Theory & Problems of Statistics

What statistics measure and what statistics mean are two completely different concepts.

Modern societies have become devoted to statistics as a means of proving an argument. Statistical studies appear irrefutable, because they are based in hard facts. However, data alone cannot always lead to a reliable conclusion--especially regarding individuals.

Source Bias

Science may be about finding truth, but studies are still done by human beings, each with their own affiliations and prejudices. When reading statistics, it is important to consider who did the study, who helped fund the study and what they wanted the study to prove.

Data Representation

Any measurement of data is inherently subjective, and will not provide all the context needed for an informed decision. If a survey says that four out of five dentists prefer a certain toothpaste, it omits who the dentists surveyed were, over what brands they preferred that particular toothpaste and why the dentists participated in the survey.

Correlation-Causation Problem

Because one thing is related to another (correlation) does not mean one directly influenced the other (causation). The parable about the rock that wards off tigers demonstrates this problem; just because there are no tigers around does not mean that the rock caused them to stay away.

References

• Cut the Knot: Misuse of Correlation and Causation in Statistics
• Science News: Odds Are, It's Wrong
• WSU.edu: The Problems With Statistics