Fixed Effects Regression Methods for Longitudinal Data Using SAS

2.7. An Example with Unbalanced Data

In the NLSY example, the data set was balanced with exactly three observations for every child. But the fixed effects method is by no means limited to balanced data; it can be applied without modification to data in which some persons have many observations and others have few. Persons with only one observation, however, are effectively eliminated from the analysis because such individuals have no within-person variation.

In this section I apply the fixed effects method to a highly unbalanced data set in which the "individuals" are schools and the "occasions" are students within each school. The National Education Longitudinal Study (NELS) began in 1988 and included 1,003 schools with a total 21,580 students. ...