jim.shamlin.com

14 - Mill's Methods

The British philosopher John Stuart Mill developed a set of methods that can be applied to an observed effect to speculate its causes. Prerequisite to all of his methods is a basic procedure:

  1. Identify a set of possible causes, events that occurred prior to the target event, and which are suspected to have been the cause of the target event
  2. Collect information about a range of situations involving the possible causes
  3. Use one of the five methods to infer the probable cause of the effect

The Method Of Agreement

The method of agreement finds two or more situations in which an effect occurred, and seeks a common factor among them. That is, if something occurred when ABC preceded it, when CDE preceded it, and when BCD preceded it, the common factor of "C" is a likely cause because it is common to all situations where the effect occurred.

The Method Of Difference

The difference method compiles situations where some of the events occurred - which sometimes resulted in the effect and sometimes did not. In this instance, the factor that was present when the effect occurred and absent when it did not is a probable cause.

To return to alphabet soup: if ABC did not cause the effect but BCD did, then "D" is the probable cause because it was present when the effect occurred and absent when it did not.

This factors heavily into the scientific method, in which a control group is used in which a factor is removed, and the failure for the effect to materialize is taken as proof that the absence of the factor is evidence to its influence on a test group in which it was present. The difficulty here is makign sure that all other conditions are identical.

The Joint Method

This method is used when there are multiple causes and multiple effects, to find correlation between one of the causes and one of the effects. It gets a little odd:

The conclusion is that A causes X is based on its agreement with the first two cases (A appears every time X appears) as well as its difference with the third (the absence of A results in the absence of X)

The joint method is applied when neither the agreement nor disagreement methods will work, In the example above, the first two cases suggest that either A or C might create X, the second and third suggest that either A or E might cause the absence of X, and the first and third suggest that either B or C might cause the absence of X.

The Method Of Residues

Where some factors can be identified as the causes of certain effects, any effect that is produced by the addition of other factors is attributable to those other factors.

When then conclude that "A creates Z" though we do not yet have proof that it is a sufficient cause or merely a necessary one on its own.

The Method Of Concomitant Variations

Concomitant variations test whether a factor is solely responsible for an outcome, or whether two factors must be present together to create a phenomenon

The conclusion here is that factor A causes X only when B or C (but not both) are also present. We then conclude that A is a necessary cause of X, but not a sufficient cause.

Limitations Of Mills Methods

Mills methods are highly popular in science, particularly chemistry, where theories are tested in a tightly controlled environment such that the inputs are known. The methods do not fare as well outside of the laboratory environment.

Mills methods are based on the assumption that one of the candidate causes is the true cause. It may not be at all, and all experimentation can lead to failure Te methods are useful in making the final cut when you have a good idea of causality, but do not help to discover the candidates.

The methods are also based on the assumption that there can be only one cause, and but for concomitant variation, can lead to failure when there is interaction among causes to an effect.

The methods are also based on the assumption that the factors that do not cause an effect are neutral, and do not prevent the effect from occurring.

The methods are also based on the assumption that a cause will always result in an effect, ignoring probability. Sometimes, a rock might bounce off a window rather than break it, which would cause it to be disqualified as a cause.

In spite of the limitations of the methods, they remain an important tool for critical reasoning, particularly for causal investigations.