Going Beyond the Average: How You Can Analyze Differences in Experiences and Compositions of Lower- and Higher-Performing Students with NAEP Data

Ebony Walton (NCES), Robert Finnegan (ETS), and David Hoff (MSG)

The average score does not always tell the whole story.

That is why the National Center for Education Statistics (NCES) reports achievement results for students at various points on assessment scales. NCES presents results from the National Assessment of Educational Progress (NAEP) with average scale scores as well as scores for

• lower-performing students, at the 10th- and 25th-percentiles;

• middle-performing students, at the 50th-percentile; and

• higher-performing students, at the 75th- and 90th-percentiles.

Changes in scores at various percentiles help us to see nuances that the average scale score does not provide.

The percentile data are also key to our understanding of contextual differences between lower- and higher-performing students. To enable data users to analyze contextual factors from NAEP surveys for lower- and higher-performing students, NCES has recently added a new feature to the NAEP Data Explorer (NDE).[1]

In that tool, Percentile Profiles are now visible within the Variable categories. This option, in conjunction with other variables in the NDE, will allow you to see the percentages of lower- and higher-performing students who have specified characteristics.

Let’s look at results from one student survey question to illustrate how percentile profile data can show the context of lower- and higher-performing students more accurately. As part of the NAEP survey, NCES asked students how many days they were absent from school in the last month. Findings from the 2022 mathematics assessment showed that, nationally, about a third (33 percent) of eighth-grade students in public schools reported being absent three or more days during the previous month. Percentile profile results, however, show a statistically significant difference between the percentage of lower- and higher-performing students who reported missing three or more days: 46 percent of lower-performing students compared to 21 percent of higher-performing students. In this case, the overall percentage does not fully describe the experiences of lower- and higher-performing students.

You can also see these percentile profile results for various jurisdictions. When looking at the results across the states and the District of Columbia, you will observe a similar result to the nation—larger proportions of lower-performing students than higher-performing students reported missing three or more days of school last month. Up to fifty-nine percent of lower-performing students and thirty-four percent of higher-performing students reported missing three or more days last month, showing how the results vary from one jurisdiction to another.

The Percentile Profiles feature can also provide insights into the activities and states of mind about learning that differentiate higher and lower performers. For instance, in reading in 2022, more higher-performing fourth-graders (70%) reported they “definitely” could recognize the difference between facts and opinion in a text than lower performers. Twenty-seven percent of lower performers reported that they definitely could do this. And, when we look at this aspect of student confidence across types of school locations, we see some statistically significant differences across the students’ reports from city, suburban, rural, and town locations. Seventy-one percent of higher-performing fourth-graders in suburban locations said they definitely could recognize facts from opinions, which is higher than the 68% that reported the same in city and rural locations. See this report in NDE to explore these results.

You may conduct statistical significance tests to compare results between percentile groups and across jurisdictions within a year. However, since percentile scores may change with each year, comparing results over time should be interpreted with caution.

The ability to analyze contextual factors for lower- and higher-performing students strengthens our understanding of student experiences, environments, and other non-cognitive factors. While not causal, the percentile profile data from NAEP are valuable resources to help explain the complexities of our nation’s students and move beyond describing the average.

NCES will publish results from the NAEP 2024 mathematics and reading assessments at grades 4 and 8 on January 29, 2025. Be sure to visit nationsreportcard.gov for the latest findings, including the new NAEP Percentile Profiles data.

[1] The NDE categorizes lower-performing students as those below the 25th percentile and higher-performing students as those at or above the 75th percentile.