Let’s talk about risk for a second. And in particular, lets talk about bacon.
If the article above is to be believed, eating two slices of bacon daily can increase your risk of pancreatic cancer by almost 20%. So here’s a simple question — if you eat bacon daily at those levels, what is your risk of pancreatic cancer?
- Above 20%
- Below 20%
- Exactly 20%
- We don’t know.
The answer is that we don’t know. If you think about this a minute, you’ll realize that this is a variant of the base rate problem we’ve mentioned before. Your risk is 20% higher — but 20% higher than what? As the article indicates later on:
But experts cautioned that the overall risk of pancreatic cancer was relatively low. In the UK, it is one in 77 for men and one in 79 for women.
In other words, the actual base risk is about 1.4%. Eating bacon will raise that risk from 1.4% to (1.4 * 1.20) or 1.7%.
Please, don’t get me wrong, there’s plenty of good health reasons to not eat bacon. But given the small absolute impact on pancreatic cancer, pancreatic cancer is probably not the biggest strike against bacon.
What you are mostly worried about in health is absolute increases and decreases in risk.
So are relative increases worth looking at? Absolutely! Outside of risk, there is a lot of use for them. For instance, when we can’t comprehend raw numbers, relative increase helps us understand them. If we were to say that the population of Manchester, NH increased by 20,000 last year, it might be difficult for us to comprehend what that means. If we say it increased 18% in one year, that gives us some context. We have a sense, in that case, that a city’s ability to absorb new people is relative to its size. The same is true of increases in the deficit, or immigration — the relative increase helps us to comprehend the rate of change, which gives us some hints as to whether the change is dramatic or run-of-the-mill.
Focusing on relative change can also be useful in comparisons. Above, we see the percentage change in state prison populations. The female prison population is a fraction of the male population in absolute terms. But when we look at percentage change we can see the pattern that absolute figures might hide — male prison population growth is approaching zero, but female population growth rates are remaining high, tracking male trends but growing at almost double the rate.
In fact, without knowing anything about the total number of prisoners in the population, the relative increases and decreases allow us to spot interesting spikes and clear long-term trends.
Rather than relying on only one or another, the smart thing is to compute both absolute and relative change, and then ask yourself which figure answers the sort of question you are asking.