Simultaneity and Its Impact on Ecological Regression Applications

Regression analysis of ecological relationships, which have been specified as simultaneous, results in biased parameter estimates and spurious inference. Examples of the problem are taken from the ecological literature. An alternative regression approach is to add a time dimension to the specified relationships, which removes the simultaneity. Alternatively, if measurements on exogenous variables are available, then standard techniques, such as two-stage least squares, can be used.