“There are three kinds of lies: lies, damned lies, and statistics” – Mark Twain and Benjamin Disraeli
In order to read and comprehend scientific research, you must have a basic understanding of statistics. If you have never taken a collegiate level statistics course, here are links to two excellent tutorials to get you started:
The data that a researcher gathers typically represents a small sample of a given group, but by using appropriate statistical techniques, the researcher is able to estimate if and how the data relates to the larger population.
A few key statistical terms that you will see in every study you read are as follows:
Sample Size: The number of a group being studied. Researchers typically want to find a sample size large enough so that they can draw statistically significant conclusions from their data.
Confidence Interval: A confidence interval is an estimated range of values representing the sample data that likely contains the actual value found in the broader population. It is expressed in terms of a confidence level (typically 95% or 99%) which suggests that the true value would be found within the confidence interval 95% or 99% of the time.
P Value: An expression of the confidence level. p=.05 would be a 95% confidence level while p=.01 would be a 99% confidence level.
Statistically Significant: Statistically significant does not refer the the importance or gravity of a particular conclusion, but rather it implies whether or not a conclusion could have been a result of chance.
Correlation and Causation: Correlation is defined as a statistical relationship between two random variables while causation means that one thing causes another. For example, it is true that the presence of a fire truck might be highly correlated with house fires; however we know that fire trucks do not cause house fires. As obvious as that example is, it is easy to read a research report and leap to a conclusion that one thing causes something else to happen. As you read, be mindful of phrases like “linked to,” “a relationship between,” or “associated with.” They imply correlation and not causation.
Blind: Not given complete information to a group involved in a study, as the knowledge from that information might somehow bias behavior.
Cohort: A group of people who share a common characteristic.
Control: A group that does not receive the experimental treatment. The results from this group are compared against the results from the experimental group that does receive the treatment.
Double-blind: A valuable research study protocol where neither the researchers nor the subjects know whether they belong to the experimental group or the control group.
Placebo: A medically ineffective treatment which in some cases is known to provide a perceived effect by the person receiving the treatment. This is called the “placebo effect.” In some experiments, a placebo is given to the control group in order to show the true efficacy of the treatment being studied.
Randomized Trials: A study in which participants are arbitrarily or randomly assigned to be a part of the group getting treatment or part of the “control group.” A control group may get a placebo, a different dosage of the treatment, or another appropriate treatment that is well understood.
“Science is organized common sense where many a beautiful theory is killed by an ugly fact.” -Thomas Huxley
Additional Information about Correlation and Causation
If you are struggling with the concepts of correlation and causation, the follow two references may be helpful:
- The Dihydrogen Monoxide Research Division which looks at the “risks” of Dihydrogen Monoxide, otherwise known as water.
- The Smell of Bakery Bread May be a Health Hazard, a humorous look at how bread might be a menace to society.