ABSTRACT
The non-experimental nature of econometrics is a main feature that helps distinguish econometrics from its parent discipline of statistics. Econometricians have little control over sample size and face uncertainty about model specifications as compared to other scientific disciplines in which experiments can be conducted. Another complication is the additional complexity caused by the fact that factors not directly of interest must be taken into account by econometric models rather than handled by experimental design. Thus, these models typically contain large numbers of unknown parameters whose true relationships and growth pattern need to be estimated. The econometrician’s task is often made much harder in complex surveys and cross-sectional and time-series applications, where many of the variables whose relationships need to be untangled and estimated exhibit reasonably similar growth patterns.
