|
When it is not possible to use another set of farmers as controls one needs to devise a strategy whereby study farmers can provide their own form of control. This can be achieved by setting up a longitudinal study that provides for an 'intervention' period to follow a 'baseline control' period. Such a study needs to be long enough, however, to allow adequate numbers of observational units to be defined over time to provide a sound statistical analysis. There could be significant levels of confounding with uncontrollable random fluctuations over time. |
||
|
The observational unit in such 'before/after' comparisons needs to be some form of unit in time. In agronomic or livestock production systems this unit of time is likely to be as much as one year. Crops tend to be planted once a year at the onset of the rains. Livestock, especially in livestock crop systems, will also tend to follow an annual pattern of production. Furthermore, the contrasting wet and dry seasons that occur within years need to be taken into account. |
|
|
