5.3: What is the specific problem that researchers working within relational meta-theories have created?

We assume that there are many different causal bundles that are sufficient to produce the same developmental pathway, and we assume that all of them are biopsychosociocultural packages. If we lived in a world where we were hunting for a single necessary and sufficient cause (like the tubercle bacillus as a cause of tuberculosis), our job would be easier because those kinds of causes create a recognizable signature of covariation (in which the outcome never occurs when the cause is absent, and always occurs if the cause is present). Before we get to feeling too sorry for ourselves, however, we should note that almost no one in the social sciences, or even in epidemiology or medicine, believes that important outcomes are caused by single necessary and sufficient causes (see box). Even exposure to our tubercle bacillus won't cause TB if there is sufficient host resistance. As pointed out by Rutter (2007), “With very few exceptions, there is no such thing as a single necessary and sufficient cause. It is not just that multifactorial traits or disorders have multiple causal influences, but also that several different causal pathways may all lead to the same endpoint” (p. 378). So we are looking for a “constellation of components acting in concert” (p. 378) or actually several constellations since there are always several different pathways. That means that we will need to sort out the possible causal factors into effective (or sufficient) bundles (what Judea Pearl, 2009, calls “causal beams”), including the identification of factors that can’t be in the bundles or else they will preempt or nullify the impact of the causal beam.

“Necessary and Sufficient” or “Insufficient but Necessary Components of Unnecessary but Sufficient Causes”? Despite the fact that most people would start out by defining a cause as something that is both necessary and sufficient to produce its effects, it turns out that the kinds of causes studied by social scientists almost never satisfy those claims (Mackie,1965). Mackie examines this idea through the illustration of a house fire that investigators conclude was caused by an electrical short-circuit, even though the short-circuit was not enough by itself to start the fire (i.e., it was not sufficient) and the fire could have started in a variety of other ways (i.e., it was not necessary). “At least part of the answer is that there is a set of conditions (of which some are positive and some are negative), including the presence of inflammable material, the absence of a suitably placed sprinkler, and no doubt quite a number of others, which combined with the short-circuit constituted a complex condition that was sufficient for the house's catching fire-- sufficient, but not necessary, for the fire could have started in other ways. Also, of this complex condition, the short-circuit was an indispensable part: the other parts of this condition, conjoined with one another in the absence of the short-circuit, would not have produced the fire. The short-circuit which is said to have caused the fire is thus an indispensable part of a complex sufficient (but not necessary) condition of the fire. In this case, then, the so-called cause is, and is known to be, an insufficient but necessary part of a condition which is itself unnecessary but sufficient for the result. The experts are saying, in effect, that the short-circuit is a condition of this sort, that it occurred, that the other conditions which conjoined with it form a sufficient condition were also present, and that no other sufficient condition of the house's catching fire was present on this occasion. I suggest that when we speak of the cause of some particular event, it is often a condition of this sort that we have in mind. In view of the importance of conditions of this sort in our knowledge of and talk about causation, it will be convenient to have a short name for them: let us call such a condition (from the initial letters of the words italicized above), an INUS condition” (Mackie, 1965, p. 245).

We should also keep in mind that, as explained by Rutter, “almost all causal pathways involve several different phases. For example, the pathway to the psychological or psychopathological end point does not begin with a causal risk factor, it must be preceded by the pathway leading to exposure to the risk factor” (2007, p. 378). As a result, the causes that lead to one step may not be the same ones that lead to another. Moreover, the causes that lead to even the same step may also be different if they happen during different developmental periods. For example, the most common way for a 17-year-old girl to get access to beer may be through association with older boys, but the most common way for a 10-year-old girl to get access to beer may be through stealing the beer from her parents. And finally, remember that lovely metaphor of the “conversation” or “dance,” which we liked so well as relational meta-theorists? Well, that metaphor implies that most of the causal effects we find are likely to be reciprocal-- meaning for example, in our illustration in Figure 23.1, that we should expect that, not only does teacher involvement influence student engagement, but student engagement is also likely to reciprocally influence teacher involvement. That does not mean, of course, that we are supposed to manufacture reciprocal causation, but it does mean that when we look for it, we have to look for it in a way that can distinguish between these two directions of effects.

