Health literature has long noted a positive correlation between health and levels of education. Two competing theories have been advanced to explain this phenomenon: (1) education "causes" health by allowing individuals to process complex information and act on it; and, (2) education and health are merely correlated through some third underlying characteristic.
Determining which of these two theories is correct is of importance to public policy. But that task is empirically difficult because, from the standard, static perspective, the theories are observationally equivalent.
We exploit a way in which the two theories have different implications regarding the sort of behaviour we should observe over time. We use smoking as a measure of health behaviour and find that smoking rates between "high" and "low" educated individuals expand when information is hard to process, and then contract as it becomes more easily processable. This approach is then repeated using physical activity as a measure of health-related behaviour to address limitations of the smoking model.
Our novel approach to estimating the differences in the behavioural responses to changes in the processability of health-related information, across education groups, provides strong evidence in support of the view that education and health are causally linked.
Author Keywords: applied statistics, education, health economics, public health, public policy, smoking