Estimating men’s fertility from vital registration data with missing values
Level 5 - mechanism / opinion, no new human data
Methodological simulation study using demographic data; graded by non-clinical design analogy.
OpenAlex W2781958466 · doi:10.1080/00324728.2018.1481992
What was done
The authors evaluated two imputation approaches for estimating men's age-specific fertility rates and related demographic measures from vital registration data with missing paternal age. The first method imputed missing paternal ages using the unconditional paternal age distribution of recorded births, while the second imputed missing values conditional on maternal age. The methods were evaluated using simulation studies modeled after vital registration patterns from Sweden, the United States, Spain, and Estonia, varying both the overall proportion and the age selectivity of missing data.
What was found
The abstract reports no numerical error metrics, bias estimates, or specific performance values. Qualitatively, the simulation results showed that the maternal-age-conditional imputation method outperformed the unconditional approach across the majority of tested simulation conditions.
Why it matters
Male fertility metrics are frequently understudied due to widespread missing paternal data in official birth registries. This study provides demographic researchers with a validated, maternal-age-dependent imputation strategy to improve the accuracy of male fertility estimates.
Limits
The abstract provides no quantitative metrics regarding the magnitude of improvement, error rates, or bounds of acceptable missingness. Findings are based on simulated data modeled on four high-income countries and may not generalize to registration systems with fundamentally different missingness mechanisms or maternal-paternal age relationships.
Cited by
- supports In Sweden, the total number of children linked to fathers via administrative registry data is about 6% to 8% less than those linked to mothers.