The media often mis-represents correlation research as factual, proven information. Think of the “breaking news” articles frequently found on social media that try to convince you that chocolate will make you live longer or that eating fish will make your teen nicer. Many of these articles are based off of the results of correlation research, not experiments. However, correlation does not prove causation and the only way to prove a cause and effect relationship is do an experiment that manipulates the variables. [The dictionary defines a variable as a factor that is liable to change.] Independent variables are manipulated based on the needs of the study while the dependent variable is the element that is expected to be altered by the independent variable.
Correlational research involves measuring the relationship between two or more relevant variables. This method is often paired with a scatter plot to create a visual image of the relationships. The points can show an ascending or descending line to represent positive linear relationships or negative linear relationships respectfully. When there is essentially no relationship between variables the points create a nonlinear pattern or a collection of random dots. The most common method for measuring the relationship between variables is the Pearson correlation coefficient. With this method an equation is used to measure the relationship in a range from -1.00 to +1.00. The strength of the relationship is determined by the distance of correlation coefficient from zero. For example, -.45 is a stronger relationship than .30 .
The most important thing to keep in mind with correlation studies is that they can not be used to draw complete conclusions about the relationships between variables. As mentioned, only an experiment can determine true relationships. This is because correlation research does not take into account outside variables. Correlation research is used to determine if further studies should be held to research possible relationships. It provides a lead for r
esearchers to start but experiments bring the results to the finish line.
“Eat Sweets, Live Longer” by sciencenews discusses a correlation study done by researchers at the Harvard School of Public Health in Boston. I-Min Lee and Ralph S. Paffenbarger Jr. collected data from 7,800 Harvard alumni men. The independent variable of this study was the amount of candy eaten and the dependent variable was the life span of the participants. The study found that “consumption of candy was associated with greater longevity.” The study did take other possibilities into consideration such as; age, drinking or smoking habits, diet, exercise and weight. Although the researchers did find a correlation between candy and a longer life, there was far from enough information gathered to prove any definitive connection. Had you read the media coverage of this study without knowing how identify a correlation study, you would think scientists are telling you to rush out and buy the biggest Hershey bar you could find. Despite the media’s positive conclusion that over indulgence in chocolate will add years to your life, Lee hopes the research will “stimulate more research in the field so that both laboratory scientists as well as scientists conducting studies in human populations can develop a more coherent picture" and not just more cavities.
Experiments are the most reliable method for determining true relationships between variables. There is a wide variety of options on how to conduct experiments.
An example of a well conducted, however cringe worthy, social experiment is described by preventdisease.com. An article titled “Wash Your Hands: Signs Only Work for Women” details an experiment to see the effectiveness of hand washing signs. In this particular experiment the hand washing signs are the independent variable while the number of people who wash their hands is the dependent variable. The experiment placed a bystander in both a womens’ and men’s restroom to gather information on how many people wash their hands after using the toilet when the “wash your hands” sign is absent and when it is present. The study found that women are over-all are more likely to wash their hands than men and are more effected by the “wash your hands” signage. The number of men who washed their hands stayed relatively the same before and after the sign was displayed. When the sign was in view in the womens restroom, the number of women who ‘rinsed’ their hands went down and the number of women who washed their hands with soap and water went up. Researchers admitted the internal validity of the study could be flawed because of the presence of a bystander in the bathroom. This means, many of the restroom users who only rinsed their hands may have not washed their hands at all had they not known someone else was in the room.
Unlike a correlation study, the results of this experiment held more validity. The media outlet preventdisease.com covered this particular experiment well by providing all the information to the public including the variables that may have affected the accuracy of the results.
Correlational research has both strengths and limitations but when authentic results need to be found, experimentation is the best method. The media may not always offer the most honest representation of correlation studies. The media has a tendency to lead the reader to believe the findings of correlation studies are as valid as those found with experimentation. It is up to the reader to understand the different between correlation and experimentation so that they can better determine the justifiability of a study results.
Comments