Statistics, I love you, but you are bringing me down

After giving this issue some thought, I have come to the conclusion that statistics are currently serving the unfortunate purpose of preventing psychology from developing into a mature science. I will not go so far as to claim that psychological research is unscientific, just under developed when compared to its older siblings biology, chemistry, and physics.

I believe that science has two primary epistemological modes: observation and theory. Observation involves simply watching. There is a precise and accurate way to watch, and this is what science tries to do. Due to cognitive biases, however, people tend to see patterns in what they watch that may not represent actual differences in nature. Thus, we have developed what some people call "social control" (see: Dr. Laura Little) so that our tendencies to see what is not there may be kept in check. This social control exists in the form of the null hypothesis significance test. This way, differences are held to a commonly agreed upon standard and if the differences appear likely to be "real" (i.e., not due to chance) then we may state they we have observed a phenomenon in nature.

Theory involves trying to understand and predict the phenomenon that we have observed. It goes beyond simply seeing if something is really there, and it attempts to look at the underlying causal mechanisms that explain or precipitate the phenomenon we are interested in. If statistics is the language of knowing whether something is actually there or not, then mathematics is the language of theory. This means that, after conducting null hypothesis significance tests, we should create parsimonious math models that explain and predict phenomena. This is not completely unheard of in psychology: it is often done in the fields of perceptual psychology, and math modeling has been used to accurately predict divorce in a social-psychological context (see: Dr. John Gottman). Yet, this type of analysis is sorely lacking in psychology as a whole.

Why is this? I believe it is because as psychologists, we have become too attached to statistics. We have worked hard (and brilliantly) to use the most advances methods to detect differences in populations, but we are highly unaware of the fact that other sciences use math modeling to create powerful theories on top of and in conjunction with statistics. Although statistics are crucial to observation, scientists should be bilingual: they must speak a language that accurately distinguishes between spurious and real phenomena (i.e., statistics) and they must speak one that is descriptive and prescriptive of both current and future events (i.e., mathematics). Psychologists need to learn to speak this second language in order to advance their science from nascence into maturity.

Psych Stories: The Coolidge Effect

As the story goes...

...President Calvin Coolidge and his wife were being toured around a government farm when Mrs. Coolidge inquired about the ratio of roosters to hens in the hen house. "Surely, one rooster could not mate with all these hens," she asked. "Indeed, it mates dozens of times each day, day after day." A smirk on her face, Mrs. Coolidge suggested that her guide impart this knowledge to Mr. Coolidge. Upon receiving this message, President Coolidge examined the hen house for a moment before responding, "This rooster, does it mate with the same hen each time?" When the guide responded, "No, sir. A different hen each time," the President asked his guide to tell that to his wife.

Based on this story, the name "Coolidge Effect" has been applied to the phenomenon, observable in nearly all mammalian species, of both sex's boost in sexual performance when introduced to new partners. This frenzy of copulation is thought to be explained by a rush of the neurochemical dopamine, which controls motivation and reward in the limbic system.