Analysing longitudinal data
Presentations by British academic Professor Stephen Jenkins have helped boost New Zealand’s capability in analysing longitudinal data.
Stephen, from the Institute for Social and Economic Research (ISER) at the University of Essex, was in New Zealand in June and July on a visit funded by a SPEaR Linkages Visiting Speaker Award and hosted by research institute MOTU. He spoke on longitudinal data analysis at a number of seminars, workshops and meetings organised by MOTU, Statistics New Zealand, the Ministry of Social Development, and Victoria University, strengthening the linkages within the academic, policy and research communities.
MOTU senior fellow Dave Maré said the visit was timely as it followed the release of the first set of results from Statistics New Zealand’s SoFIE survey, a longitudinal survey which, as more results came on-stream over the years, would provide insight into the dynamics of how people’s lives unfold.
“Stephen’s visit reinforced the value of those data sources. His visit will help build capability here for analysing longitudinal data and gaining social policy insights, so we can get the best out of the information from SoFIE. To get the most out of longitudinal data, researchers need to develop skills and use special statistical tools that are different from those used in analysing cross-sectional surveys. One workshop, held at Victoria University of Wellington, focused on helping researchers gain some of these skills, and to understand the strengths and problems of longitudinal research. Other workshops, attended by academics and public sector and independent researchers, provided hands-on experience in using statistical software,” Dave said.
“One of the challenges in analysing longitudinal data is allowing for the fact that people are different in their life trajectories or in their responses to change. Understanding these differences is crucial for policy analysis, and is not possible without longitudinal data. Longitudinal data helps you to understand what is really going on behind the trends in social outcomes that we monitor.”
Richard Arnold, organiser of the Victoria University workshop, also emphasised the importance of building capability in longitudinal data analysis.
“By tracking the same set of individuals over time, longitudinal surveys can detect the effects on a person’s life of a significant event, such as becoming unemployed or retraining. These effects on income or health, for example, may only become apparent some time later in the person’s life, and cannot be seen in repeated cross-sectional surveys, where a new sample of respondents is selected each time the survey is run,” Richard said.
