Students of Color Were Closing the Achievement Gap. COVID-19 Is Projected to Erode that Progress.

By Henry Kronk
July 20, 2020

In March, COVID-19 turned American education on its head. The nearly overnight switch to online learning was seamless for the few with experience with online learning. For everyone else, it was jarring. While the digital divide had been all but closed in brick-and-mortar schools, it was still far too wide for both students and teachers at home. Many have speculated and projected learning losses for large swaths of American students. But, increasingly, it’s clear that students of color are affected at higher rates than others. After decades of gains by Black and Latino students in high school achievement and graduation, many wonder how far progress is going to walk back.

a graph of the high school achievement gap for students of color and whites. The gap was closing.
High school completion rates of 18- to 24-year-olds, by race/ethnicity: 2000 through 2016. SOURCE: U.S. Department of Commerce, Census Bureau, Current Population Survey (CPS), October 2000 through 2016. See Digest of Education Statistics 2017, table 219.65.

Students of Color Participate Less in Online Learning

Earlier this month, the Los Angeles Unified School District (LAUSD) published a landmark study investigating the digital learning habits of its students from March 16 to May 22.

The district uses the learning management system Schoology, and they were able to get detailed information regarding how learners participated in remote classes. (For example, researchers could see when students initiated a session, accessed content, completed assignments, and more.)

Researchers found that students of color, students with disabilities, and those that live in low-income neighborhoods participated significantly less than their peers. At the middle school level, 14% of Black students and 10% of Latino students did not log on to Schoology once during the period of the study.

Only 60% of all students were active to some degree each day. Weekly participation among Black high school students peaked at 71%. For Latino students, it was 73%.

These findings support an earlier survey of 400 LAUSD parents by the advocacy group Speak Up. They polled parents to see how much live online interactions their kids were getting from their teachers. Students of color received significantly less than their White peers. 10.94% of Black parents and 16.51% of Latino parents said their child never received any live online instruction.

How Will COVID-19 Affect Learning?

It will be years before we can accurately measure the effect that COVID-19 has had on learning. However, researchers have begun to model and predict likely outcomes. In a working paper, researchers from Brown University projected that students left the 2019-20 school year with 63%-68% of the learning gains in reading compared to a typical year. With math, it was much worse: 37%-50% of normal gains.

The researchers speculate that losses will be worse for students of color. Native, Black, and Latino Americans are affected by COVID-19 at much higher rates than their Asian and White counterparts.

“Furthermore,” the authors write, “the so-called “digital divide” in technology and internet access by race/ethnicity and socioeconomic status likely contributes to greater inequalities during the COVID-19 pandemic …”

Researchers at McKinsey and Company modeled three different scenarios for learning losses and gains. They estimate that average remote learning results in three-four months of learning losses over the school year. In a scenario in which in-class learning doesn’t resume until January 2021, they estimate that learners of color, and especially low-income students, will lose far more months of academic gains than Whites. They project that Black students will lose 10.3 months, Latino students will lose 9.2 months, and low-income learners will lose over a year of learning.

The Achievement Gap Was Closing. What Will Happen to Those Gains?

These losses may occur after a time when the high school achievement gap was closing in the U.S. In 1972 (the first year that the National Center of Education Statistics began compiling this data), learners of color had abysmal high school completion rates. For Black adults aged 18-24 72.1% earned a high school diploma. For Latinos, it was 56.2%. Over the following decades, however, those rates steadily climbed. In 2016, 92.2% of Black learners aged 18-24 had earned a high school diploma, while 89.1% of Latino students had done the same. White (94.5%) and Asian (96.8%) rates remained higher, but the gaps were a fraction of what they were.

These gains are now in jeopardy.

The McKinsey researchers conclude their report, writing, “These numbers are sobering—but they are not inevitable. If the United States acts quickly and effectively, it may avoid the worst possible outcomes. But if there is a delay or a lack of commitment, COVID-19 could end up worsening existing inequities.”

Featured Image: Joseph Perez, Unsplash.


  1. “As bots enter the classroom, both teachers and learners will have to reflect on their uses and outcomes. They will need to adopt an awareness of AI’s presence. Teachers must recognize AI’s short comings, such as inherently developing biases and its inability to process human emotions.”

    This statement is correct as it relates to AI, generally; however, it assumes that AI exists as THE entity that students directly interact with. There are many potential expressions of AI, including a human-in-the-loop approach, in which it is configured in such as way as to facilitate dialogs and interactions between people, either studentteacher or studentstudent.

    For example, we’re building an L2 language speaking practice app (Language Hero Smart Chat). We use AI to enable beginning students, who speak different languages, to have natural, real life conversations in each other’s language from Day 1. They speak directly to each other, interacting with the system only to select from multiple content choices suggested by it, designed to facilitate a real free-ranging dialog resulting in real bonding, to the extent it’s possible, rather than to practice a particular lexical structure (they can also text or go off the grid to have pure video chat).

    Teachers can use this system as well for group chat. They can upload their own curriculum as well (the Smart Chat system configures it as multiple vector (branching script) chat or merges it with the system curriculum (focused on real life useful topics like travel, food, shopping, social chat, expressing ideas, etc.). Everything they say is comprehensible to their students, and so are all student responses.

    When such a system is implemented in a manner that pays particular attention to the affective components that make human interaction so effective for creating the desire to learn (and corresponding openness to processing L2 content, in this case), we think it can be a more effective tool than bot chat.