Understanding Longitudinal Analysis


Cause and effect are natural concepts to most people. In our everyday world we naturally link what has happened before to what happened later and make judgments about whether one event led to the other.

But how can we be sure that one event led to the other?

For some questions we will be able to carry out a controlled experiment - that is, we can vary the treatment received by two or more groups of people who are identical in all other respects but this isn’t always possible or ethical. Does breastfeeding improve child development? It isn’t ethical to randomly allocate children to be breastfed or not. Does a bad neighborhood make someone more likely to commit criminal acts? It isn’t practical set the ‘badness’
of neighborhoods.

The only answer is to use observational studies which observe the strength of the relationship between X and Y without manipulating the relationship. But this to has its own problems. If we measure X and Y at the same time and the relationship is strong, how do we know that X caused Y and not vice versa?

The answer is to use longitudinal studies

Longitudinal studies follow the same individual over time and test the relationship between causes and effects by observing whether X always proceeds Y. Using this approach we can even get round the fact that we didn’t control all the other conditions that might influence the relationship between X and Y. With a large enough number of individuals there will be variation in all the important factors allowing us to construct statistical models of the relationship between X and Y.

© 2012 CLSI Suffusion theme by Sayontan Sinha