464
fertilizer use by accelerating the release of nutrients into the soil while producing
long-term benefits for sustainable crop production. The adoption of maize-legume
intercropping and rotation also exhibited a significant and positive (complementary)
relationship (32.5%). This may be explained by the fact that smallholder farmers are
advised to follow a two- to three-year adoption cycle of legume diversification in the
maize-based farming system as promoted by the Legume BB project than with crop
rotation at the centre. This implies that the adoption of a particular cropping system
in the current year was dependent on the cropping system practised by the farmer in
the previous season, while the cropping system on the next growing system is conditional on the present cropping system employed by the farmer.
3.3 Determinants of Disadoption of Maize-Legume
and Legume-Legume Intercropping and Rotation
Among the four technologies under consideration, only two of the technologies had
significant number of disadopters to merit regression analysis. These are maizelegume intercropping and rotation. It is worth noting that these are the only practices that are being promoted by the Legume Best Bets project. Thus, bivariate
probit regression model was applied to test whether the two disadoption decisions
are correlated or not. The aim here is to test whether the same factors that encourage
adoption also discourage disadoption of the ISFM technologies. Table 5 presents
the results of the bivariate probit model for the determinants of ISFM technology
disadoption, with particular focus on maize-legume intercropping and rotation.
The results of the bivariate probit regression show that the model is highly significant overall, with the Wald x
2
(20) = 53.57; p = 0.0001. This implies that the
model fits the data reasonably well and the null hypothesis that all regression coefficients are jointly equal to zero is rejected. The Wald test of rho = 0 with
x
2
(1) = 9.82887; p = 0.0017) supports the application of the bivariate probit model
and conclude that the error terms in the two disadoption decisions are correlated.
This is also confirmed by the positive and significant correlation coefficient (rho) of
0.5972 (about 60%) suggesting that a strong complementary relationship exists
between the disadoption of maize-legume intercropping and legume-maize rotation.
In other words, the two technologies are abandoned jointly. This may be so because
of the central role of legumes in the two cropping system, and thus whatever happens to legumes will affect the sustainability of the two systems. This has a critical
policy implication in that efforts to avert the abandonment of one practice will most
probably reduce the likelihood of abandonment of the other.
Five of the explanatory variables were significant in explaining disadoption of
maize-legume intercropping and legume-maize rotation. These are sex of the household head, farming experience, total land size, access to legume seed and access to
extension. An interesting result was found here in that a symmetrical relationship
exists whereby some of the variables that encourage adoption ISFM technologies
J. S. Kanyamuka et al.
fertilizer use by accelerating the release of nutrients into the soil while producing
long-term benefits for sustainable crop production. The adoption of maize-legume
intercropping and rotation also exhibited a significant and positive (complementary)
relationship (32.5%). This may be explained by the fact that smallholder farmers are
advised to follow a two- to three-year adoption cycle of legume diversification in the
maize-based farming system as promoted by the Legume BB project than with crop
rotation at the centre. This implies that the adoption of a particular cropping system
in the current year was dependent on the cropping system practised by the farmer in
the previous season, while the cropping system on the next growing system is conditional on the present cropping system employed by the farmer.
3.3 Determinants of Disadoption of Maize-Legume
and Legume-Legume Intercropping and Rotation
Among the four technologies under consideration, only two of the technologies had
significant number of disadopters to merit regression analysis. These are maizelegume intercropping and rotation. It is worth noting that these are the only practices that are being promoted by the Legume Best Bets project. Thus, bivariate
probit regression model was applied to test whether the two disadoption decisions
are correlated or not. The aim here is to test whether the same factors that encourage
adoption also discourage disadoption of the ISFM technologies. Table 5 presents
the results of the bivariate probit model for the determinants of ISFM technology
disadoption, with particular focus on maize-legume intercropping and rotation.
The results of the bivariate probit regression show that the model is highly significant overall, with the Wald x
2
(20) = 53.57; p = 0.0001. This implies that the
model fits the data reasonably well and the null hypothesis that all regression coefficients are jointly equal to zero is rejected. The Wald test of rho = 0 with
x
2
(1) = 9.82887; p = 0.0017) supports the application of the bivariate probit model
and conclude that the error terms in the two disadoption decisions are correlated.
This is also confirmed by the positive and significant correlation coefficient (rho) of
0.5972 (about 60%) suggesting that a strong complementary relationship exists
between the disadoption of maize-legume intercropping and legume-maize rotation.
In other words, the two technologies are abandoned jointly. This may be so because
of the central role of legumes in the two cropping system, and thus whatever happens to legumes will affect the sustainability of the two systems. This has a critical
policy implication in that efforts to avert the abandonment of one practice will most
probably reduce the likelihood of abandonment of the other.
Five of the explanatory variables were significant in explaining disadoption of
maize-legume intercropping and legume-maize rotation. These are sex of the household head, farming experience, total land size, access to legume seed and access to
extension. An interesting result was found here in that a symmetrical relationship
exists whereby some of the variables that encourage adoption ISFM technologies
J. S. Kanyamuka et al.