Distinguished Educator Award winner Lynette Hudiburgh cautions how statistics in the media can be misleading

Written by Caroline Igo, CAS communications intern

Lynette Hudiburgh (R) receives the CAS Distinguished Educator Award from Senior Associate Dean Renée Baernstein.

Lynette Hudiburgh, a prestigious Miami statistics lecturer, is the recipient of the Distinguished Educator Award from the College of Arts and Science. She gave her lecture, "Statistics: Confused, Misused, and Abused, or How to Avoid Being Swept Away in the Data Deluge" on Tuesday, September 17 in front of a full Upham Hall classroom.

Hudiburgh has always had a thing for numbers. She received her undergraduate degree from Adams State University, and then moved onto achieve a master's in statistics from the University of Northern Colorado. Since then, Miami has been fortunate enough to be a home for Hudiburgh these past 9 years, where she teaches various introduction to statistics courses.

"You would not believe the reactions I get when I tell people that I teach statistics, Hudiburgh joked at the start of her lecture. "While many people may see statistics as boring and dry, statistical literacy is very important."

What is Statistical Literacy?

Hudiburgh framed her talk in terms of the concept of statistical literacy, which is the way in which we understand and interpret data in everyday life. Especially in the digital age, a plethora of data is readily available at our fingertips.

Hudiburgh referenced a study that stated that 90% of the world's data was created in only the last two years. "To put that in perspective, that means there are more bits of data on earth than there are stars in the universe," she said.

To avoid being swept away by this overwhelming amount of data, Hudiburgh stressed that we must become familiar with statistical analysis.

"While most of us aren't statistical scientists, we are all capable of learning how to read data," Hudiburgh said. Quoting mathematician Samuel S. Wilks, she added, "Statistical thinking will one day be as necessary for efficient citizenship as the ability to read and write!"

Becoming More Statistically Literate

Everyone should be able to read and interpret statistics in the media, Hudiburgh explained, especially in today's age when there is so much weight placed on the facts. It is nearly impossible to find an informative article without a graph, chart, or data.

"But, without the knowledge of how to read that data, we could be misguided," she said.

Hudiburgh pulled up a recent study on the board. It was a poll about the percentage of Americans that agree with Trump's presidency at this point in his career. A higher percent said 'YES' rather than 'NO,' so at first glance, one would come to the conclusion that most Americans agree with President Trump.

However, Hudiburgh pointed out the fine print on the bottom: "among Republicans."

Understanding Bias in Statistics

Lynette Hudiburgh shows how graphical representations of data can be misleading.

The TV station that reported the poll was biased, as they 'cherry-picked' the data, only using part of the data which supported the point they were trying to make. In this case, they made it seem that President Trump is more popular than he actually is among Americans.

"Bias is a part of everyday life, but we have to work hard to minimize bias in statistics by utilizing proper sampling techniques," Hudiburgh explained.

She showed the audience another example: a 1936 poll from The Literary Digest. Its goal was to predict the winner of the '36 presidential election, but it was very wrong. The large sample size of more than 2.3 million voters that was used included mostly wealthy, white men. The voters were not randomly selected; therefore the results were biased.

"It doesn't matter how large a sample may be," Hudiburgh explained. "If bias is present, the results will be faulty."

Because data can be biased, it can also be misleading. Even if a chart or a graph is from a good source, that does not necessarily mean it was constructed correctly.

"Charts and graphs can mislead readers because of one or more of the following: an omitted baseline, manipulated axis, cherry-picked data, wrong use of a graph, or broken conventions," Hudiburgh said.

She concluded her argument with more examples from the media, such as a pie chart that was larger than 100% and a line graph that had stretched its axis too far.

Hudiburgh told her audience that in order to avoid being swept away, one must ask themselves if the data in front of them tells the whole story. Where was the data collected from? How was the data collected? Why was the data collected? Are there any confounding variables presents?

"Use statistical literacy skills to analyze data and charts in the media," she said. "Don't be misled!"