Charts, graphs and tables appear in newspapers, pop up on websites and fill the margins of textbook pages. Statistics are the basis of discoveries and arguments presented on the evening news and morning radio shows. But what do the numbers really mean, and how reliable are they? On Monday, Oct. 1, Rebecca Goldin, associate professor of George Mason University’s Department of Mathematical Sciences, came to campus to speak to St. Olaf students about exactly that.
Passionate about statistics, Goldin has established credibility through her experience in the field of mathematics. In addition to teaching, Goldin is also the director of research at Statistical Assessment Service STATS. The mission of this non-profit organization is to “improve the public debate on how science, quantitative and mathematical issues are covered in the media.”
Goldin’s well-attended presentation described how to interpret statistics presented in the media. Using many colorful examples, she illustrated the ways data can be inaccurate, skewed or even invalid. She also spoke of the relevance and integrity statistics gain when they are viewed in an accurate context. Golden gave several suggestions of points to keep in mind when encountering statistics in the media.
She encouraged students to begin by thinking about where the data came from. Looking at the location in which a survey was distributed can provide clues as to who the data actually came from.
“If a survey was given at a college, it would most likely have different results than if it was given at a shopping mall,” Goldin said. The economic, social and political backgrounds of subjects are variables in the results. Pondering how these factors influence the findings of an experiment makes the reader more aware of their actual meaning.
Consider how the data is being revealed. Goldin explained, “In many instances, people reporting a set of statistics can present them in a way that proves whatever bias they may have.” It is important to look at the question that was asked by the researchers and consider if it is leading in any way. The question may be presented in a way that encourages specific responses, as it is extremely difficult to conduct a completely unbiased survey.
“There will never be enough data to control for all of the variables you should control for,” she clarified.
Be leery of studies with conflicting results. If one researcher discovers a connection that clashes with a previously accepted truth, it does not mean all findings up to that point should be discredited. Goldin repeated, “Remember that correlation does not mean causation.”
This can be difficult to remember while looking at the morning’s headlines.
“The media presents information in a way that gives equal time to both sides, but the reality is that usually most researchers are in consensus,” she pointed out. The people who are most knowledgeable about a specific area usually agree. Conflicting views should encourage further research rather than provoke an instant change in views.
When reflecting on published statistics, it is important to note that a group study is precisely what it sounds like, the study of a group.
“When it comes to social experiments, some of what you get is experiments that tell you about large groups of people and not just one person,” Goldin cautioned.
“We are all individuals. Groups may have large patterns, but statistics will never be fully predictive of individual behavior and there is no way we can get past that aspect of it.” Statistics work splendidly as evidence for general trend analyses, but applying the same information to an individual will rarely be accurate.
Golding is working to show how mathematics is applicable outside of the classroom. Statistical comprehension and analysis has become an integral part of understanding the world. With the election coming up, a dialogue about statistics and their implications is extremely relevant. Try to view polls and predictions critically. Keeping Goldin’s advice in mind can help voters understand the information put before them, and interpret it using an educated approach.
As scholars gain a deeper understanding about the way studies are conducted, they are becoming more aware of the limits on quantifiable research. Modern technology allows information to be collected more precisely and at a larger scale than ever before. It also provides people with the resources to understand data in more accurate ways.
As Goldin put it, “There are going to be some very big changes on the way we look at it, but there will always be a role for standard statistical inference.”
