--- george owuor wrote:
> I am interested to evaluate effect of a credit programme on
> farm production besides other factors of production such as
> fertiliser. My data have participants and non-participants.
> How would I go about this knowing well that credit influence
> output via purchased factors such as fertiliser, hired
> labour.
This is actually quit simple assuming that you will use
-regress- in order to estimate these effects. Lets say your
dataset consists of 4 variables, called production, credit,
labour, and fertiliser. You are assuming that credit has both
a direct effect on production, and an indirect effect through
credit and fertiliser. You can get the direct effect by typing:
regress production credit labour fertiliser
You claimed you wanted to know the effect of credit while
controlling for fertiliser and labour (the direct effect). In this
case this is probably the most uninteresting effect. How would
participating in a credit program influence production other then
by allowing people to buy means to increase production, like
fertiliser and labour? The direct effect is a sort of residual: the
effect of participating that cannot be explained by any of the
variables in the model. Residuals are important in guiding future
research or give clues on how to improve our model, but are not
really of substantive interest. This is not always true. Looking at
direct effects does make sense in cases where one does expect a direct
effect and is theoretically interested in that effect. I think this is
not the case here, since I don't see any reason to expect a direct
effect of participating in a credit program on production. In your case
I would be more interested in the indirect and total effects. For more
on those see: http://www.ats.ucla.edu/stat/Stata/faq/pathreg.htm .
Hope this helps,
Maarten
