Abstract:

In 2019 Growing Up in New Zealand, a large longitudinal cohort study of ~6000 children, introduced “Our Voices”, an innovative five-year project that aims to develop novel ways of collecting and analyzing longitudinal information from young people, so we can get a fuller picture about what shapes their overall well-being – including their educational and home lives. A key component of this project has been the experimental co-design of digital platforms to enable young people to engage in research and share their voices in ways that are less constrained by traditional questionnaire design. We worked directly with young cohort members to co-create a web-based app called “Tō Mātou Rerenga - Our Journey”, capable of collecting large sample multi-modal qualitative data in an engaging way (including free text, audio, photos and videos), about various aspects of their lives, on their own terms and in their own voices.

Our Journey is currently being trialed with the existing Growing Up in New Zealand cohort of 13-year-olds. The cohort has been followed longitudinally via more traditional questionnaires on multiple occasions with their parents, from before their birth until the most recent Growing Up data collection at age 12. The Our Voices project was developed in an attempt to reduce the biased attrition that is common in longitudinal birth cohort studies across the teenage years globally, as well as to ensure that voices that are least often heard (and more likely to be lost) could continue to enrich the longitudinal data as the cohort moved into early adulthood.

Additionally, in order to analyze the multi-modal qualitative data obtained from cohort members who engage with the app, innovative machine learning techniques have enabled us to analyze multi-model data in a timely way. In parallel, these techniques are being tested against more traditional, manual qualitative approaches to assess applicability for the New Zealand context, and to determine any bias or error in the existing tools. We are also integrating and comparing the qualitative data with the existing quantitative longitudinal information to provide robust and rich evidence about young peoples' well-being.

This presentation will share some of our preliminary findings, including: the benefits and challenges of data collection based on a co-designed tool; the methodological learnings in terms of the extent to which existing machine learning techniques are capable of analyzing complex and nuanced multi-modal qualitative information from young people in New Zealand; and some interesting preliminary insights shared by the young people in the study, such as their perceptions of the quality of their educational experience.

Keywords:

Co-design, longitudinal research, large sample qualitative, machine learning.