Analysis of factors affecting strawberry farming (case: Dolat Rayat District, Karo Regency)

This study aims to analyse the factors affecting strawberry farming in Dolat Rayat District. The analytical method used in finishing the study is the multiple linear regression model through the Cobb Douglas production function. The variables used in this study were land area, use of seedlings, use of NPK fertilizer, total labours and use of Joker pesticide. The results obtained indicate that land area, use of seedlings, use of NPK fertilizer, total labours and use of Joker pesticide together have a significant effect on strawberry production with a coefficient of determination of 0.699. Partially, land area, labours and the use of Joker pesticide did significantly influence the strawberry production. Meanwhile, the use of seedlings and the use of NPK fertilizer did not significantly affect the strawberry production.


Introduction
Strawberries are one of the most important fruit commodities in the world, especially for countries with subtropical climates. Consumer demand for strawberries tends to increase from year to year. T he higher market absorption rate reflects that the strawberry agribusiness has bright prospects in the future. Based on Agricultural Statistics in 2017, strawberry production in Indonesia has decreased every year. Strawberry production reached 169,796 tons in 2012, 90,352 tons in 2013, 58,882 tons in 2014, 31,798 tons in 2015 and 12,091 tons in 2016 [1].
The high demand for strawberries is inversely proportional to the lower production of strawberries, thereby increasing the import of strawberries to Indonesia. To prevent the increasing in strawberry imports, some efforts should be made to increase the production of strawberries in Indonesia. The authors will analyse factors affecting the production of strawberry farming.

Research methods
To analyse the effect of several factors on the farmers' strawberry farm production in the s tudy area, multiple linear regression models were used, where previously classical assumptions were tested in order to meet to the requirements of the econometric criteria such as normality, multicollinearity and heteroscedasticity tests.

Multiple linear regression model
To determine the factors that affecting the production of strawberries, the model used is a multiple linear model through the Cobb-Douglas production function. The formula of the production func tion in the linear model mathematically is as follows: In order to estimate the parameters used, which were initially in multiple linear model, the for mula must be transformed first into a natural double logarithm (ln) by using the ordinary least square method as formulated as follows: looking at the graph plot between the predicted values of endogenous variables, namely Y: ZPRED with the residual X: SRESID. With the following test criteria: 1. If a certain pattern is found, such as the existing points forming a certain regular pattern (widening, wavy, and narrowing): heteroscedasticity occurs. 2. If no clear pattern is found and all the points are scattered below and above the number 0 on the Y axis: there is no heteroscedasticity.

Normality test.
The use of the normality test is to determine whether the data distribution approaches or follows the normal distribution. In principle, normality can be discovered by looking at the data distributions or the points on the diagonal axis of the graph or by looking at the his togram of the residual. If the data is spreading around the diagonal line or histogram graph, it means the regression model has met the assumption of normality [2].

Model compatibility test
is a statistical value calculated from the sample data. This coefficient indicates the percentage variation of all dependent variables that c an be identified by looking at the changes in independent variables (explanatory variables). The coefficient is a measure of the extent to how far an independent variable can change a dependent variable in a relationship [3].

Simultaneous influence of variable test (F test).
The F test is a test that is used to show the simultaneous influence of the related variables in accordance with all independent variables used in the model. With the following criteria: a. If calculated F < F statistics or if the significance of F > α: H0 is accepted. b. If calculated F > F statistics or if the significance of F < α: H1 is accepted Information: H0 = variables of land area, labours, seedlings, NPK fertilizer and Joker pesticide simultaneously have no real effect on strawberry production H1 = variables of land area, labours, seedlings, NPK fertilizer and Joker pesticide simultaneously have a real effect on strawberry production

Partial variable influence test (t test).
The T test is a test used to show how far the influenc e of an independent variable is partially in explaining the variation of dependent variables. With the following test criteria: a. If calculated t < t statistics or if the significance of t > α: H0 is accepted. b. If calculated t > t statistics or if the significance of t < α: H1 is accepted Information: H0 = variables of land area, labours, seedlings, NPK fertilizer and Joker pesticide simultaneously have no real effect on strawberry production H1 = variables of land area, labours, seedlings, NPK fertilizer and Joker pesticide simultaneously have a real effect on strawberry production

Results and discussion
The number of samples used in this study is 30 samples of strawberry farmers. This is directly proportional with Agung Statement said that the central limit theorem has been applied to a sample size of at least 30. Even stated for a sample size greater than 20, the normal distribution has been us ed to approach the binomial distribution [4]. The production function used to look at the factors that influence strawberry farming is the Cobb-Douglass function which consists of dependent variables and independent variables. The dependent variables are strawberry production (Y) and its independent variables are land area (X1), seedlings (X2), NPK fertilizer (X3), labours (X4) and Joker pesticide (X5). Based on the analysis that has been done, that the result of the Kolmogorov-Smirnov test is obtained the significance value of 0.200 > 0.05 (α). It can be deduced that the data used are distributed normally. This is also in accordance w ith the approximation graph, the display of P-P normal plot of regression standardized residual of data is s aid to be normally distributed plots depicted data spread or docked to the diagonal line.

