Science seeks to show probability of relationships

In response to the letter in which Gregg Calkins disputes A.M. Costa Rica’s claim that proving causation in science is difficult, I’m afraid that Mr. Calkins is mistaken.

Actually, science never proves anything. The scientific method seeks to disprove the null hypothesis, which asserts that there is no relationship between hypothesized variables. Insofar as experiments or other investigations can disprove the null hypothesis, scientists (and the public) can be increasingly certain that the hypothesized relationship exists between the variables. However, that certainty is never 100 percent, and logically can’t be, since no one can ever be sure that the results won’t be different tomorrow. In practice, science has adopted statistical conventions for reporting findings. These get complicated fast, but they basically calculate the likelihood that a discovered relationship between variables did not occur by chance.

However, the statistics are almost never 100 percent, and, in fact, even statistically significant relationships between variables usually only explain only a small part of their association.

With respect to cause, things get even more complicated.  Most scientific findings are simply correlations — two variables have some statistically significant relationship with one another — but it’s anyone’s guess which caused which or whether a third unknown variable is causing both. In general, the only way to get a handle on cause (and it is only a handle, since science remains in the business of disproving the null hypothesis) is with controlled laboratory experiments. These are necessary in order to eliminate as many of the other variables as possible in order to see whether the hypothesized causal variable is really doing the causing.

Enter the obvious difficulty with demonstrating cause in the case of the global warming thesis: You can’t put the globe in a laboratory. As a result, almost all scientific evidence (some questions can of course be replicated in laboratory settings) comes from the messier real world where there are thousands of uncontrolled variables, any of which could be causal. Mr. Calkins provides an excellent example of this scientific difficulty by noting (I assume correctly) that the earth has actually cooled rather than warmed over the last decade. The problem here is that temperatures are going to fluctuate, some simply by chance.

Another problem, noted in the A.M. Costa Rica column, is that there are long-term trends operating that cool or warm the earth over centuries. To find out whether global warming is or isn’t occurring as a result of humanity’s (poor?) stewardship of the earth requires somehow teasing both random and long-term trends apart from those that can be plausibly attributed to human stewardship.

I don’t have the faintest idea where the bulk of the evidence lies on this crucial question, so won’t offer an opinion. However, I believe the discussion of this issue would be more fruitful if all concerned has a better understanding of the scientific enterprise that should ideally inform it.

Ken Morris
San Pedro

This entry was posted in Reader Opinion. Bookmark the permalink.