of analysis.
Suppose, for the sake of argument, that people born under Aquarius do show a significantly higher propensity to become composers than those born under Virgo. Consider these three possible explanations…
- astrologically, the characteristics of Aquarians make them more likely to compose music than Virgos;
- the seasonal variations in daylight, work and sleep patterns, food and other factors affect the growth of the developing foetus in such a way that babies born in February (Aquarians) go on to have a greater propensity for composing music (rhythm and pitch perception, dexterity and coordination, ability to practice for long periods), on average, than those born in September (under Virgo);
- working musicians are more likely to conceive in May (perhaps due to typical seasonal social or working patterns), compared to the population as a whole, than during the busy period around Christmas, and their children are more likely than average to be musically gifted.
I am no expert on astrology, foetal development or the mating habits of musicians, but these explanations seem to be at least plausible, although they are all complex and would be difficult to test, for multiple and different reasons. They might all be partly true: equally, there may be other explanations that would have the same result. To ‘prove’ any of them as the sole explanation would require a convincing argument against each of the alternatives – again, very difficult to do.
And if one of the non-astrological explanations were the ‘real’ one, that could be regarded as supporting evidence in favour of the astrological interpretation – just in different (i.e. more modern, scientific) terms.
Even before we get to this stage, we need to ask exactly what the statistics are telling us. Most of the time, the best that statistics can demonstrate is a correlation – a tendency for one thing to vary in line with something else. Statistics does not usually help in sorting out whether one thing is causing another (perhaps indirectly), or whether both are being influenced by some unknown third factor (which might simply be some sort of bias in the data).
In the previous article, I argued that variation in the distribution of composers’ star signs is largely explained by the normal seasonality in birth rates for the population as a whole. However, this ignores the likelihood that the seasonality varies quite significantly between countries and time periods, or between different socio-economic classes, religious groups, or other factors. The fact that the overall distribution of composers does not look unusual in aggregate does not imply that there are not some particular sub-groups where there is a significant correlation between musical aptitude and star signs.
The point is that statistics (or indeed any other research methodology) can only ever give you evidence – not a definitive answer. It has to be carefully weighed up alongside all of the other evidence (whether qualitative or quantitative), taking account of the characteristics of the data and methods used, and considering all of the possible alternative theories of what might actually be going on. Even in the physical sciences, this is a very difficult thing to do, and requires experience, deep understanding, and careful analysis. In the humanities – in music history as much as in politics, business, economics or astrology – it is often impossible fully to reconcile all of the evidence (but that of course does not mean that we should not try). Research often generates more questions than it answers. We have to keep developing new data, ideas and techniques to be able to understand these things better, but we must also recognise that they may well be too complex for us ever to comprehend completely.