Figure 1. Normal P-P plot of regression standardized residual
Multicollinearity test is used to see whether there is a great linear correlation between some or all of the variables that included in the regression model. The existence of multicollinearity can be identified from the VIF (Variance Inflection Factor) value or the tolerance value. Multicollinearity does not occur when the VIF value is smaller than 10 and also the tolerance value is larger than 0.1. The VIF values obtained are 2.311, 4.003, 1.865, 1.662 and 2.695. While the tolerance values obtained were 0.433, 0.250, 0.536, 0.602, and 0.371 respectively. This indicates that the data used has met the requirements of the multicollinearity test. This clearly proves that there is no multicollinearity in the regression model. The heteroscedasticity test can also be done by looking at the distribution pattern of the dots on the diagram scatterplot. Based on Figure 2, it can be seen the dots are spreading out without forming any certain clear pattern. Therefore, it can be concluded that there are no s ymptoms of heteroscedasticity.

Figure 2. Scatterplot
The R 2 value is 0.699 based on the analysis of the coefficient of determination (R 2 ) has been done. It means variations in bean production can be explained by variations in production factors of 69.9%. This means that the ability of the independent variables to be included in the production function model, namely land area, seedlings, NPK fertilizer, labours and Joker pesticide has a considerable Based on the results of the F test analysis, the calculated F (11.141) > F table (3.75) and a significant value of 0.000, at the 95% confidence level and error rate α = 5% so that the use of production factors of land area, seedlings, NPK fertilizer, labours and Joker pesticide together have a significant effect on strawberry production. Based on Table 1, the results of the analysis that influence the amount of use of chemical Joker pesticide are included in the equation for the use of Cobb Douglas production inputs as follows:

Land area (X1)
The results of the partial test analysis that have been carried out show that the coefficient value of land area is 0.316. The coefficient of power X1 (0.316) shows the magnitude of the effect of land area on strawberry production. If X1 (land area) increases by 1%, then Y (strawberry production) will increas e by 0.316%. Conversely, if X1 (land area) decreases by 1%, then Y (strawberry production) will decrease to 0.316%. The significance value of t is X1 (0.000) < α (0.05). This shows that H1 is accepted and H0 is rejected. This means that the independent variable of land area has a significant effect on changes in strawberry production.

Seedlings (X2)
The results of the partial test analysis that have been carried out show that the coefficient of seedlings is -0.037. The coefficient of power X2 (-0.037) shows the magnitude of the effect of the s eedlings on strawberry production. If the use of X2 (seedlings) increases by 1%, then Y (strawberry production) will decrease by 0.037%. Conversely, if the use of X2 (seedlings) decreases by 1%, then Y (strawberry production) will increase to 0.037%. The significance value of t is X2 (0.088) > α (0.05). T his s hows that H0 is accepted and H1 is rejected. This means that the independent variable use of seedlings has no significant effect on changes in strawberry production.

NPK fertilizer (X3)
The results of the partial test analysis that have been carried out show that the coefficient of NPK fertilizer is 0.083. The coefficient of power X3 (0.083) shows the magnitude of the effect of NPK fertilizer on strawberry production. If the use of X3 (NPK fertilizer) increases by 1%, then Y (strawberry production) will increase by 0.083%. Conversely, if the use of X3 (NPK fertilizer) decreases by 1%, then Y (strawberry production) will decrease to 0.083%. The significance value of t is X3 (0.920) > α (0.05). This shows that H0 is accepted and H1 is rejected. This means that the independent variable use of NPK fertilizer has no significant effect on changes in strawberry production.

Labours (X4)
The results of the partial test analysis that have been carried out show that the labours coefficient value is 8.095. The coefficient of power X4 (8.095) shows the magnitude of the influence of labours on strawberry production. If the use of X4 (labours) increases by 1%, then Y (strawberry production) w ill increase by 8.095%. Conversely, if the use of X4 (labour) decreases by 1%, then Y (strawberry production) will decrease to 8.095%. The significance value of t X4 (0.027) < α (0.05). This shows that H1 is accepted and H0 is rejected. This means that the independent variable of use of labours has a significant effect on changes in strawberry production.

Joker pesticide (X5)
The results of the partial test analysis that have been carried out show that Joker pesticide coeffic ient value is 55.714. The coefficient of power X5 (55.714) shows the magnitude of the influenc e of J oker pesticide use on strawberry production. If the use of X5 (Joker pesticide) increases by 1%, then Y (strawberry production) will increase by 55.714%. Conversely, if the use of X5 (Joker pesticide) decreases by 1%, then Y (strawberry production) will decrease to 55.714%. The significance value of t is X5 (0.003) < α (0.05). This shows that H1 is accepted and H0 is rejected. This means that the independent variable of use of Joker pesticide has a significant effect on changes in strawberry production.

Conclusions
The variables used in this study were land area, use of seeds, use of NPK fertilizer, total labours and use of Joker pesticide. The land area, use of seedlings, use of NPK fertilizer, total labours and us e of Joker pesticide together have a significant effect on strawberry production with a coefficient of determination of 0.699. Partially, land area, labours and use of Joker pesticide have a significant effect on strawberry production. Meanwhile, the use of seedlings and the use of NPK fertilizer did not significantly affect the strawberry production. This explains that help from government in the form of capital and provision of production facilities to increase production in strawberry farming is still very much needed.