Core Concepts
Assessing predictability of fertility outcomes using Dutch survey and register data to advance understanding of fertility behavior.
Abstract
The content discusses a data challenge, PreFer, focusing on predicting fertility outcomes in the Netherlands by combining Dutch survey (LISS panel) and register data. It highlights the importance of assessing predictability, comparing theory-driven and data-driven methods, and utilizing data challenges for scientific progress. The methodology, phases, submission process, evaluation metrics, and determining winners are detailed.
Abstract:
Research on determinants of fertility outcomes.
Lack of predictive evaluation in social sciences.
Introduction to datasets (LISS panel & Dutch register).
Description of fertility prediction data challenge PreFer starting in Spring 2024.
Introduction:
Importance of quantifying predictability in social sciences.
Overview of explanatory vs. predictive modeling.
Significance of out-of-sample predictive ability.
Data Extraction:
"Approximately 25% of people in the LISS dataset had a new child between 2021 and 2023."
"For each person in the sample aged 18-45 in 2020, we calculated the number of children in each year between 2021 and 2023."
Stats
"Approximately 25% of people in the LISS dataset had a new child between 2021 and 2023."
"For each person in the sample aged 18-45 in 2020, we calculated the number of children in each year between 2021 and 2023."