Ratios in the evolutionary behavioural sciences: problems and solutions
Ratios are widely used in the social and life sciences, particularly the evolutionary behavioural sciences. However, over the past few decades, several problems have been identified with them leading many disciplines to abandon their use. Ratios show statistical properties that are not always obvious, are difficult to interpret properly and can have unintended implications. Uncertainty about their precise meaning impedes the development of effective theory. Arbitrary choices of which variable is used as the numerator and which is used as the denominator can lead to different results. Additionally anthropometric variables rarely remove the influence of body size and can induce spurious effects. Fortunately, advances in statistical techniques and software mean that most of these issues are easily resolved and have rendered ratios mostly unnecessary. Any researcher with a firm grasp of multiple regression can avoid the pitfalls associated with ratios. This review aims to introduce readers to the many issues associated with ratios and how to overcome them, with a special emphasis on the ratios most widely used in the evolutionary behavioural sciences. Using open access datasets I demonstrate both the potential problems with ratios and their solutions. I outline the times when ratios are acceptable and useful, and when our science would be more rigorous and productive without them.