Agriculture Reference
In-Depth Information
Given the rather large number of variables enumerated, the likelihood of correlation among
independent or predictor variables is high. For this reason, the test of multi-colinearity was
applied. Assuming two variables, X 1 and X 2, co-linearity is suggested if:
X 1 = λX 2
(3)
However, Equation 2 demands that a more robust function be developed to cater for the
several predictor variables in the model. This can be presented as:
λ 1 X 1 i + λ 2 X 2 i + ... + λ k X ki = 0
(4)
where λ 1 are constants and X 1 are the exploratory variables that might be linearly correlated.
The speed with which variances and covariances increase can be seen with the variance-
inflating factors (VIF), which shows how the variance of an estimator is inflated by the
presence of multi-colinearity. A formal detection tolerance or the variance inflation factor
(VIF) for multi-colinearity as illustrated by Gujarati (2003) can be used as follows:
VIF = 1 / tolerance
(5)
where tolerance = 1-R 2
Tolerance of less than 0.21 or 0.10 and/or VIF of 5 or 10 and above indicates multi-
colinearity of variables. Where multi-colinearity was detected on the basis of the value of
the VIF, the highly collinear variable, that is those with very high VIF, were deleted from
the model.
Finally, a test was conducted to detect any possible serial correlation indicated by the size
of the Durbin-Watson (DW) statistic by establishing that:
μ t = ρμ t -1 + ε t
(6)
Or that the error terms are not correlated.
11.3.2 Model variables
Socio-economic and technical variables that hinder efficiency in trading of agricultural
produce by newly resettled farmers have been analysed. Table 11.3 presents a summary of
these variables, their units of measurements, types, and hypothesized relationships with the
dependent variable.
Search WWH ::




Custom Search