Unpacking a paradox

Doctoral student Kyle Nickodem finds better ways to understand and communicate data about student achievement

In 2007, Minnesota’s fourth-grade mathematics score on a national exam was fifth highest in the country, a significantly higher score than Texas. But when test scores were grouped by race, the subgroups of children in Texas—white, black, Hispanic, and Asian/Pacific Islander—scored higher than the same subgroups in Minnesota.

Kyle Nickodem wanted to understand how all of the subgroups in one state could get better scores and still fall behind the overall score of another state. He had encountered the phenomenon called Simpson’s Paradox.

Simpson’s Paradox materializes when scores do not accurately represent a state’s education system because variables like race distribution and socioeconomic status are correlated with test performance. Because Minnesota has a bigger proportion of white students, who tend to score higher on standardized tests than minority students, Minnesota’s overall average goes up. Texas, with a more diverse population, scores lower.

“The states that have a more balanced distribution of ethnic groups are being penalized,” says Nickodem.

Nickodem is now a second-year graduate student in quantitative methods in education, part of the Department of Educational Psychology. He and his adviser, associate professor Ernest Davenport, are working to present the numbers in a more accurate way and correct misperceptions in education.

Nickodem and Davenport created the Relative Performance Index (RPI), a tool that corrects for Simpson’s Paradox. They base their research on a standardized test given to U.S. fourth- and eighth-graders every two years called the National Assessment of Educational Progress (NAEP).

The RPI produces a single score that provides a more accurate depiction of how a state is doing. The score is based on how a subgroup compares to the national average of its respective subgroup before it is rescaled back to the NAEP numbers.

For example, in one year’s original NAEP data, Georgia ranked 24 and Pennsylvania 10. After correcting for Simpson’s Paradox using the RPI, Georgia jumped up to 12 while Pennsylvania fell to 14.

“Pennsylvania isn’t necessarily doing anything wrong,” says Nickodem. “The Relative Performance Index is a way of showing who is actually educating all of their students, as compared to just the high performing students.”

Nickodem is interested in the role that standardized assessments play in education and how they influence the decisions that people make. He has seen the effects of assessments on many levels, from work with kids in schools and communities to the political arena.

Personal and political

Living in Milwaukee before graduate school, Nickodem spent two years working with College Possible, a nonprofit education organization that began 15 years ago in the Twin Cities, helping low-income students go to college. Nickodem began by helping high school juniors prepare for their upcoming ACT. The following year, he worked with the same students as they toured colleges, wrote essays for scholarship applications, and applied for financial aid.

“Learning about their backgrounds and what factors influenced them to get to where they were and why they wanted to pursue a college degree played a big role in the reason I decided to come to the University of Minnesota and focus on improving educational opportunities for all students,” he says.

The University provided more ways for Nickodem to continue and expand his work to close gaps. One of them is with Generation Next, an organization seeking to close achievement and opportunity gaps in Minneapolis and St. Paul.

Generation Next aims to help citizens and policymakers better understand the data behind statistics. Nickodem has been able to organize information in a way that helps Generation Next representatives speak knowledgeably in the political realm.

“This work really opened my eyes to how much of a role politics plays in education,” Nickodem says.

Last year, for example, Governor Dayton called for a revaluation of the mandated assessments in the state of Minnesota in his 2014 State of the State address. Though improving education is widely agreed upon across the political spectrum, people often disagree on what changes need to be made and how they should be implemented. For significant advances to take place, Nickodem says, everyone must be on board.

Through his work with College Possible and Generation Next and his research on standardized test scores, Nickodem hopes to make a direct impact on changing the educational system as a whole, even if it takes time.

“The achievement gap is a big thing right now, and we are learning that it is on individuals to change,” he says. “It was created due to systematic inequalities, so it requires everyone to do their part to change the system that created these inequalities—whether they are in high school, academia, government, or a citizen in the community.”

Nickodem is doing his part.

Learn more about the quantitative methods program, the Department of Educational Psychology, and Generation Next.

Story by Ali Lacey | Photo by Susan Andre | February 2015