Triumph of Reason: Statistics
  by Groff Schroeder

Benjamin Disraeli wrote, "There are three kinds of lies - lies, damned lies and statistics." While lies and damned lies obviously are lies, statistical measures such as grade point averages, homicide numbers and infant mortality rates successfully represent reality and enable architecture, engineering and medicine. The results of mathematical calculations do not change, so statistical analysis of a data set always provides the same answer - unless you create (falsify) the value(s) you want, or make a mistake during the calculation.

Statistics is a branch of applied mathematics capable of inferring a system's future behavior through probabilistic analysis of past data (measurements) of the system. Statistical inferences based on large, representative data sets and mathematically rigorous probability calculations are extraordinarily successful in the analysis of reality. Statistical inferences can detect election fraud; they also allow the discrimination of association and causality, verifying mathematically if one thing is merely associated with another thing or is actually caused by it. Without statistics, modern technology would not exist.

Hundreds of years of statistical analysis provide extremely accurate and precise information, upon which the research and development of almost any product is critically dependent. Statistically derived facts of nature stemming from repeated analysis of physical properties of materials (melting points, configurations, conformation, etc.), characteristics of structures (statics and dynamics), energy exchange (thermodynamics), charge flow (electronics) and behaviors of systems (cybernetics) allow engineers to correctly predict the future behavior of their designs. However, since the standard deviation is a key concept of statistics, exact predictions are unlikely. Instead, statistics quantifies the observer's level of uncertainty about a system, allowing informed action based upon known probabilities.

Statistical relationships help insurance underwriters calculate premiums based upon risk factors like age and smoking - making profits possible despite regularly writing huge checks. Sports fans rarely reject batting averages and other statistical records as lies, even though steroids or other mechanisms of cheating might make them incorrect. We might interpret a weather report predicting a 20% chance of rain as incorrect if it does not rain at our location - while those in the 20% of the forecast region receiving precipitation see the forecast as correct.

Statistics supporting what we want to believe are easier to accept than those that do not. However, a major value of statistics is their ability to repeatedly verify what we do not want to hear.

When purposefully falsified or plucked from the air, statistics actually are lies. However, statistical inferences correctly calculated from valid data provide stunningly accurate, precise and repeatable tools for the evaluation of the past, present and future. Despite their awe-inspiring power as a predictive and retrospective tool, many people do not understand statistics (or even mathematics in general), making it easy for some to misrepresent the meaning of statistical inferences - even after the correct application of probabilistic relationship to valid data. Here, the statistics are not lies, but the conclusions drawn from them are. Because people rarely question even easily verifiable claims, outrageous misrepresentations of statistics can be wildly successful, as long as key information remains inaccessible.


While opportunity for misrepresentation and abuse always exists, without statistics none of our technology would work. Statistics offers an extremely powerful but easily manipulated tool for understanding nature, so skepticism and critical thinking play an important role in assessing their validity.

 

Send us an Email

Or write to us at:
Freethinkers of Colorado Springs
P.O. Box 62946
Colorado Springs, CO 80962-2946
Phone: 719-594-4506