US over 41 years: 9-year-olds read 13 points better, 13-year-olds 8, and 17-year-olds as in 1971
In maths since 1978: plus 25, 21 and 6 points; the older the pupils, the smaller the gain. This is the national test with the same kind of tasks; 17-year-olds have not taken it since 2012.
More
The US long-term national test has given 9-, 13- and 17-year-olds tasks of the same design since the early 1970s. Over forty years the younger pupils gained clearly in both reading and maths, while at the graduating age the reading gain stayed within the error and the maths gain was small.
Bars: change in the average reading score from 1971 to 2012 for each age. Scores of different ages are not compared with each other, only changes within each series.
- US: reading at age 9 points
1971: 207.6; 2012: 220.8
- US: reading at age 13 points
1971: 255.2; 2012: 262.8
- US: reading at age 17 points
1971: 285.2; 2012: 286.9
- US: maths at age 9 points
1978: 218.6; 2012: 244.0
- US: maths at age 13 points
1978: 264.1; 2012: 285.0
- US: maths at age 17 points
1978: 300.4; 2012: 306.3
NCES, NAEP Long-Term Trend: national mean scores, 1971 and 2012 (reading), 1978 and 2012 (maths); significance from NCES's comparison with the first year. Reading: age 9 207.6 → 220.8, age 13 255.2 → 262.8, age 17 285.2 → 286.9 (not different from 1971).
Seventeen-year-olds are tested only if still in school: over these years fewer dropped out, so more weaker pupils may have been tested, and the composition changed; a flat average can hide gains within groups. The test carries no stakes for older pupils; motivation is another possible explanation. After 2012 the 9- and 13-year-olds fell (2020–2023); that is another card.
A question for this picture
And adults who left school long ago: do they read better or worse than ten years before?