Cargando…

Using machine learning and qualitative interviews to design a five-question survey module for women’s agency

Open-ended interview questions elicit rich information about people’s lives, but in large-scale surveys, social scientists often need to measure complex concepts using only a few close-ended questions. We propose a new method to design a short survey measure for such cases by combining mixed-methods...

Descripción completa

Detalles Bibliográficos
Autores principales: Jayachandran, Seema, Biradavolu, Monica, Cooper, Jan
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Pergamon Press 2023
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9693692/
https://www.ncbi.nlm.nih.gov/pubmed/36597415
http://dx.doi.org/10.1016/j.worlddev.2022.106076
_version_ 1784837607075610624
author Jayachandran, Seema
Biradavolu, Monica
Cooper, Jan
author_facet Jayachandran, Seema
Biradavolu, Monica
Cooper, Jan
author_sort Jayachandran, Seema
collection PubMed
description Open-ended interview questions elicit rich information about people’s lives, but in large-scale surveys, social scientists often need to measure complex concepts using only a few close-ended questions. We propose a new method to design a short survey measure for such cases by combining mixed-methods data collection and machine learning. We identify the best survey questions based on how well they predict a benchmark measure of the concept derived from qualitative interviews. We apply the method to create a survey module and index for women’s agency. We measure agency for 209 married women in Haryana, India, first, through a semi-structured interview and, second, through a large set of close-ended questions. We use qualitative coding methods to score each woman’s agency based on the interview, which we use as a benchmark measure of agency. To determine the close-ended questions most predictive of the benchmark, we apply statistical algorithms that build on LASSO and random forest but constrain how many variables are selected for the model (five in our case). The resulting five-question index is as strongly correlated with the coded qualitative interview as is an index that uses all of the candidate questions. This approach of selecting survey questions based on their statistical correspondence to coded qualitative interviews could be used to design short survey modules for many other latent constructs.
format Online
Article
Text
id pubmed-9693692
institution National Center for Biotechnology Information
language English
publishDate 2023
publisher Pergamon Press
record_format MEDLINE/PubMed
spelling pubmed-96936922023-01-01 Using machine learning and qualitative interviews to design a five-question survey module for women’s agency Jayachandran, Seema Biradavolu, Monica Cooper, Jan World Dev Article Open-ended interview questions elicit rich information about people’s lives, but in large-scale surveys, social scientists often need to measure complex concepts using only a few close-ended questions. We propose a new method to design a short survey measure for such cases by combining mixed-methods data collection and machine learning. We identify the best survey questions based on how well they predict a benchmark measure of the concept derived from qualitative interviews. We apply the method to create a survey module and index for women’s agency. We measure agency for 209 married women in Haryana, India, first, through a semi-structured interview and, second, through a large set of close-ended questions. We use qualitative coding methods to score each woman’s agency based on the interview, which we use as a benchmark measure of agency. To determine the close-ended questions most predictive of the benchmark, we apply statistical algorithms that build on LASSO and random forest but constrain how many variables are selected for the model (five in our case). The resulting five-question index is as strongly correlated with the coded qualitative interview as is an index that uses all of the candidate questions. This approach of selecting survey questions based on their statistical correspondence to coded qualitative interviews could be used to design short survey modules for many other latent constructs. Pergamon Press 2023-01 /pmc/articles/PMC9693692/ /pubmed/36597415 http://dx.doi.org/10.1016/j.worlddev.2022.106076 Text en © 2022 The Author(s) https://creativecommons.org/licenses/by/4.0/This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
spellingShingle Article
Jayachandran, Seema
Biradavolu, Monica
Cooper, Jan
Using machine learning and qualitative interviews to design a five-question survey module for women’s agency
title Using machine learning and qualitative interviews to design a five-question survey module for women’s agency
title_full Using machine learning and qualitative interviews to design a five-question survey module for women’s agency
title_fullStr Using machine learning and qualitative interviews to design a five-question survey module for women’s agency
title_full_unstemmed Using machine learning and qualitative interviews to design a five-question survey module for women’s agency
title_short Using machine learning and qualitative interviews to design a five-question survey module for women’s agency
title_sort using machine learning and qualitative interviews to design a five-question survey module for women’s agency
topic Article
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9693692/
https://www.ncbi.nlm.nih.gov/pubmed/36597415
http://dx.doi.org/10.1016/j.worlddev.2022.106076
work_keys_str_mv AT jayachandranseema usingmachinelearningandqualitativeinterviewstodesignafivequestionsurveymoduleforwomensagency
AT biradavolumonica usingmachinelearningandqualitativeinterviewstodesignafivequestionsurveymoduleforwomensagency
AT cooperjan usingmachinelearningandqualitativeinterviewstodesignafivequestionsurveymoduleforwomensagency