#spurious relationship

[neu2neuro-blog1]

Do "average" students make better physicians regardless to challenge?

Science succeeds or fails in its efforts to produce valuable technology and policy based upon the quality of its theories. When theory is incorrectly formulated or hypotheses are incorrectly deduced from theory, then technology and social policy fail. Unfortunately, it is very easy to build incorrect theory or incorrectly deduce from theory. And all too often, mistakes in theory induction or deduction are not revealed until people experience the negative consequences of flawed technology or social policy.

Mistakes in theory induction and deduction can be summarized as two types of false relationships between variables: spurious relationships and suppressor relationships.

Spurious Relationships

Spurious relationships are false indications of causality. They occur when theory omits an extraneous variable that affects two other variables that have no causal connection. The effects of the extraneous variable on the two other variables leads one to conclude, in error, that the two other variables are causally linked. This silly example illustrates a spurious relationship. City Manager Sapp, of Sappville, has collected data on the past 100 fires to occur in the city. He has information on the number of fire trucks dispatched to each fire and the amount of damage (in dollars) caused by these fires. Mr. Sapp notices a very strong relationship between the number of fire trucks dispatched and the amount of fire damage. He concludes, logically, from the data at hand, that "if we stop sending fire trucks to fires we will greatly reduce fire damage in Sappville." What Mr. Sapp did not realize from his limited data was that a third variable--size of fire--caused variation in both of the events he recorded: number of fire trucks dispatched and amount of fire damage.

Suppressor Relationships

Suppressor relationships are false indications of no causality. In this case, the extraneous variable has a positive relationship with one variable and a negative relationship with another variable. Although these two other variables are causally connected, it seems like there is no relationship between them because the extraneous variable "cancels out" the correlation between them. Consider this theoretical proposition: The greater the academic performance, the greater the job performance. For challenging jobs, there is a causal connection between academic performance and job performance. For routine jobs or those involving much repetition, however, there is little correlation between academic performance and job performance. The extraneous variable is boredom with the job. For routine or repetitive jobs, persons who excel academically have poor job performance because they tend to get bored.

Courtesy of: http://www.soc.iastate.edu/sapp/soc415SSRelationships.html

[lying-with-statistics]

Trying to predict the likelihood that a professional athlete will become bankrupt at some point in the future is a dodgy effort at best. Most of the factors that one would actually want to take into account probably wouldn’t have much to do with their on-court or on-field performance. Yet, a labor economist has managed to find a correlation between a basketball player’s 3-point shooting frequency and their bankruptcy rates. And, he’s concocted an entire explanation (basically out of thin air, as there is no existing theory) to justify why this isn’t simply a spurious correlation. The problem is that there are a lot of more reasonable predictors of bankruptcy that he doesn’t look at. Arguing that taking 3-point shots is the best measure of risk acceptant behavior is problematic in and of itself (by this measure, someone like Dennis Rodman is a very risk averse person— does this make any sense?). This study is the double whammy of bad research— in that it ignores potentially good measures of player behavior (on and off the field) and other potential indicators of financial stability like pay rates, number of years in the league, etc., and instead emphasizes a lousy measure. Henry has to essentially make up a reason why looking at 3-point frequency is legit in the first place for this study to be an exercise in anything but spurious correlation. Generally, the more tortured the explanation, the more unlikely it is (this is known as Occam’s Razor— the simplest explanation is most often the correct one).

[notesontheory]

In statistics, a spurious relationship (or, sometimes, spurious correlation or spurious regression) is a mathematical relationship in which two events or variables have no direct causal connection, yet it may be wrongly inferred that they do, due to either coincidence or the presence of a certain third, unseen factor (referred to as a "confounding factor" or "lurking variable"). Suppose there is found to be a correlation between A and B. Aside from coincidence, there are three possible relationships: