Causation When C causes E, C helps to make E happen. Learning about - PowerPoint PPT Presentation

9/26/2016 Causation When C causes E, C helps to make E happen. Learning about causes allows us… Reasoning about … to predict what effects will occur Causation … to bring about desirable outcomes … to prevent undesirable outcomes Science often seeks evidence of what causes what. Justin C. Fisher Pseudo-science often makes bad causal claims. SMU Dept. of Philosophy Which causal claims should we believe? The Cause of E versus A Cause of E Particular Events vs Types of Events Most events had many causes working together. We can often say which particular billiard ball caused Q: What caused the fire? another to move. A: That the match was struck, that the match-tip contained certain chemicals, that the room contained oxygen, etc… We can’t yet say which particular cigarettes caused someone’s cancer. But we do have strong evidence that one type of event (smoking) causes another type of event (cancer). 1

9/26/2016 Post Hoc, Ergo Propter Hoc Deterministic Causation versus ( After this, therefore because of this .) Probabilistic Causation Decapitation causes death (virtually 100% of the time) Smoking causes death (it increases the probability of dying sooner) Correlation Causation and Counterfactuals “Trait A is correlated with Trait B in a particular sample” = the percentage of sample members with B is higher among those that have A than among those that lack A. = learning that a sample member has A statistically “C caused E” is roughly equivalent to increases the probability that it has B (and vice versa ). “ if C had differed in certain ways, then E In the strongest possible (aka perfect ) correlation , everything that is an A is a B, and vice versa . would have differed too.” In a very weak correlation , A’s are only a tiny bit more Counterfactuals may be hard to assess for particular events. likely to be B’s than non-A’s are. Our best guide usually is to look at other similar events In a negative correlation , A’s are less likely to be B’s than where the alleged cause differed, and to observe whether non-A’s are (i.e., not-A is positively correlated with B). the alleged effect differed too. (Correlation can also be defined for measurable traits, like height and mass, not just binary traits that you either have or lack.) 2

9/26/2016 Causation = Correlation Causation without Correlation (Correlation is always defined with respect to a sample. Depending on what’s in the sample, correlations can be weird.) Let’s imagine: Eating olive oil causes longer life. Eating garlic causes shorter life. In our sample, the olive oil We’ll consider: eaters also are the garlic eaters. Cases of Causation without Correlation In our sample, eating olive oil causes, but need not be Cases of Correlation without Causation correlated with, longer life. Causation without Correlation Causation without Correlation (Correlation is always defined with respect to a sample. (Correlation is always defined with respect to a sample. Depending on what’s in the sample, correlations can be weird.) Depending on what’s in the sample, correlations can be weird.) Let‘s imagine: Let‘s imagine: Moral #2: If you can’t get (or make) a diverse sample Moral #1: If you want causal relations to show up as Eating olive oil causes Eating olive oil causes without other correlations built into it, then you’ll need correlations in a sample, it’s best to use a sample that is longer life. longer life. to “ control for” other factors. E.g., if we attend just to diverse, without other potential inhibitors being correlated Eating garlic causes Eating garlic causes garlic eaters , and if this sub-sample is large enough, then a with the potential causes we’re considering. shorter life. shorter life. correlation between olive oil and longevity should appear within it (and similarly within the non-garlic-eaters). In our sample, the olive oil In our sample, the olive oil eaters also are the garlic eaters. eaters also are the garlic eaters. In our sample, eating olive In our sample, eating olive oil causes, but need not be oil causes, but need not be correlated with, longer life. correlated with, longer life. 3

9/26/2016 Correlation without Causation (1) Correlation without Causation (2) Some correlations are Some observed between two effects of a common cause. correlations are E.g., a visible storm might pure coincidences. cause pilot to turn on sign, then also cause turbulence. One way to address this concern is to enlarge the Observational studies can attempt to “ control for ” sample to include many instances where C is present, potential common causes (as described earlier). many where C is absent, many where E is present and many where E is absent. In random controlled trials , randomly triggering either effect of a common cause The larger the (random) sample, the less risk a strong will not increase the likelihood of the other effect. correlation would be produced by mere coincidence. How to distinguish Cause from Effect Correlation without Causation (3) Pay close attention to timing : causes precede effects. Use random controlled trials . The effect should occur more often in cases where we randomly trigger the cause than in cases where we don’t. In contrast, the cause should not vary in response to our randomly triggering the effect . 4