Causation correlates

You can put almost any two variables on a graph and visually demonstrate a correlation. One very tongue in cheek chart that is making the rounds this week is the impact of the declining prevalence of Microsoft Explorer in the browser wars and the correlation with murder rates in the US.

From Gizmodo

Given the experience most of us had with IE in the past, there may actually be some truth to this.

Most would recognize that this relationship is as useful as the correlation between pipe smoking and the use of microwave ovens.

One goes up, ones goes down. A definition connection, but which causes which? Or merely coincidence.

From Quarks to Quasars has an article about how complicated teasing causation out of correlation can be.

In Latin, the phrase generally used is “cum hoc ergo propter hoc” which translates more literally to “With this therefore because of this.” The opposite, however, is true. Causation proves correlation, but not the other way around.

When you have two (or more) data points that ‘line up just right’ (or, that correlate with one another) there are one of three ways of looking at the data. Either:

  1. A causes B
  2. B causes A
  3. Both A and B are consequences of a common cause
  4. Some combination of 1, 2, and 3 may be in place (which normally describes a self-reinforcing system, such as the predator-prey system, where one intrinsically affects the other and vice versa.
  5. A and B aren’t related and the correlation is a coincidence.

Some criticize the cum hoc ergo propter hoc fallacy by claiming that while correlations does not prove causation, but it is most certainly a hint. As the charts above show, it really doesn’t necessarily provide anything useful all, and it requires a combination of logical thought, data gathering, and analysis to provide answers. It is also important to note that over time, further  data may either verify or disprove the original assumption.

For example, a study conducted by the University of Pennsylvania’s Medical Center in 1999 concluded that:

Young children who sleep with the light on are much more likely to develop myopia later in life (myopia is near-sightedness). Therefore, sleeping with the light on causes myopia.

However, upon further analysis, the actual cause was exactly the opposite, parental myopia led to nightlight use.

Later, thanks to the peer review process, the Ohio State University disproved this claim and showed that myopia wasn’t caused by sleeping with the light on. Instead, OSU demonstrated a strong link between parental myopia and the development of child myopia, the study also noted that nearsighted parents where more likely to leave their child’s light on at night.

It is this cum hoc ergo propter hoc fallacy that is primarily behind the studies that are used to promote alternate medicine. Most people who have studied these treatments carefully have realized that the most common reasons people seek treatment at alternative practitioners are for conditions that have symptoms that vary over time, or are prone to heal themselves over time. Sufferers of recalcitrant pain, for example, are aware that their degree of pain is not constant. It waxes and wanes over time. A visit to a chiropractor or acupuncturist may precede a relief from symptoms that would have happened without the treatments.

This problem is not unique to alternative medical treatments, but can be found in a number of scientific disciplines. Re-examining data, cautious reproduction, and funding and publishing those results are important in verifying original results. However, this is not happening for a number of reasons. In October of this year, The Economist ran an article outlining some of these reasons and emphasized the importance of rectifying the problems.

Careful examination of hypotheses is critical in establishing relationships between two or more events, and a critical examination of preconceived ideas and biases in research. Scientific research is ultimately self-correcting, but those corrections can be dangerously delayed without the proper controls.

This entry was posted in Science and tagged , . Bookmark the permalink.

One Response to Causation correlates

  1. Wesley E. Long says:

    Post hoc ergo propter hoc (after this, therefore because of this). This is the fallacy of assuming that A caused B simply because A happened prior to B. A favorite example: “Most rapists read pornography when they were teenagers; obviously, pornography causes violence toward women.” The conclusion is invalid, because there can be a correlation between two phenomena without one causing the other. Often, this is because both phenomena may be linked to the same cause. In the example given, it is possible that some psychological factor — say, a frustrated sex drive — might cause both a tendency toward sexual violence and a desire for pornographic material, in which case the pornography would not be the true cause of the violence.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s