West Bangka Regency was chosen as a research site because this area is a producer of Muntok white pepper, a brand name of white pepper prominently exported from the island of Bangka since the Dutch colonial era. Seventy-seven samples of white pepper farmers were randomly selected. The cross-sectional primary data were collected by interviewing directly the farmers in the study sites using a list of questions in the questionnaire previously served by the researcher.

The analytical tool used to analyze factors influencing adaptation of white pepper farmers to climate change is the logit regression model approach. The model is formulated as follows (Gujarati, 2004):

(1)

where:

Pi = the probability that farmer will make the choice of adaptation strategy to climate change

β₀ = intercept

βi = regression coefficient

Xi = the ith of the independent variable

Both sides of equation (1) are multiplied by 1 +e^-zi to obtain:

(2)

If equation (2) is divided by Pi and then subtracted 1, the following equation is obtained:

(3a)

Or could be represented as follow:

(3b)

Equation (3) is transformed into a natural logarithmic model, and then the following equation is obtained:

(4)

Because ln e^zi = Z_i, then equation (4) can be represented as follows:

(5)

Equation 5 is well known as a logit model.

Farmer adaptation strategies to climate change are positioned as qualitative binary dependent variables where the value is 1 if the farmer implemented this strategy and 0 if he/she did not implemented it. In this study those variables were approximated from the probability (Pi) of the five adapted strategies as follows: 1) Farming diversification, 2) mixed cropping, 3) using farming residues as compost, 4) building wells for water harvesting and 5) using organic fertilizer.

The independent variables (Xi) in this study are age, gender of farmers, farming experience, white pepper price, household expenditure, total income of farmers from all his/her farming business, cooperative membership, farmer group membership, and other rural organization membership and educational status. However, there were additional considerations related to the econometrical reason for placing independent variables in the logit regression equation as follows:

Based on the cross-sectional primary data, the value of the white pepper price did not varied. Thus, the price of white pepper cannot be chosen as an independent variable in all logit model equations in this study.

Table 2. Table 2. Parameter estimation of the logit model of adaptation strategy (n = 70) Source: own source ****, ***, **, *

are significant at 1, 5, 10, and 15 % probability level, respectively.

Diversification of farming is adopted to minimize losses due to the farming failure of certain types of cropping in anticipation to extreme climates. Factors affect the diversification of farming can be explained as follows: firstly, the farming total revenues have a statistically significant effect at α = 1 percent on the farming diversification, where the higher revenue from farming they have, the higher the probability to implement on-farm diversification. This evidence is the same as the finding by Asfaw et al. (2017) in Ethiopia that the better livelihood of farmer household, the higher the propensity to diversify their farming.

Secondly, membership in agricultural extension groups has a statistically significant effect on farming diversification at the α = 10 percent. Positive coefficient value shows that farmers who participate in extension groups have a higher probability to apply farming diversification than those farmers who do not participate. Ehiakpor et al. (2016) said that involvement in extension groups enabled farmers to share the technical experience of adaptation strategies to one another, so that their involvement in farmer groups would improve the probability to implement farming diversification. The same findings are indicated by several studies (Asfaw et al., 2017). Meanwhile, the membership in cooperative and other village organizations did not have a statistically significant effect on farming diversification. This phenomenon shows that membership in cooperatives and other village organizations does not increase farmers’ probability to diversify their farming as one of their adaptation strategies to climate change.

Mixed cropping is meant in this study as planting two or more cropping in the same area at the same time with white pepper as the main crop and the others as subsidiaries. This study found that as many as 78.57 percents of white pepper farmers cultivate mixed cropping in research location.

The logit model equation of mixed cropping (table 2) shows only one independent variable with a significant effect on the mixed cropping strategy, i.e., the farmer’s gender. The coefficient value of this binary qualitative dummy variable is negative and statistically significant at α = 15 percent. This means the probability of female farmers to adopt a mixed cropping strategy is higher than in male ones. This can happen according to Singh et al. (2016) because these female farmers are assessed by all their family members who have the character and experience to be trusted and to make the best decisions for their farming business.

Kassam and Friedrich (2012) and Bhaktikul (2012) said that farmers’ strategy to use crop residues as compost can be seen as an adaptation behavior to climate change because this action is categorized as conservation farming, which is one of the ways to adapt one’s farming to climate change. The logit regression equation (table 2) relates to factors that affect the use of crop residues as compost and found these facts: firstly, farming’s total revenues have statistically a significant effect at α = 1 percent on the use of crop residues as compost. This evidence supported the previous study that found that the better the farming revenue they had, the higher the propensity to implement an adapted strategy to climate change (Asfaw et al., 2017), i.e., the use of crop residues as compost.

Secondly, the experience has statistically a negative significant effect on the use of crop residues as compost at the level of α = 10 percent. The fact that all samples in this research cannot explain what climate change means and how to anticipate it, shows their limited information about climate change. Singh et al. (2016) said that farmers who had many experiences, but minimal access to information about climate change tend to be reluctant to implement an adaptation strategy. This is the explanation why farming experience negatively affects the use of crop as compost.

Building wells for water harvesting is a behavior or action that reflects the farmers’ adaptation to climate change (Rejekiningrum et al., 2011) because this action is an implementation of agricultural conservation as a kind of land management technology to improve plant resistance to drought. The results show that 47.14 percent of farmers take this action.

The logit model equation of building wells for water harvesting (table 2) shows only one independent variable that has a significant effect on this strategy, namely, the farmer’s gender. The coefficient number of this binary qualitative dummy variable is negative and is statistically significant at α = 1 percent. This means that the probability of female farmers to build a well for water harvesting is higher than the male ones. This can happen according to Singh et al. (2016) because these female farmers are assessed by all their family members who have the character and experience to be trusted and make the best decision for their farming business, in this case, building wells for water harvesting.

The logit regression equation (table 2) relates to factors affecting the use of organic fertilizer and provides the following facts: firstly, gender has a significant effect on this strategy. The coefficient of this binary qualitative dummy variable is negative and is statistically significant at α = 1 percent. This means the probability of female farmers to adopt the use of organic fertilizer strategy is higher than the male ones. Singh et al. (2016) explained that this phenomenon can happen because female farmers agreed with all their family member to make the decision for choosing adapted strategy regarding the climate change and, especially in this case, using organic fertilizer.

Secondly, membership in the cooperative has a significant effect on using the organic fertilizer at the level of α = 5 percent. The regression coefficient value of this binary qualitative dummy variable is negative, indicating that farmers who are not cooperative members have a higher probability in using organic fertilizer than those farmers who are members.

Thirdly, membership in the farmer group for extension has a significant effect on the use the organic fertilizer at the level of α = 10 percent. The coefficient value of this binary qualitative dummy variable is a positive indication that farmers who are members of the group for an extension have a higher probability to use organic fertilizer than those farmers who are not members.

Fourthly, education is not a determinant of the use of organic fertilizers. The effect of this variable is not significant on the use of organic fertilizer. This means there is no difference in the probability of using organic fertilizer between farmers whose education is high school or above and farmers whose education is below the high school. So does experience, this independent variable is not a determinant that affects the use of organic fertilizer by farmers.