Farmers' Perceptions and Adoption of New Agricultural Technology: Evidence From Analysis in Burkina Paso and Guinea, West Mrica
Farmers' Perceptions and Adoption of New Agricultural Technology: Evidence From Analysis in Burkina Paso and Guinea, West Mrica
Farmers' Perceptions and Adoption of New Agricultural Technology: Evidence From Analysis in Burkina Paso and Guinea, West Mrica
ECONOMICS
ELSEVIER Agricultural Economics 13 (1995) 1-9
a West Africa Rice Development Association, OJ BP 2551, Bouake OJ, Ivory Coast
b International Crops Research Institute for the Semi-Arid Tropics, Niamey, Niger
Accepted 14 May 1995
Abstract
Economists investigating consumer demand have accumulated considerable evidence showing that consumers
generally have subjective preferences for characteristics of products and that their demand for products is
significantly affected by their perceptions of the product's attributes. However, the role of farmers' preferences in
adoption decisions have received very limited attention in adoption studies conducted by economists. This paper
tests the hypothesis that farmers' perceptions of technology characteristics significantly affect their adoption
decisions. The analysis, conducted with Tobit models of modern sorghum and rice varietal technologies in Burkina
Paso and Guinea, respectively, strongly supports this hypothesis. Our results provide a strong case for future
adoption studies to expand the range of variables used away from the broad socio-economic, demographic and
institutional factors to include farmers' subjective perceptions of the characteristics of new agricultural technologies.
subjective preferences for characteristics of prod- stable yield performance under both traditional
ucts and that their demand for products is signifi- and improved soil fertility management condi-
cantly affected by their perceptions of the prod- tions and good grain quality, especially for mak-
uct's attributes (Jones, 1989; Lin and Milon, 1993). ing 'to', a very popular local porridge across the
For example, using a double-hurdle model, Lin Sahel (ICRISAT, 1980-1983). As striga-a para-
and Milon (1993) found that commodity at- sitic weed-has been a major problem facing
tributes and consumers' safety perceptions were sorghum (and, indeed, cereal) farmers in West
significant in explaining decisions to consume and Africa, a complementary program on screening
the frequency of consumption of shellfish in the for striga-resistant varieties formed a major part
USA Similarly, Jones (1989) found, using Cragg's of ICRISAT's research program (Ramaiah, 1991).
double-hurdle framework, that consumers' sub- A number of improved varieties was developed
jective perceptions influenced cigarette smoking and tested in on-farm tests in Burkina Faso, e.g.
decisions. Economists investigating the adoption ICSV 1049 BF, ICSV 1002 BF, etc. While farm
of new agricultural technology have lagged be- tests have been conducted on these varieties in
hind in investigating how farmers' (the consumers eastern Burkina Faso (Matlon, 1985; Wildschut
of agricultural technologies) subjective percep- and Ouattara, 1987), little is known about the
tions of technology characteristics affect their factors affecting the adoption decisions of farm-
adoption decisions. In extensive reviews of adop- ers.
tion studies in developing countries (Feder et al., Research on the development of mangrove
1985; Feder and Umali, 1993), no study was found rice in West Africa is carried out by the West
to have analyzed the direct effects of farmers' Africa Rice Development Association (WARDA).
subjective assessments of agricultural technology Starting in 1976, WARDA maintained a regional
characteristics on adoption decisions. In their re- rice improvement program in Sierra Leone, tar-
cent study, Adesina and Zinnah (1993a) found geting the approximately 200 000 ha of mangrove
that farmers' perceptions of the characteristics of swamps cultivated to rice in West Africa. The
modern rice varieties significantly affected adop- mangrove rice ecologies, located on tidal estuar-
tion decisions in Sierra Leone. The paucity of ies that are near the ocean, are important rice
empirical economic studies on this issue justifies growing environments for six countries in West
further investigation. The objective of this paper Africa: Guinea, Guinea Bissau, Senegal, Gambia,
is to determine the effects of farmers' percep- Sierra Leone and Nigeria. In Guinea, about
tions of agricultural technology characteristics on 64 000 ha of mangrove rice area is under cultiva-
adoption decisions, using comparative analysis of tion, representing 12% and 18% of national rice
two technologies in West Africa (i.e. improved area and rice production, respectively. Seeds of
sorghum varieties in Burkina Faso, and improved the modern varieties (e.g. ROK-5, ROK-10, Ku-
mangrove rice varieties in Guinea). atik Kundur) were recently diffused to neighbor-
A historical background of the development of ing Guinea (Zinnah et al., 1993). One of the
the new agricultural technologies examined in varieties, ROK-5, has been increasingly adopted.
this paper is briefly discussed below. Sorghum In the Coyah region (the country's major man-
improvement activities in West Africa have been grove swamp rice growing zone), it was estimated
a major focus of research conducted by the Inter- that the percentage of farmers cultivating ROK-5
national Crops Research Institute for the Semi- increased from 1% in 1989 to 15% in 1990
Arid Tropics (ICRISAT) which has the global (Adesina and Zinnah, 1993b). On-farm tests con-
mandate for sorghum improvement. ICRISAT's ducted from 1982 to 1985 showed that these
sorghum improvement work started in Burkina modern mangrove rice varieties consistently out-
Faso in 1975/1976. Most of the initial work was yielded the best local varieties by more than 30%
on the screening of introduced and local (Agyen-Sampong, 1990). Little is known, how-
germplasm. Breeding started much later. Among ever, about the determinants of farmers' adop-
the principal breeding objectives were: good and tion of these varieties.
AA. Adesina, J. Baidu-Forson /Agricultural Economics 13 (1995) 1-9 3
The rest of the paper is divided into three > 0). Otherwise, if bmi- bli ~ 0 (or V; * ~ 0) no
sections. Section 2 describes the conceptual and adoption is observed (i.e. V; = 0). The model per-
empirical models, while Section 3 discusses the mits the investigation of the decision of whether
empirical results. Section 4 presents conclusions or not to adopt and the conditional level of use of
and implications of the results for adoption stud- the technology if the initial adoption decision was
ies. made. The Tobit model also has an advantage in
that its coefficients can be further disaggregated
to determine the effect of a change in the ith
2. Methods variable on changes in the probability of adopting
the modern variety and the expected use intensity
A given agricultural technology embodies a of the variety. Following McDonald and Moffitt
number of important characteristics that may in- (1980), it can be shown that
fluence adoption decisions. The observed adop- E(V;) =<I>(z)E(v;') (2.1)
tion choice on an agricultural technology (e.g.
modern crop varieties) is hypothesized to be the where E(v/) is the expected value of V; for
end result of a complex set of inter-technology those farmers that have already made the adop-
preference comparisons made by farmers. Let the tion decision, and <I> is the cumulative normal
perceived benefits derived from growing a mod- distribution function at z, where z is X/3 I a.
ern variety and the local variety be represented Differentiating with respect to any element of
by b(m) and b(l), respectively. Let the ith X gives
farmer's perceptions of the varietal-specific char- aE( V;) ;ax;= <I>( z) {aE( v;') ;ax;}
acteristics of the modern and local rice varieties
be given by Pim and Pi!, respectively. Let the +E(v;"){a<I>(z);ax;} (2.2)
other socio-economic and demographic charac- Multiplying through by XJE(V;), the relation
teristics of farm households affecting the adop- can be converted into elasticity forms
tion decision be denoted as ci. For simplicity, let
these relations be represented as bmi = q(pim' {aE( V;) ;ax;}XJE( V;)
pil;c) and bli = w(pim,Pi!;c), respectively. To =<I>( z) {aE( v;') jaX;}XJE( V;)
model this effect of perceptions of technology
attributes on adoption decisions a Tobit model is + E( v;' ){a<I>( z) ;ax;}XJE( V;) (2.3)
used. This is represented below using an index Re-arranging this using Eq. (2.1) gives
function approach
{aE( V;) ;ax;}XJE( V;)
V; • = {3TXi + E; (1.1)
= {aE( v;') ;ax;}XJE( v;')
V;=O if V;* ~0 (1.2)
+ {a<I>( z) ;ax;}XJ<I>( z) (2.4)
V;=V;* if V;*>O (1.3) The total elasticity consists of two effects: (1)
where V;, a limited dependent variable, is the the change in the probability of the expected
perceived benefit of adoption of the modern vari- level of use intensity of the modern varieties for
ety; V; * is an underlying latent variable that those farmers that are already adopters, and (2)
indexes adoption; X is the vector of socio-eco- the change in the elasticity of the probability of
nomic and demographic characteristics of the being an adopter.
farm household, and the technology perceptions The empirical models are discussed below. For
of the farmer; f3T is a vector of parameters to be the sorghum model, the dependent variable is the
estimated; and E; is an error term. When the share of the total sorghum area that is cultivated
inter-varietal attribute preference comparisons in improved varieties (ISSHARE), while the de-
are such that bmi- bli > 0 (i.e. V; * > 0) the farmer pendent variables are, respectively, farmers' sub-
is observed to adopt the modern variety (V; = V; * jective assessments of: yield performance
4 A.A. Adesina, J. Baidu-Forson /Agricultural Economics 13 (1995) 1-9
(YIELD), the quality of the local porridge (to) of these variables allows us to compare our re-
that farmers make from sorghum (TO), perfor- sults with those obtained earlier by Adesina and
mance under poor soil conditions (SOIL), toler- Zinnah (1993a) in Sierra Leone. Based on these
ance to striga weed infestation (STRIGA), and results, it is hypothesized that these factors are
drought tolerance (DROUGHT). It is hypothe- positively related to adoption decisions in Guinea.
sized that these factors are positively related to Ease of cooking is likely to be important to these
adoption decisions. In addition to these factors, rice farmers because of the scarcity of fuel-wood
the following socio-economic and demographic for preparing food in the mangrove areas. Farm-
variables were included as explanatory variables ers are obliged to travel several kilometers to
in the model: the age of the farmer (AGE), search for fuel-wood. Ease of cooking of rice
whether or not the farmer was a village head and varieties is linked to the gelatinization tempera-
the number of years since becoming village head ture of the rice. Varieties with high gelatinization
(HEAD), participation of the farmer in on-farm temperature take longer periods to cook (Un-
tests of the improved sorghum varieties (TEST), nevehr, 1986). This increases the demand for
number of times the farmer was visited by exten- fuel-wood, a very scarce input in the mangrove
sion agents in the survey year (VISIT), and environments where most of the mangrove trees
the number of times the farmer participated in have been cut down by farmers. Thus, farmers
extension related workshop training activities may tend to adopt varieties that puts less pres-
(WKSHOP). It is hypothesized that TEST and sure on the scarce fuel-wood supply. Tillering
AGE are positively related to adoption decisions. capacity is also very critical in the mangrove
Testing the varieties in on-farm tests can be ex- ecology because of the high damage to young rice
pected to influence adoption decisions. The ex- seedlings from crabs. Varieties that have the ca-
pected sign on AGE is an empirical question: it pacity to produce tillers fast, and which repro-
may be that older farmers have more experience duce secondary tillers following initial crab dam-
in cultivation and are better able to assess the age, are likely to be adopted by farmers. The high
characteristics of modern technology than labor demands for manual threshing in this farm-
younger farmers. However, it could also be that ing system, where labor is the most limiting fac-
older farmers are more risk averse than younger tor, may create incentives for farmers to adopt
farmers and have a lesser likelihood of adopting varieties that are easier to thresh. Finally, better
new technologies. There is no agreement in the yield performance over local varieties is expected
adoption literature on this as the direction of the to positively affect adoption decisions. The other
effect is generally location or technology specific socio-economic and demographic factors of the
(Kebede et al., 1990; Polson and Spencer, 1991). farmer in the model are: age of the farmer (AGE),
Based on the innovation-diffusion literature, it is the farm size (FSIZE), participation in on-farm
hypothesized that extension visits (VISIT) and tests (TEST), visit by extension agents (VISIT),
participation in workshops (WKSHOP) are posi- and years of experience in mangrove rice farming
tively related to adoption by exposing farmers to (EXP). The expected signs of these variables fol-
'new information'. low that discussed above for the Burkina Faso
For the Guinea model, the dependent variable case. In addition, it is expected that EXP is
is the share of the total mangrove swamp rice positively related to adoption as more experi-
area that is cultivated in improved varieties enced farmers may be able to better assess new
(RSHARE). Two classes of explanatory variables technologies.
are included in the model. The first set (i.e. In each model, the subjective perception vari-
technology perceptions) are farmers' subjective ables were defined as dichotomous variables that
assessments of the following characteristics of the take on the value of 1 if the farmer considered
rice varieties: yield performance (YIELD), ease that the new agricultural technology is better
of cooking (COOK), tillering capacity (TILLER), than the local technology for a given technology
and ease of threshing (THRESH). The inclusion attribute, and a value of 0, otherwise. The empiri-
A.A. Adesina, J. Baidu-Forson /Agricultural Economics 13 (1995) 1-9 5
cal models were estimated using an iterative max- tion of the improved sorghum varieties. Signifi-
imum likelihood algorithm (White, 1978) in order cant (at the 5% level) were quality for making
to obtain asymptotically efficient parameter esti- sorghum paste (TO) and performance under poor
mates. The data for the Burkina Faso study were soil conditions (SOIL). The coefficient on yield
collected in 1989/1990 from a sample of 57 farm- performance (YIELD) is barely significant at
ers in the east central region. Data for the Guinea 10%. Farmers' assessment of the relative toler-
study were collected in 1990-1991 from a sample ance of the improved varieties to striga
of 110 farmers in the Coyah region. (STRIGA), though positively related to adoption
decisions, was not significant. The negative sign
on the drought tolerance variable (DROUGHT)
3. Results is unexpected and may be due to negative corre-
lation between this variable and other varietal
The results for the sorghum model in Burkina characteristics not included in the model. Two
Faso are given in Table 1. The analyses show that farmer characteristics were significant in explain-
farmers' perceptions of four varietal technology ing adoption decisions: the age of the farmer
characteristics were positively related to the (AGE) and farmers' participation in on-farm tests
probability of adoption and intensity of cultiva- (TEST). The latter variable was the most highly
Table 1
Estimated tobit model for factors affecting the adoption of modern sorghum varieties in Burkina Faso, 1989/1990
Normalized Standard T-ratio Elasticity of
coefficient error Adoption Expected level
index of use intensity
Intercept -5.8974 1.7202 -3.4283 •••
Perception of yield 0.9635 0.5820 1.6555 • 1.2401 0.5242
(YIELD)
Perception of quality of 1.8214 0.7268 2.5058 •• 0.8930 0.3775
local sorghum porridge (TO)
Perception of adaptability 1.5951 0.7103 2.2456 •• 2.9328 1.2397
to poor soils (SOIL)
Perception of tolerance 0.2679 0.4750 0.5642 0.6405 0.2707
to striga weed (STRIGA)
Perception of drought -2.1313 0.7477 -2.8506 •• -2.8681 -1.2123
tolerance (DROUGHT)
Age of the farmer (AGE) 0.0613 0.0325 1.8850 • 20.0953 8.4941
Village head (HEAD) -0.0326 0.0299 -1.0875 -4.1800 -1.7669
Participation in 1.2103 0.2578 4.6944 ••• 10.4590 4.4209
on-farm tests (TEST)
Number of times farmer 0.0064 0.0753 0.0855 0.0288 0.0122
was visited by extension
agents (VISIT)
Number of times farmer -0.3342 0.1376 -2.4281 •• -2.8681 -1.2123
participated in extension-
related workshops
(WKSHOP)
The parameters were evaluated using one-tailed t-tests. Asterisks indicate significance at: * • • 1%; * * 5%; • 10%.Dependent
variable, ISSHARE.
Log-likelihood function, 9.8697.
Mean square error, 0.8218E-03.
Mean error, 0.9482E-03.
Squared correlation between observed and expected values, 0.5183.
6 A.A. Adesina, f. Baidu-Forson I Agricultural Economics 13 (1995) 1-9
significant of all the explanatory variables. This is those that are adopted. Secondly, the analyses
not unexpected, because extensive sorghum on- show that sorghum technology development ef-
farm trials were carried out in the survey area. forts should not only concentrate on agronomic
The negative sign on WKSHOP is unexpected characteristics. The strong significance of the food
and may be explained by the possibility that the quality variable indicates that farmers are con-
workshops farmers attended were not specifically cerned about non-agronomic characteristics. As
on sorghum. Some authors have also argued that sorghum is principally used in West Mrica for
what is important is not this contact with exten- making to, research efforts need to be focused on
sion but how farmers assess the relevance of the screening for lines with appropriate good grain
issues discussed at such farmer workshops for and food quality for making this local porridge.
their actual production decisions (Zinnah et al., One way to ensure this is to have farmers, espe-
1993). cially women, assess the food quality of all
These results have implications for sorghum sorghum varieties that are proposed for on-farm
technology improvement research in West Africa. testing in West Mrica.
First, it shows that yield performance of sorghum Results for the Guinea model are given in
technologies needs to be evaluated by farmers Table 2. The results show that all the four tech-
under their own soil conditions. The results nology-characteristic perceptions were strongly
showed that varieties that farmers judge as having significant in explaining farmers' adoption deci-
superior yield performance over local varieties sions. These results are very similar to those
under on-farm soil fertility conditions tend to be obtained by Adesina and Zinnah (1993a) in Sierra
Table 2
Estimated tobit model for factors affecting the adoption of modern mangrove rice varieties in Guinea, 1990/1991
Normalized Standard T-ratio Elasticity of
coefficient error --------------------
Adoption Expected level
index of use intensity
Intercept -0.9534 0.5395 -1.7671 *
Perception of yield 0.7610 0.2875 2.6469 * * * 0.2676 0.2491
(YIELD)
Perception of ease of 0.7782 0.3146 2.4735 * * 0.3741 0.3482
cooking (COOK)
Perception of tillering 1.0347 0.2934 3.526 * * * 0.3805 0.3542
capacity (TILLER)
Perception of ease of 0.6621 0.2838 2.3324 * * 0.2542 0.2366
threshing (THRESH)
Age of the farmer (AGE) -0.0009 0.0116 -0.0771 -0.0604 -0.0563
Farm size (FSIZE) -0.1092 0.1166 -0.9369 -0.2171 -0.2021
Participation in on-farm 0.2221 0.3109 0.7142 0.0382 0.356
tests (ON-FARM)
Number of times farmer 0.0175 0.0600 0.2926 0.0061 0.0056
was visited by extension
agents (VISIT)
Years of experience 0.0115 0.0110 1.0399 0.4141 0.3854
in rice farming
(EXP)
The parameters were evaluated using one-tailed t-tests. Asterisks indicate significant at: * * * 1%; * * 5%; * 10%.
Dependent variable, RSHARE.
Log-likelihood function, -24.985.
Mean square error, 0.3353E-Ol.
Mean error, - 0.1382E-Ol.
Squared correlation between observed and expected values, 0.502.
A.A. Adesina, J. Baidu-Forson I Agricultural Economics 13 (1995) 1-9 7
Leone. It is also important to note that a non- new agricultural technologies are very important
agronomic trait (ease of cooking) significantly af- determinants of adoption behavior. The results
fected farmers' adoption decisions. When taken are robust in that this conclusion is the same for
together with the results for sorghum in Burkina the two different technologies examined in Burk-
Faso, it is clear that farmers consider factors ina Faso and Guinea. The results agree with the
related to food quality as very critical in adoption evidence provided by studies of consumer de-
decisions. This suggests that the conventional ap- mand showing that consumers critically evaluate
proach of focusing technology development characteristics of a product and that demand is
mostly on enhanced agronomic characteristics affected by consumers' subjective assessments of
needs to be changed to also significantly encom- product attributes (Jones, 1989; Lin and Milon,
pass food quality characteristics. 1993). Farmers are consumers of the products of
The Guinea results show that none of the agricultural research and their subjective prefer-
so-called 'contact' variables were significant in ences for characteristics of new agricultural tech-
influencing adoption decisions. The lack of signif- nologies affect their adoption decisions.
icance of the extension variable supports the find- Our results, which lend support to the earlier
ing by other researchers that researcher-exten- finding of Adesina and Zinnah (1993a), call for
sion-farmer linkages are extremely weak in more theoretical and empirical studies to exam-
Guinea and that most of the mangrove varieties ine the effects of farmers' subjective preferences
currently grown by farmers were obtained via of technology characteristics on adoption deci-
farmer-farmer contacts (Zinnah et al., 1993). sions. An advantage of the approach we have
Comparative analyses of the results in Guinea used is that it is possible to identify the key
and Burkina Faso with the results earlier re- characteristics of technologies that need to be
ported for Sierra Leone by Adesina and Zinnah targeted for improvement if adoption is to be
(1993a) show some important patterns in the role achieved. Future studies in this area should over-
of extension and other farmer contact variables in come a limitation of the current analyses: the
influencing farmers' adoption decisions. In the data we had did not permit us to measure the
case of mangrove rice varieties in Guinea and intensity of the perceptions, as was done by Lin
Sierra Leone, extension had very little to do with and Milon (1993) in their study of consumer
technology diffusion: this occurred mainly through demand in the USA. We suggest that future
farmers self-experimentation, evaluation, ex- studies collect data that measure not only the
change and transfer. In the case of sorghum in dichotomous comparison of preferences for tech-
Burkina Faso, extensive on-farm testing of the nology attributes but also the intensity of those
varieties in a number of test-villages appears to preferences. Another possible limitation of the
have affected adoption decisions. This suggests current analyses is the possibility of estimation
that extension workers do not influence technol- bias from simultaneity if the perception variables
ogy transfer in all cases. Farmers are also impor- and the other 'contact' variables (i.e. contact with
tant as sources of technology information and extension, participation in on-farm tests etc.) are
agents of technology transfer. When farmers as- endogenous. It was not possible for us to examine
sess the characteristics of new technologies and this issue owing to the evidently formidable
find them to match their preferences, they often econometric problems associated with estimating
give the technologies to other farmers to test and simultaneous equation systems with dummy or
evaluate thereby setting into motion an endoge- other limited dependent variables. Also impor-
nous process of technology diffusion. tant is the need for studies to examine non-adop-
tion of new agricultural technologies and how
farmers' subjective preferences for technology at-
4. Conclusions
tributes play a role (Yapa and Mayfield, 1978;
The analyses in this paper showed that farm- Nowak, 1992). Nonetheless, our results provide a
ers' subjective preferences for characteristics of strong case for future adoption studies to expand
8 A.A. Adesina, J. Baidu-Forson I Agricultural Economics 13 (1995) 1-9
the range of variables used away from the broad Tropics, 1980. Annual Report Cooperative Program/
socio-economic, demographic and institutional Upper Volta. ICRISAT, Upper Volta.
International Crops Research Institute for the Semi-Arid
factors (Feder et al., 1985; Feder and Umali, Tropics, 1981. Annual Report Cooperative Program/
1993), to include farmers' subjective perceptions Upper Volta. ICRISAT, Upper Volta.
of the characteristics of new agricultural tech- International Crops Research Institute for the Semi-Arid
nologies. Tropics, 1982. Annual Report Cooperative Program;
Upper Volta. ICRISAT, Upper Volta.
International Crops Research Institute for the Semi-Arid
Tropics, 1983. Annual Report Cooperative Program/
Acknowledgments Upper Volta. ICRISAT, Upper Volta.
Jones, A.M., 1989. A double-hurdle model of cigarette con-
sumption. J. Appl. Econometrics, 4: 23-29.
The authors are grateful to two anonymous
Kebede, Y., Gunjal, K. and Coffin, G., 1990. Adoption of new
reviewers of this journal for providing us with technologies in Ethiopian agriculture: the case of
very helpful comments. We are also grateful to Teguelet-Bulga District, Shoa Province. Agric. Econ., 4:
Peter Matlon, Kofi Debrah and Timothy 0. 27-43.
Williams for helpful comments and discussions Kivlin, J.E. and Fliegel, F.C., 1966. Attributes of innovations
as factors in diffusion. Am. J. Sociol., 72: 235-248.
on earlier versions of the paper. Responsibility Kivlin, J.E. and Fliegel, F.C., 1967. Differential perceptions of
for errors remains with us. Part of this study was innovations and rate of adoption. Rural Sociol., 32: 78-91.
conducted while the first author was a Rocke- Lin, C.T.J. and Milon, J.W., 1993. Attribute and safety per-
feller Social Science Research Fellow and Assis- ceptions in a double-hurdle model of shellfish consump-
tant Principal Economist at the International tion. Am. J. Agric. Econ., 75: 724-729.
Matlon, P.J., 1985. A critical review of objectives, methods,
Crops Research Institute for the Semi-Arid Trop-
and progress to date in sorghum and millet improvement:
ics, while based at the West Africa Sorghum a case study of ICRISAT/Burkina Faso. In: H.W. Ohm
Improvement Program in Mali. The funding of and J.G. Nagy (Editors), Appropriate Technologies for
The Rockefeller Foundation for his work is grate- Farmers in Semi-Arid West Africa. Purdue University,
fully acknowledged. The assistance provided by Lafayette, IN, pp. 154-179.
McDonald, J.F. and Moffitt, R.A., 1980. The uses of Tobit
our colleagues at INERA and CRPA/ Koupela analysis. Rev. Econ. Stat., 62: 318-321.
in Burkina Faso throughout the field work is Nowak, P., 1992. Why farmers adopt production technology.
gratefully acknowledged. J. Soil Water Conserv., 47: 14-16.
O'Mara, G., 1980. The micro-economics of technique adop-
tion by small-holding Mexican farmers. World Bank De-
velopment Research Center, Washington DC.
References Polson, R.A. and Spencer, D.S.C., 1991. The technology
adoption process in subsistence agriculture: The case of
Adesina, A.A. and Zinnah, M.M., 1993a. Technology charac- cassava in South Western Nigeria. Agric. Syst., 36: 65-77.
teristics, farmer perceptions and adoption decisions: a Ramaiah, K.V., 1991. Breeding for striga resistance in sorghum
tobit model application in Sierra Leone. Agric. Econ., 9: and millet. In: S.K. Kim (Editor), Combating Striga in
297-311. Africa. International Institute of Tropical Agriculture,
Adesina, A.A. and Zinnah, M.M., 1993b. Impact of modern Ibadan, Nigeria, pp. 75-80.
mangrove swamp rice varieties in Sierra Leone and Guinea. Shakya, P.B. and Flinn, J.C., 1985. Adoption of modern
Int. Rice Res. Notes, 18: 36. varieties and fertilizer use on rice in the eastern Tarai of
Agyen-Sampong, M., 1990. History of rice research in Rokupr: Nepal. J. Agric. Econ., 36: 409-419.
Evaluation and overview of WARDA's mangrove rice pro- Strauss, J., Babosa, M., Teixeira, S., Thomas, D. and Gomes,
gram, and relationship to national program. West Africa R., 1991. Role of education in adoption of technology: a
Rice Development Association, Rokupr, Sierra Leone. study of upland rice and soybean farmers in central-west
Feder, G. and Umali, D., 1993. The adoption of agricultural Brazil. Agric. Econ., 5: 341-359.
innovations. A review. Techno!. Forecast. Soc. Change, 43: Unnevehr, L.J., 1986. Consumer demand for rice grain quality
215-239. and returns to research for quality improvement in south-
Feder, G., Just, R.E. and Zilberman, D., 1985. Adoption of east Asia. Am. J. Agric. Econ., 68: 634-641.
agricultural innovations in developing countries: a survey. White, K.J., 1978. SHAZAM-A general computer program
Econ. Dev. Cult. Change, 33: 255-297. for econometric methods. Econometrica, 46: 239-240.
International Crops Research Institute for the Semi-Arid Wildschut, J. and Ouattara, A., 1987. Rapport Technique.
A.A. Adesina, I. Baidu-Forson j Agricultural Economics 13 (1995) 1-9 9
Resultats de Ia campagne 1987. O.R.D. du Centre-Est, Zinnah, M.M., Compton, J.L. and Adesina, A.A., 1993. Re-
Burkina Faso. search-extension-farmer linkages within the context of
Yapa, L.S. and Mayfield, R.C., 1978. Non-adoption of agricul- the generation, transfer and adoption of improved man-
tural innovations: evidence from discriminant analysis. grove swamp rice technology in West Africa. Q. J. Int.
Econ. Geogr., 54: 145-156. Agric., 32: 201-211.