Longitudinal Study Design

Cohort Studies

Cohort studies are defined by a common characteristic -one or more groups (cohort) are followed longitudinally -relationship between outcome of interest and certain characteristics (risk factors, exposure) is analyzed -STAYS INTACT FROM THE BEGINNING of the study

Time Series Design

Investigators collect data from the same participants at different time points in order to observe the pattern changes within (intra-individual) and across (inter-individual) individuals

Multi-level modeling

-Hierarchical linear models -statistical models of parameters that vary at more than one level -ex: a model of student performance that contains measures for individual students as well as measures for classrooms within which the students are grouped

Types of Longitudinal Studies

-Panel Studies -Cohort Studies -Time series design -Multi-level modeling

Longitudinal Study Design

-Two or more observations of a certain characteristic within the same group -usually an OBSERVATIONAL study design ---no intervention ---observation of naturally occurring phenomenon -variation: natural experiments (researcher cannot assign people to certain groups) -track changes over a long time (several weeks, months) -dependent variable/outcome measures determine whether the test will be longitudinal or pre/post test assessment only -usually longitudinal studies are 6 weeks or longer

Panel Studies

-provides longitudinal, nationally representative data on a group (panel) on various indicators -panel members may or may not be the same over years -usually a survey -carry over the people that you can; and you refresh the panel with new people ---people who were assessed at first may not respond years later for a follow-up, so researchers can add more people

Issues

-understanding of cross-sectional data -requires time and resources -assumption of "local independence" -even when the group is refreshed in a panel, the longitudinal study is always correlated ---the 2 groups are never independent; don't assume local independence -ANOVA - good for three or more observations

Purpose and Considerations

Purpose: -detect changes over time ---within group ---within and between groups -establishes causality based on temporal order of events Considerations: -time intervals --> even vs. uneven -interactions Usually done with one group of people over an extended period of time -although you could have a control group