Causality and Causal Beams In his fascinating book on causality, Judea Pearl (2009) distinguishes several shades of causality which are important to consider: 1. Production: capacity of a cause to bring about the effect in situations where both are absent. Requires us to step outside the world momentarily and imagine a new world where the particular cause and effect are absent. We apply the cause and see if the effect sets in. 2. Dependence: an aspect of causation that appeals to the necessity of a cause in maintaining certain effects of in the face of certain conditions that would otherwise negate these effects. 3. Sustenance: enriches the notion of dependence with features of production while remaining in a world where both cause and effect are true. The cause alone would be sufficient for maintaining the effect no matter what the other circumstances. The cause takes responsibility for sustaining the effect under such adverse conditions. 4. Causal beam: Set of sustaining parent variables that are effective under each set of conditions. 5. Natural beam: Sustaining set of parent variables that are effective when we freeze all the variables outside the sustaining set at their actual values. 6. Actual cause: Exists as a natural beam. 7. Contributing cause: Causal beam but no natural beam. (p. 316- 319)

So how do we deal with these issues?

Let’s take them one at a time, starting with the issue of “pre-existing conditions” or the issue that people may have been different even before they encountered the potential causal factor. Let’s just take an obvious example. We discover that people who go to hospitals are much more likely to die than people who stay home, and so we wonder whether spending time in hospitals can be fatal. Before we get carried away, of course, someone will immediately point out that the reason that people go to hospitals is that they are very sick or injured, so hospitalized people are very different from non-hospitalized people before they ever even arrive at the hospital, and it is this pre-existing condition that is more likely to cause death than the hospital stay—both the hospital visit and death are likely to be the effects of pre-existing illness or accident.

In fact, that is one of the mains reason we are grateful to experimental methods-- because random assignment helps us with this problem. As we mentioned previously, the ideal test of causality requires a time machine, or a “reality bifurcation device” in which we could split reality in two and send the exact same participants on two different trains simultaneously, one in which they encounter the potential causal factor and one in which they do not. As explained by Rutter (2007), “All causal reasoning requires an implicit comparison of what actually happened when an individual experienced the supposed causal influence with what would have happened if simultaneously they had not had that experience. Even in a controlled experiment, that observation can never be made” (p. 378). And, because we can’t do this, we are always stuck with the fact that we have to send different groups of participants on those two trains, and that’s why we appreciate random assignment, because it means that our groups are as similar as we can possibly make them before they go, and we can even estimate (and, if needed, adjust for) that similarity by comparing treatment and control groups on their pretest measures before they board their trains (i.e., before they are exposed to the potential causal factor).

But in naturalistic designs, we only get to watch our participants take their different trips, and since we know that different people prefer different vacations, we are forever fretting about whether they were already different to begin with before they even encountered the cause (or did not encounter the cause). Because if they were different (and we have every reason to believe that they would be), it could be that it is those pre-existing differences that are creating the differences we detect between the two groups that took the different trains and not the effects of the different trains (the cause) at all.

Any third variable could represent a selection effect, that is, the reason that the person took that particular train, and so provide an alternative possible explanation for the observed covariation between the potential cause and the outcome (besides a causal connection between the potential cause and outcome). For example, if we are interested in determining whether American students who spend their junior year abroad in England versus in Zimbabwe come back with a broader worldview, we can look at differences in their worldviews when they return. Say that we find one: the students who stayed in Zimbabwe have broader worldviews than the students who spent time in England. It is plausible that their different experiences may have produced the differences in their worldviews. But it is just as likely that students with a broader worldview would be more likely to choose Zimbabwe over England, and it could be those preexisting differences selected them into the different experiences. So the students were already different in their worldviews before they even embarked on their different experiences, and it is these pre-existing differences that we detect upon their return, and not any causal effects of their different experiences while they were away.

Well, why not just look at them before and after they went, and see how they changed?

Yes! You have stumbled across one of our secret weapons in the fight to document causal processes, and that is time. If we include time in the design of our naturalistic studies, we have several advantages.