Machine Learning Models Trained on a Longitudinal Survey Dataset Poorly Predict Hundreds of Life Outcomes
Abstract
Understanding the predictability of life outcomes is important for public policy and science, but existing studies focus on a limited number of outcomes and therefore struggle to describe the broad patterns of predictability. We developed a high-throughput approach that uses a single dataset to study the predictability of many future outcomes using past data. We deployed this approach with hundreds of outcomes and thousands of machine learning models per outcome in the Future of Families and Child Wellbeing Study. Three findings emerged. First, most outcomes were not very predictable, but some were much more predictable than others. Second, even for the most predictable outcomes, our complex prediction procedure typically exhibited little skill relative to a simple procedure. Third, we found modest patterns in how predictability varied across outcomes. We conclude by discussing scope conditions and how future research might address them.
Related
- Can Intra-Operative Stimulation Data Predict Post-Operative Deep Brain Stimulation Outcomes? — A Machine Learning Approach Based on Probabilistic Maps
- Comparison of parametric versus machine-learning multiple imputation in clinical trials with missing continuous outcomes.
- Constraint Learning by Optimizing over Trained Neural Networks: A Pragmatic Tutorial
- Sample size determination for randomized decentralized clinical trials with longitudinal outcomes
- Bayesian joint modeling of longitudinal patient-reported outcomes and survival: an application to chronic obstructive pulmonary disease
BibTeX
@misc{cantrell2026machine,
title = {{Machine Learning Models Trained on a Longitudinal Survey Dataset Poorly Predict Hundreds of Life Outcomes}},
author = {Emily M. Cantrell and Pranay Anchuri and Hanzhang Ren and Matthew J. Salganik},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.17115},
eprint = {2610.17115},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.17115},
}
RIS
TY - UNPB
TI - Machine Learning Models Trained on a Longitudinal Survey Dataset Poorly Predict Hundreds of Life Outcomes
AU - Emily M. Cantrell
AU - Pranay Anchuri
AU - Hanzhang Ren
AU - Matthew J. Salganik
PY - 2026
DA - 2026/10/12
PB - arXiv
DO - 10.48550/arxiv.2610.17115
UR - https://arxiv.org/abs/2610.17115
AB - Understanding the predictability of life outcomes is important for public policy and science, but existing studies focus on a limited number of outcomes and therefore struggle to describe the broad patterns of predictability. We developed a high-throughput approach that uses a single dataset to study the predictability of many future outcomes using past data. We deployed this approach with hundreds of outcomes and thousands of machine learning models per outcome in the Future of Families and Child Wellbeing Study. Three findings emerged. First, most outcomes were not very predictable, but some were much more predictable than others. Second, even for the most predictable outcomes, our complex prediction procedure typically exhibited little skill relative to a simple procedure. Third, we found modest patterns in how predictability varied across outcomes. We conclude by discussing scope conditions and how future research might address them.
ER -