¶ … correlation and regression using an example from public health to illustrate.

"Regression analysis is a statistical tool for the investigation of relationships between variables. Usually, the investigator seeks to ascertain the causal effect of one variable upon another," such as the effect upon a student population's BMI of stocking vending machines with healthier food vs. allowing the vending machines to offer 'junk food' (Sykes 2010). Will the presence of healthier food in vending machines result in a lower average BMI in the population or will it not (because children are bringing junk food from home or other sources)? Correlation merely suggests that two factors tend to occur together -- for example, schools with a higher average student BMI may tend to have more vending machines on-premises. But these schools may also be poorer, and have less healthy cafeteria food, so the vending machines may not be the causative factor in obesity.

Interpret the following statement: Correlation is not causation. Again, provide an example from public health to illustrate

Correlation is not causation because two common traits may be associated, but one trait does not 'cause' the other trait. For example, children whose last names end with a vowel may have a statistically higher chance of manifesting type II diabetes. The reasons for this might be the poverty and discrimination faced by many Latinos, the letters of whose names more often end with a vowel than other ethnicities. Poverty often leads to a poorer diet, less access to places to exercise safely, less access to healthcare, and a lack of opportunities to become educated about how to live a healthy lifestyle. It is poverty that is the causative factor to the child's ill health, however, not the child's name, and changing the child's name will not affect the outbreak of type II diabetes in the Latino community.

Reference

Sykes, Alan O. An introduction to regression analysis. University of Chicago Law School.

Retrieved February 8, 2010 at http://www.law.uchicago.edu/files/files/20.Sykes_.Regression.pdf