This paper is based on a combination of quantitative and qualitative data collected in Sumatra, Indonesia. This section introduces the local context, the surveyed communities, the gathered data, and the analytical methods.
Transportation in Indonesia has been dominated by heavily subsidized motorized road transport (Hook and Replogle
1996). The number of motorized vehicles in Indonesia has been growing at an annual rate of over 20 % (Susantono
2011). Motorcycles, which can operate even on roads in poor condition, are particularly popular (Hook and Replogle
1996). In 2012, there were already over 76 million registered motorcycles in Indonesia, i.e., approximately one registered motorcycle for every three people (Badan Pusat Statistik
2013). Non-motorized transport is often considered demeaning, primitive, and outmoded; sidewalks are mostly missing; and the safety of pedestrians has become a concern (Hook and Replogle
1996; Dick
2000). Increasing pollution has led to calls for transportation demand management (Susantono
2011). Mobile phone penetration has also been increasing rapidly in recent years: 81 % of Indonesians had a working mobile phone in their household in 2012—a significant increase from 67 % in 2011 (Broadcasting Board of Governors
2012).
Surveyed communities and gathered data
We gathered data from farmer groups in two subdistricts, Sumberejo and Pulau Panggung, of Tanggamus Regency. A fixed-form face-to-face-administered questionnaire was targeted at all household heads in 16 coffee- and cocoa-producing farmer groups, which were randomly selected among all 36 groups present in the two subdistricts. The 16 selected farmer groups listed 398 households as their members in 2008, when the lists were first compiled by the government. During the survey in September 2012, we managed to access and administer the questionnaire to 315 of them (79 %).
The self-identified household heads were asked to name persons from whom they received agricultural information. Specifically, the English translation of the prompt is, “Please list all people you can recall from outside this household whom you seek for advice, whom you can learn from, or who can generally provide useful information regarding farming practices.” In total, 1575 information-sharing ties were elicited. We refer to the number of persons whom each respondent named (in network terms, the ego’s outdegree) as the total number of information sources in the manuscript. The information sources are further split into the number of information sources inside the respondent’s farmer groups and the number of information sources outside the respondent’s farmer groups in the main analysis (in network terms, the number of alters in the respondent’s personal network who are based inside and outside his farmer group). The exact household location of each information provider was also identified. To assess the strength of the ties, we asked the respondents how they would describe their relationship with the information provider, how long they have known each other, and how often they communicate. The respondents were asked about the main mode of contact (menghubungi) with each communication partner. If the respondent provided more than one mode, the interviewer asked which mode was the most common and recorded his or her answer. Based on the respondents’ descriptions, the interviewers classified the relationships into “family/relative”, “neighbor”, “community organization member”, “know through work”, and “other,” but these variables were not significant in the multivariate models.
In addition to the network component, 13 pages of the questionnaire were dedicated to detailed questions about the socio-economic characteristics of the respondents and their households. The GIS coordinates and altitudes of the respondents’ households were also recorded, and the straight distance between all respondents and their informants (both inside and outside of the interviewed sample) was calculated. The basic characteristics of the sample are presented in Table
1.
Table 1
Sample description
Income (million Rupiahs per annum) | 296 | 29.40 | 42.10 | 0.00 | 176.00 | 449.00 |
Age (years) | 299 | 45 | 12 | 16 | 43 | 87 |
Distance to paved road (min of walking) | 271 | 4 | 7 | 0 | 1 | 60 |
Altitude (m) | 309 | 422 | 108 | 245 | 418 | 688 |
Personal network |
Total number of information sourcesa
| 315 | 5.00 | 3.84 | 1 | 4 | 20 |
Number of information sources inside the respondent’s farmer group | 315 | 3.64 | 3.55 | 0 | 3 | 19 |
Number of information sources outside the farmer group | 315 | 1.36 | 1.72 | 0 | 1 | 10 |
In August 2013, after statistically analyzing the data, we separately conducted qualitative interviews to substantiate the quantitative results. We interviewed 20 farmers from nine groups, averaging approximately two informants in each qualitative interview session. The interviews took a maximum of 2 h and were simultaneously interpreted between Indonesian and English by two interpreters to ensure accuracy. The informants described their information-gathering strategies and explained their attitudes toward different modes of contact. They also described the internal functioning of “farmer groups,” which are important elements in our theoretical framework, sampling, and analysis. We include a brief description of the farmer groups and the surveyed villages below.
Although some of the farmer groups have existed informally for decades, they were formalized mainly in 2007–2008 in response to a new governmental policy regarding the provision of official information, financial support, subsidized inputs, and equipment. According to information from the local government officials, government support to farmers, such as organic pesticides, fertilizers, or information about product competitions and training, is now channeled only through registered organizations. The typical size of a farmer group is approximately 20 households. The organizations have regular monthly meetings, which are held on a rotational basis in the members’ houses. The current agricultural activities of the coffee and cocoa farming groups mostly focus on sharing their experiences with new bio-agriculture practices because demand for bio-agriculture products has increased. The meetings may include religious rituals and arisan. In the local version of this Indonesian tradition, households regularly contribute to a common fund, which is given every month to a randomly selected winner among the households.
