Stability and change in male fertility patterns by cognitive ability across 32 birth cohorts
Data files
Jun 28, 2023 version files 182.80 KB
-
model_1_data.RData
-
model_2_data.RData
-
README.md
Abstract
The relationship between cognitive ability (CA) and childbearing remains unsettled. Using Norwegian administrative registers with population coverage, we study how male lifetime fertility patterns differ across cognitive score groups, and how these changed across the 1950–1981 birth cohorts, covering a period characterized by rapid social and economic change. The analyses reveal systematic differences in fertility and fertility timing across CA groups, with high-scoring males having delayed but ultimately higher fertility than lower-scoring males. This pattern remains stable over time despite strong trends towards delayed and reduced fertility. The overall positive relationship between CA and fertility is primarily driven by high rates of childlessness in the lowest-scoring group, with low-scoring males showing higher rates of parity progression conditional on having children.
Methods
The data was derived from Norwegian administrative registers on loan from Statistics Norway. The population registry contributed information on births, deaths, citizenship category and parents, while data from the armed forces contained cognitive ability test scores from military conscription. A set of potential fathers was defined across the included birth cohorts, with the sample conditioned on being born to two Norwegian parents and being alive at age 18. Pseudonymous individual-level identifiers were used to identify each male's conscription test score (if scored) and their children (individuals of both sexes in the population registry with this individual as their assigned father). Father age at birth was calculated as the difference between father and child's birth cohort, and parity was assigned by sorting a father's children by birth year.
Usage notes
The data for each model is provided in two formats, bundled in an RData file: As a data.table, and as a list of data-inputs as required by the Stan estimation model. The data can be opened by R.