Germany drops to the OECD science average for the first time since 2006

Two proportions in the German results carry more diagnostic weight than any score. About a fifth of 15-year-olds clear proficiency level II in none of the three domains ^[14], while 3% reach the top two levels in all three ^[17]. That is close to seven students below the basic threshold everywhere for each one above the high threshold everywhere ^[5], and PISA's own reading of level II is that students beneath it lack the skills to meet the ordinary demands of education, work and everyday life ^[12].

The thinning at the top is the part that resists a purely remedial diagnosis. Measured against 2015, the high-performing group has lost about 38% of its share in mathematics and 42% in reading, against 18% in science ^[4]. Samuel Greiff, who leads the German arm of the study at TUM, describes a decline visible across the entire performance spectrum rather than in particular subjects or particular groups of students ^[18].

A pandemic explanation is the one this round was best placed to test, and it does not survive well. A single disruption predicts a dip and then partial recovery; instead German scores came in below the 2022 round in all three domains ^[3]. The decline Greiff describes has been running for about a decade ^[5], which puts its onset near the 2015 baseline against which the below-basic shares doubled ^[7]. He describes the recent findings as a clear downward trend that has taken hold, not a brief dip ^[6].

Ranking comparisons are the least useful numbers here. Germany trails the four Chinese regions of Beijing, Shanghai, Jiangsu and Zhejiang by 148 points in mathematics and by 111 in science, and Singapore by 70 points in reading ^[1][2][3], gaps too large to work as targets. The comparison that bites is with Estonia, Canada, Switzerland and Austria, which the report identifies as significantly ahead of Germany under comparable economic and social conditions ^[10]. Other OECD countries have drifted down too, but Germany's decline is the pronounced one ^[8]. The newly examined digital domain provides no offset: in model-based problem solving using digital tools, German students also sit at the OECD average ^[11].

Socioeconomic status emerges as by far the strongest of the three background factors the study weighed, ahead of gender and immigrant background ^[19], and its correlation with science performance is the highest recorded in the series ^[20]. Among students from disadvantaged families, 2% reach the high proficiency levels, against roughly 25% of students from privileged families ^[21], a ratio of about twelve to one ^[6]; the study puts the influence of socioeconomic status in Germany at three times what it is in Canada ^[22]. That association does not identify a mechanism. PISA measures what 15-year-olds can do at one moment in one cohort, not why, and a single cross-section will not separate schooling from everything else that tracks family circumstance.

The evidence here is strong enough to treat the decline as a standing property of the system rather than the residue of one closure, and thin on what is driving it.