In the German state of Brandenburg, a child's school career starts with a date on a calendar: September 30th. Turn seven before it and you are older than the state's key age. Between 2009 and 2025, about 22.6 percent of Brandenburg's third-graders crossed that line, meaning roughly one child in five began school later than the standard timetable. That fraction did not hold steady. It bent sharply upward, and the researchers can name the year it started.
Maxi Woelke, Fabian Arntz and colleagues at the University of Potsdam and Freie Universität Berlin worked with data from EMOTIKON, a project that has measured the physical fitness of third-graders in Brandenburg every year since 2009 and is mandatory for the state's public primary schools. Because every child's enrolment date is recorded, the team could look backwards from third grade and ask a simple question of 292,181 children in 550 schools: did this child start school on time? For 66,188 of them, the answer was no.
The interesting part is when those delays happened. On 29 November 2013, Brandenburg changed the rules. Until then, a school principal decided whether a child was ready, working from a school entrance examination. After the change, following pressure from parental interest groups, parents' own preferences entered the process. The researchers treated that date as a boundary and used a regression discontinuity design, a method that looks for an abrupt step in a trend at a known cutoff, to see whether the line moved.
It moved. Before 2013, the share of late-enrolled children was drifting slightly downward, from 19.7 percent in the 2009 cohort to 18.4 percent in 2014. Afterwards it climbed steadily, cohort after cohort, reaching 26.4 percent by 2025. In relative terms that is an increase of roughly 33.7 percent from 2014 onward. The statistical jump at the regulation date was strong and appeared in both of the team's models.
The pandemic is the more ambiguous case. The authors set a second boundary at 30 January 2020, when the WHO declared a public health emergency, and found no clear additional jump: the numbers simply kept rising along the path the regulation change had already set. They tested alternative cutoffs 30, 60 and 90 days later and got the same answer. Their own reading is careful. The data cannot separate a continuation of the post-2013 trend from a genuine pandemic effect layered on top of it, and they say so plainly.
Who starts late
Boys were about 35.4 percent more likely than girls to be enrolled late, a gap that mirrors what other studies find for diagnosed developmental disorders. The authors raise the possibility that some of this is a detection effect rather than a difference in development. They point to research on ADHD suggesting girls must display a wider range of behavioural and emotional problems before a diagnosis is made, and wonder whether school readiness assessments work the same way.
Geography mattered too. Children living far from Berlin were about 33.8 percent more likely to be late than those in the capital's orbit. Regionally deprived areas showed higher rates, and after 2013 the increase was steepest there, driven mainly by the work-related component of the deprivation index rather than the education component. The team's own explanation is speculative: where jobs are insecure, the costs that come with starting school may weigh more heavily, and a year's delay may look like relief.
One result surprised the authors. They expected rural areas to fare worst. Instead, midsize centres had the highest rates, about 12.2 percent above rural areas, with no reliable difference between rural areas and urban centres at all. They offer a guess (higher living costs than the countryside without a city's health and education infrastructure) and label it as a guess in need of testing.
Why it matters
Delayed enrolment is often read as a health signal. Children who start late are more likely to have been born prematurely, more likely to have hearing or behavioural problems or asthma, and tend to score lower on later academic and fitness tests. That is why researchers watch the number: it is one of the few developmental measures collected on nearly every child in a population.
This analysis complicates that reading. If a single administrative change can shift the rate by a third while children's development presumably did not shift by a third in the same window, then the measure is tracking policy and parental decision-making alongside whatever it tracks about the children themselves. Anyone comparing regions, or comparing years, needs to know which rules were in force.
The practical takeaway sits with the assessments themselves. If boys are being flagged more readily than girls with comparable difficulties, some girls are missing support they would benefit from. And if midsize towns really are harder places to grow up in than either cities or the countryside, that is not where most support is currently aimed. This is one German state, and correlations rather than causes, but it is a large and unusually complete window on both.