Applying Causal Discovery to Intensive Longitudinal Data

Brittany Stevenson, Erich Kummerfeld, Jennifer Merrill
Proceedings of The 2021 Causal Analysis Workshop Series, PMLR 160:20-29, 2021.

Abstract

Intensive longitudinal data (ILD) could be a solution for two problems in psychology: 1) In traditional experiments and survey studies, ndings are not necessarily representative of the real-life constructs and relationships studied, and 2) Group-level analyses commonly mischaracterize or obscure relationships for individuals. Popular analytic methods within psychology are currently not well-equipped to use ILD for causal discovery and causal inference, however. We have performed the first causal discovery analysis on ILD, encountered some challenges, and developed some solutions to these challenges. This paper describes our application of causal discovery to an example ILD dataset, and addresses two particular challenges that arose: 1) How should one address variables measured on different timelines, and 2) What number of observations is needed for individual-level analysis.

Cite this Paper


BibTeX
@InProceedings{pmlr-v160-stevenson21a, title = {Applying Causal Discovery to Intensive Longitudinal Data}, author = {Stevenson, Brittany and Kummerfeld, Erich and Merrill, Jennifer}, booktitle = {Proceedings of The 2021 Causal Analysis Workshop Series}, pages = {20--29}, year = {2021}, editor = {Ma, Sisi and Kummerfeld, Erich}, volume = {160}, series = {Proceedings of Machine Learning Research}, month = {16 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v160/stevenson21a/stevenson21a.pdf}, url = {https://proceedings.mlr.press/v160/stevenson21a.html}, abstract = {Intensive longitudinal data (ILD) could be a solution for two problems in psychology: 1) In traditional experiments and survey studies, ndings are not necessarily representative of the real-life constructs and relationships studied, and 2) Group-level analyses commonly mischaracterize or obscure relationships for individuals. Popular analytic methods within psychology are currently not well-equipped to use ILD for causal discovery and causal inference, however. We have performed the first causal discovery analysis on ILD, encountered some challenges, and developed some solutions to these challenges. This paper describes our application of causal discovery to an example ILD dataset, and addresses two particular challenges that arose: 1) How should one address variables measured on different timelines, and 2) What number of observations is needed for individual-level analysis.} }
Endnote
%0 Conference Paper %T Applying Causal Discovery to Intensive Longitudinal Data %A Brittany Stevenson %A Erich Kummerfeld %A Jennifer Merrill %B Proceedings of The 2021 Causal Analysis Workshop Series %C Proceedings of Machine Learning Research %D 2021 %E Sisi Ma %E Erich Kummerfeld %F pmlr-v160-stevenson21a %I PMLR %P 20--29 %U https://proceedings.mlr.press/v160/stevenson21a.html %V 160 %X Intensive longitudinal data (ILD) could be a solution for two problems in psychology: 1) In traditional experiments and survey studies, ndings are not necessarily representative of the real-life constructs and relationships studied, and 2) Group-level analyses commonly mischaracterize or obscure relationships for individuals. Popular analytic methods within psychology are currently not well-equipped to use ILD for causal discovery and causal inference, however. We have performed the first causal discovery analysis on ILD, encountered some challenges, and developed some solutions to these challenges. This paper describes our application of causal discovery to an example ILD dataset, and addresses two particular challenges that arose: 1) How should one address variables measured on different timelines, and 2) What number of observations is needed for individual-level analysis.
APA
Stevenson, B., Kummerfeld, E. & Merrill, J.. (2021). Applying Causal Discovery to Intensive Longitudinal Data. Proceedings of The 2021 Causal Analysis Workshop Series, in Proceedings of Machine Learning Research 160:20-29 Available from https://proceedings.mlr.press/v160/stevenson21a.html.

Related Material