Large villages have more than one farmer group, but each household joins only the nearest one. Multiple memberships are not allowed. Sumatran villages—and thus farmer groups—tend to be ethnically segregated owing to a legacy of massive government-led group migration programs from other parts of Indonesia by previous generations. Each village has one or more mosques for everyday worship.
According to our informants, motorbikes first appeared in the surveyed villages in the 1980s, when the first paved roads were built. They became more popular at the end of the 1990s and the beginning of 2000s, when affordable 100cc models became widely available. Mobile phone towers were erected in the surveyed subdistricts in the early 2000s, and according to the informants’ explanations, mobile phones became widely popular in about 2005, partly owing to the farmers’ increasing affluence because of rising coffee prices.
Motorbikes were present in 85 %, mobile phones in 82 %, bicycles in 36 %, and cars in 6 % of the surveyed households. Moreover, 75 % of respondents had both a mobile phone and a motorbike in their household, and 8 % had neither.
Analysis
As described in Tables
1 and
2, our final dataset includes 1575 observations, i.e., 1575 links reported by 315 households, along with the geographical coordinates of the respondents’ households and their reported information sources. First, we create cumulative distribution functions to visualize the spatial extent of the relationships of mobile and motorbike owners and non-owners. We also compare the distance, the length of time, and the frequency of contact between different types of relationships, i.e., those in which walking, motorbikes, or mobile phones are the main modes of contact. We further display the distribution of geographical distance for the different types of relationships. This first step shows that the local inhabitants highly value face-to-face contact; thus, mobile phones are extremely rarely used as the main mode of information gathering.
Table 2
Description of agricultural information-sharing relationships
Relationships with all agricultural information-providing communication partners | 1575 | | | | | |
Straight distance (km) | 1544 | 2.7 | 6.55 | 0 | 0.62 | 69.92 |
Length of relationship (years) | 1553 | 18.73 | 11.18 | 1 | 20 | 60 |
Contact every day or every other day (yes = 1, no = 0) | 1561 | 64 % | | | | |
Relationships with communication partners who are mainly contacted… |
By walking | 1061 | | | | | |
Straight distance (km) | 1049 | 1.62 | 3.5 | 0 | 0.43 | 64.04 |
Length of relationship (years) | 1053 | 20.75 | 10.79 | 1 | 20 | 60 |
Contact every day or every other day (yes = 1, no = 0) | 1061 | 77 % | | | | |
Through private motorized vehicle use | 443 | | | | | |
Straight distance (km) | 438 | 3.93 | 7.77 | 0 | 1.48 | 69.92 |
Length of relationship (years) | 443 | 15.37 | 10.86 | 1 | 12 | 50 |
Contact every day or every other day (yes = 1, no = 0) | 443 | 38 % | | | | |
By mobile phone | 47 | | | | | |
Straight distance (km) | 47 | 10.94 | 14.95 | 0.17 | 7.82 | 68.36 |
Length of relationship (years) | 47 | 8.83 | 7.67 | 2 | 6 | 27 |
Contact every day or every other day (yes = 1, no = 0) | 47 | 40.4 % | | | | |
By landline phone | 5 | | | | | |
Through the use of public transport | 5 | | | | | |
Second, using the relationship-level data, we examine how motorbike owners choose between motorized transport and walking as their main mode of information gathering. We focus only on relationships for which the main mode of contact is face-to-face meetings, which the interlocutors attend either by walking or by motorbike, and exclude relationships for which the main mode of contact is by mobile phone because mobile phones are used in less than 3 % of relationships. Moreover, the number of farmers who have ties via mobile phone was too low to conduct the multilevel logistic regression. Therefore, only an approximate analysis that disregards the personal-level variation could be conducted with all three contact options, and this analysis is presented in the “
Appendix” section. The main analysis employs logistic regression with random intercepts to compare the two face-to-face options, while simultaneously considering both personal-level and relationship-level covariates. This “multilevel” approach is applied because error terms may not be mutually independent in a “single-level” model with only relationship covariates. In other words, because both personal and relationship characteristics affect the mode of contact, error terms of any single-level model include observed and unobserved characteristics of the respondent and are therefore correlated (see van Duijn et al.
1999). The use of multilevel analysis can alleviate the possible biases of single-level analyses, and the present results confirm that the variation at the personal level is considerable; thus, the multilevel approach is highly preferable.
Our personal-level covariates indicate the socio-economic status of the respondents (wealth, income, land ownership, and education) and other characteristics (age, migration experience) that have been found to be relevant in previous studies on technology adoption, usage, and mobility (Rogers
2003; Metz
2000). We also experimented with variables that may affect the ease of motorized transport usage (distance to paved and unpaved roads, household altitude, and distance to other households). Furthermore, our relationship-level variables indicate the distance between the interlocutors and the strength of their ties (kinship, frequency of contact, and length of the relationship) because tie strength has previously been found to be related (negatively) to information access (Granovetter
1973).