The utility of longitudinal studies and the differences between longitudinal and cross-sectional designs are described more fully in the Learning Hub’s Introduction to Longitudinal Studies. There are data analysis methods that allow us to make use of the rich data collected by longitudinal studies and to explore the temporal relationships between measures collected across different life stages. Each of these is suited to the analysis of different types and combinations of variables. Some variables are continuous (e.g. age) and others are categorical (e.g. a list of occupations). We call categorical variables with two levels ‘dichotomous’ (e.g. deceased or living) and, where they are coded as 0 or 1, we can also call them ‘binary’. This guide will teach you about different analytic approaches to exploring how certain types of outcomes are associated with potential explanatory factors.
Dissimilar outcomes can occur even among people who share the same characteristics. The term ‘heterogeneity’ is often used to refer to differences like these. Longitudinal data can help control for such differences by including a wide range of explanatory variables across the life course in statistical models. The problem of ‘omitted variable bias’ is also improved by using longitudinal data, but always remains, as there are connections between the outcome and explanatory variables that have not or could not be included as they are unmeasurable.
We will use an extract from the National Child Development Study (NCDS) CLOSER Training Dataset to illustrate some of the different methods that can be used in analysing longitudinal data. The NCDS is a cohort study of people born in England, Scotland and Wales during a single week of 1958. In the NCDS, detailed information has been collected on participants from childhood, through adolescence into early adulthood and later life, allowing us to look at different outcomes and potential explanatory variables.
Measurements that have been collected over time include assessments of physical health (e.g. Body Mass Index (BMI) measured at ages 7, 11, 16, 23, 33, 42 and 50), as well as a series of mental health (e.g. Malaise inventory), socio-economic position, and behavioural factors (e.g. smoking), measured at ages 23, 33, 42, and 50. These measures are examples of the variety of data available in the NCDS and other longitudinal studies.
The Learning Hub is a resource for students and educators
tel | +44 (0)20 7331 5102 |
---|---|
closer@ucl.ac.uk |
Sign up for our email newsletters to get the latest from CLOSER
Sign up