LBJ School team makes research discoveries, places 2nd at Microsoft and ODI's Education Open Data Challenge

Share this content

May 6, 2021

Internet access is a greater driver for remote learning than COVID cases and death rates, and Texas did better than the national average in closing the digital divide, LBJ School team discovers.

Professor Kenneth Flamm and a team of graduate public policy students from the LBJ School of Public Affairs at The University of Texas at Austin identified broadband infrastructure as more important than the number of COVID cases and death rates in a school district's decision to switch to remote learning, while placing second in an international data science competition run on the X-Prize platform. The team also discovered that Texas did better than the national average in closing the digital divide, and that in Texas, Dallas did best.

The competition, the Microsoft and ODI's Education Open Data Challenge, was launched to better understand the relationship between access to digital infrastructure and education outcomes for young students. This year's competition focused on closing the education digital divide.

Microsoft announces the winners of the Education Open Data Challenge (May 5, 2021)

"A necessary first step in understanding the impact of the pandemic on U.S. students' educational progress over the last two years is to understand how and why U.S. school districts made the decision to move their classes online," Flamm said. "When academic performance data for U.S. students become available later this year or next, understanding how and why some school districts opted to go online while others did not will be critical to correctly assessing the impact of online classes on educational performance."

Alongside Professor Flamm, LBJ school students Tyler Baines, Cosme Dominguez, Jack English, Lillian Hatcher, Gina Hinojosa and Lindsay Hodge — students in Professor Flamm's Data Visualization, Statistics and Econometrics class — entered the competition. To better understand school district decisions during COVID-19, the team constructed a large scale database of relevant data from a variety of public and private sources and estimated a linear probability model of school district decisions to move online in the current academic year.

They looked at school districts that went remote, assessed when and why they decided to go remote, and worked to uncover if the digital divide played a role. Using Microsoft and BroadbandNow data, as well as data on demographics and school district size, the team's preliminary conclusion was that broadband infrastructure quality seemed to have a significant impact on a school district's decision to switch to remote learning and entirely online teaching methods. Conversely, cumulative COVID-19 case and death rates had very small effects on the probability of a school district choosing online teaching modes. The team concluded that, along with the importance of the local political sentiment to shaping the U.S. response to COVID-19, broadband infrastructure quality also was a significant factor affecting policy outcomes.

"We made some interesting discoveries," Flamm said. "Broadband infrastructure quality and availability, school district size (in students), statewide factors (for example, state-level policies and mandates), and local political sentiment had large and statistically significant impacts on "onlining" decisions, while within-district household median income levels, COVID intensity (case and death rates), prior academic performance and broadband price seemed to play little or no direct role in shaping school district decisions."

"We made some interesting discoveries. The data science and policy analysis skills that the LBJ students learned at the LBJ School were critical to our success in this competition." —Kenneth Flamm

The team also discovered that nationwide, the "digital divide" between households with good, reliable access to the internet for education, and households with poor or unreliable access to the internet, had narrowed somewhat during the pandemic. Most interestingly, Texas had narrowed this digital divide more than the rest of U.S. average. And within Texas, the Dallas-Fort Worth metropolitan area seems to have narrowed educational internet access disparities more than other areas in Texas, on average.

"The apparently superior performance of Dallas in narrowing the digital divide "popped out" in our analysis of available data, and guided us toward investigating what Dallas was doing that was different from the strategies utilized (with positive results) in other parts of Texas," Flamm said. "We learned that Dallas, in addition to distributing cellular internet hotspots to families (a model used in other parts of Texas), had adopted a very innovative policy of extending in-school internet networks into disadvantaged neighborhoods using newly available, inexpensive private fixed wireless networking technology. Similar experiments now appear to be underway in other regions of Texas."

Between late spring of 2020, and the resumption of classes in January 2021, the share of Dallas households with kids lacking reliable home internet access for education fell by almost 7 percentage points, compared to a 3 percentage point drop in Houston, and slightly under a 6 percentage point drop in the average for the rest of Texas. Nationally (excluding Alaska), the drop in households lacking reliable internet access for school kids dropped just 2 percentage points, so Texas generally did better than the US national average in narrowing this educational digital divide.

If access to the internet for education is broken out by schoolchildren's household income level, the share of households earning less than $50,000 that lacked reliable internet access dropped by about 2 percentage points outside of Texas over that same time period. In Texas, by contrast, the share of these poorest households lacking reliable access for kids dropped by 15 percentage points in Dallas, 8 percentage points in Houston, and 5 percentage points in the rest of Texas.

Flamm said he and his teammates were excited to learn Texas was making such tangible progress in closing the education digital divide. The team is now working on a followup analysis.

"The data science and policy analysis skills that the LBJ students learned at the LBJ School were critical to our success in this competition," Flamm said. "The LBJ School itself introduced an innovation in its curriculum last year that recognized the growing importance of this skill set, by creating a new Dean's Certificate in Data Science and Policy Analysis that is available to students in all our graduate degree programs. The first five recipients of this new LBJ School certificate graduated in May of 2020, and we are likely to have a similar number graduating this May."

The team has received $35,000 in prize money for a nonprofit of their choice.

LBJ School nonprofit and data studies professor Ji Ma also entered the competition with teammates Dani Lindo, Sung Joon Roh and Janani Ravikumar. They studied digital access in Texas and identified areas that would benefit from digital infrastructure and support during the pandemic.


News category: