The research was carried out in Ethiopia's Abuna Gindeberat area, in the western Shoa zone of Oromia Region. The district is one of twenty-two (22) rural districts in the western Shoa zone, and is roughly 138 kilometers from Ambo, the zonal town's seat. Bake-Qalaxe, the district town, is 178 kilometers from Addis Ababa. The district is bordered on the north by the Amhara Region, on the west by Gindeberat Districts, on the east by Meta Robi District, and on the south by Jaldu District. Along the boundary with Gindeberat, the district's altitude ranges from 1000 masl to 2640 masl (Abuna Gindeberet district administration office, 2021).
Sampling procedures
A three-stage (Multi-stage) sampling process was used to choose both the study Kebeles and the respondents in order to conduct the study in a representative manner and to increase its reliability and validity. The steps of the sampling technique were as follows:
In the first stage, Abuna Gindeberat from the west Shoa Zone was purposefully chosen based on the district's maize variety production practices. Maize is a high-yielding cereal crop that plays a vital role in the district's food security, but a variety of issues prevents the district from adopting improved maize varieties. As a result, the area is well-known for its food insecurity issues. So this study at the micro level gives clue at wider level. Farmers were chosen through a process that included numerous steps. The district is divided into 44 Kebeles for administrative purposes. According to the district Agricultural Office's report for 2020/2021, out of 44 kebeles, 28 were low adopters and 16 were high adopters of improved maize varieties.
In the second stage, two Kebeles from each group were chosen at random. Then one Kebele was chosen at random from the district's high adopters, and the second Kebele was chosen at random from the district's low adopters.
In the third stage, Farmers in each randomly chosen Kebele were classified into adopters and non-adopters of improved maize varieties. To this end, the district agricultural office of (2020) and development agents at each sampled Kebeles provided lists of adopter families. Using sample size determination formula, total of (143) sampled households were chosen.
There were 49 adopters and 94% non-adopters in the entire sample of (143) household heads. Then, in accordance to their population size, households from each Kebeles were chosen. As a result, out of the (143) sample houses, 77 were chosen from Gute-Andode Kebele, while the remaining (66) were chosen from Oborra Kebele. Finally, a random sampling technique was used to choose a sample of adopters and non-adopters in each Kebeles.
Sample size determination
Using the sample size determination, the sample size for collecting data through a household survey was determined. Yamane (1967:886) offers a simple formula for determining sample sizes (Israel, 1992). At a 95% confidence level, the required sample size was calculated. The following formula was used to calculate sample size for the investigation.
n = \(\frac{\left(N\right)}{1+N\left(e\right)2}\) =\(\frac{\left(621\right)}{1+621\left(0.05\right)2}\) =143
Source: Abuna Gindeberat Agricultural and Natural Resource Office, (2021)
Methods of Data Analysis
Before analysis of explanatory variables by binary logistic regression in SPSS, explanatory variables were examined for multi-co-linearity using co-linearity diagnostics index. The researcher used descriptive, inferential and econometric method to analyze the primary data. Since the data collected were qualitative and quantitative in nature, this research relies on both quantitative and qualitative data analysis.
Model specification
The logit model was used to investigate the factors that influence the adoption of better maize varieties. Adoption is the dependent variable in the logit model, with values of 1 or 0. A farmer with a score of 1 has accepted better maize varieties, while a farmer with a value of 0 has not. Farmers who planted at least one of the improved maize varieties at least for the 2020/2021 cropping season were classified as adopters, whereas those who did not plant the improved varieties in that cropping year were classified as non-adopters.
Y ί = βo + βί + Xί +uί ………………………………………………………… (1)
Y ί = Stands for adoption of improved maize varieties with a value of 1 for adopters and 0 for non-adopters.
X ί = refers to a farmer’s characteristics e.g. age of household head for the ith farmer.
uί = refers to the error term which is an independently distributed random variable with a mean of zero.
A typical linear regression model is Eq. (1). However, because the dependent variable is adoption, which can be either 1 or 0, the use of linear probability models (LPM) is a serious issue. The anticipated value may fall outside of the relevant probability range of 0 to 1. As a result, the logit model was employed to solve the problem with the linear probability model. As a result, for adopters, the logistic cumulative probability function is:
Pί= \(\frac{1}{1+{e}^{-Z}}\) = \(\frac{1}{1+{e}^{Z}}\) ……………………………………………. (2)
P ί = is the probability farmer adopted the new varieties and that Pi is nonlinearly related to Zί (i.e. and Xί)
Z ί = βo + β₁X₁+……… βnXn and e represents base of natural logarithms
Then, (1-p) probability of non-adopter of improved maize varieties is described as:
1- Pί= \(\frac{1}{1+{e}^{Z}}\) ……………………………………………………………. (3)
By dividing Eq. 2 by Eq. 3, the odds ratio in favor of adopting the improved variety was obtained as;
\(\frac{Pί}{1- Pί}\) = \(\frac{{e}^{Z/(1+{e}^{Z)}}}{1/1+{e}^{Z}}\) =\({e}^{Z}\) …………………………………………………… (4)
To estimate the logit model, the dependent variable was transformed by taking the natural log of
Equation 4 as follows:
Lί= ɭ n ( \(\frac{Pί}{1- Pί}\)) = Zί = βo + β₁X₁+……… βnXn………………………… (5) Where,
The log of the odds ratio is L. The logit probability model is denoted by the letter L. It should be emphasized that Eq. 5's logistic model is based on the logit of Z, which is the stimulus index. This proves that when Z goes from 0 to +, Pi stays between 0 and 1. According to the literature on adoption, a farmer's decision to embrace agricultural technology is influenced by socioeconomic, institutional, and environmental factors (Feder et al., 1985). However, in adoption research, no economic theory governs the selection of specific independent variables. As a result, the outcomes differ depending on the situation. As a result, in this Logit Mode, the explanatory variables considered are; Yί = βo + β1AGHH + β2SEXHH + β3FSZ + β4EDUHH + β5TLU + β6FARMSZ + β7FARMEXP + β8OFINC + β9FERTAV + β10IMSAV + β11CREDIT + β12EXCONTACT + β13DEMON + β14TRAINING + β15PERCEP + ε. Where: Yί: stands for the log odds of adoption for the ίth farmer; the explanatory variables are briefed in Table 7 and ε is error term.