On Friends, Rachel becomes pregnant with Ross in spite of using a condom. It comes as a shock to all when they hear that condoms are only 97 percent effective. Or, as Rachel puts it, condoms only work 97 percent of the time. But what does this really mean?
The causal mechanism of a condom is simple: it prevents pregnancy by keeping the semen contained, thus avoiding physical contact between the sperm and the egg. The condom is thus what we can call an ‘additive interferer’ (see figure below). This means that an interferer (the condom) is added to counteract the effect to which the causal situation (uninterrupted intercourse) tends (pregnancy).
Causes have an intensity, and the stronger they tend towards their effects, the stronger a preventer is needed to counteract them. But we don’t want to use only the required effort. Even a preventer can be counteracted. So the more unwanted the effect is, the stronger we will work to prevent it. If pregnancy does occur, then obviously the sperm and the egg somehow managed to meet after all, and the additive interferer failed to counteract the effect. A condom can break, for instance, if it is old or defective. This would be a case where the additive interferer fails to do its causal work, much like a security system that breaks down. But one can also fail to use the additive interferer correctly. A security system must for instance be switched on to do its causal work.
Considering the statistics, therefore, 3 percent failure to prevent pregnancy does not look too bad. There are, without doubt, a number of drunken mistakes involving condoms. So it shouldn’t surprise us if in some of these cases, the condoms were either defective or not correctly used. An important distinction must thus be made between how often condoms successfully protects against pregnancy (the total amount of times it works) and the degree to which it does so in a particular situation (the causal strength).
Say a couple is about to use a condom. How are they supposed to interpret the information that condoms are only 97 percent efficient? If the condom is intact and not too old, shouldn’t they feel safe to use it? Or is using a condom more like a lottery, where some unlucky few get pregnant no matter what? On a frequentist interpretation of probability, the effectiveness of a particular condom depends on the effectiveness of all other condoms. So if a huge batch of defective condoms reaches the market while our couple is inspecting their condom, its (statistical) effectiveness drops accordingly. But the condom itself has not changed.
Against frequentism, it seems reasonable to assume that the properties of the condom are what do the causal work of protecting against pregnancy. The latex is a suitable material because it doesn’t let any fluids through, thus also protecting against STDs. One can even fill the condom with water or blow it up into a balloon. So how can we reasonably say that the condom is less than a perfect means of counteracting pregnancy? Quite the contrary, the condom seems overdisposed for the modest task of containing a tiny bit of semen. Arguably, condoms are fool proof and more than fit for working in spite of alcohol, shaky hands, bad lighting, nervousness and other less than ideal conditions.
Now recall that Rachel said that ‘condoms only work 97 percent of the time’. This indicates a part-whole relationship, where we count the successful cases against all instances of condom usage. Ross, however, said that ‘condoms are only 97 percent effective’. This suggests that the causal powers of a condom are not strong enough to perfectly prevent pregnancy, and one could wonder why anyone would want to buy or even produce such a safety mechanism.
What this shows is that we must not confuse causal strength with part-whole relations, where the part is the number of successful cases and the whole is the total number of attempts. But in addition, one must not confuse causal success in the particular case with the percentage of causal success on the population level. What other people do with condoms elsewhere does not influence what happens to this particular couple and their condom.