ABSTRACT
There are only two ways of forecasting – trend extrapolation or scenario modelling. Trend extrapolation involves gathering historical data, such as the mature height of people in a population, and after concluding that on average they are getting taller at a regular rate,
estimating future average height. This is a fairly primitive form of forecasting and it is much
better to build a model. Models involve the measurement of relationships and assume that
if a change in one or more related factors resulted in a change in other factors, then it will
happen again in similar circumstances. In crop experimentation, for example, it is normal to
take several sets of seeds and subject them to different influences. One might be planted in a
new compost, one might receive more light, another may be sprayed daily with a chemical
and so on. One set will be allowed to grow without new methods and in due course it will be
possible to measure the effect of each of the treatments in comparison to the normal group. In
this way the increased yield from investing in a new treatment can be estimated and the most
successful and cost effective adopted for the full crop. If the analogy needs to be taken further,
consider what will happen to a seed that is planted (the first sale) and then left alone – it will
probably wither and die.