Diffusion on social networks: Survey data from rural villages in central China

Empirical studies on social diffusions are often restricted by the access to data of diffusion and social relations on the same objects. We present a set of first-hand data that we collected in ten rural villages in central China through household surveys. The dataset contains detailed and comprehensive data of the diffusion of an innovation, the major social relationships and the household level demographic characteristics in these villages. The data have been used to study peer effects in social diffusion using simulation models, “Peer Effects and Social Network: The Case of Rural Diffusion in Central China” [1]. They can also be used to estimate spatial econometric models. Data are supplied with this article.


a b s t r a c t
Empirical studies on social diffusions are often restricted by the access to data of diffusion and social relations on the same objects. We present a set of first-hand data that we collected in ten rural villages in central China through household surveys. The dataset contains detailed and comprehensive data of the diffusion of an innovation, the major social relationships and the household level demographic characteristics in these villages. The data have been used to study peer effects in social diffusion using simulation models, "Peer Effects and Social Network: The Case of Rural Diffusion in Central China" [1]. They can also be used to estimate spatial econometric models. Data are supplied with this article.
& 2016 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Table   Subject  The network data can be used to quantitatively analyze social structure of rural communities. The dataset can be used to demonstrate how social relationships shape economic activities.

Data
We present the latest survey data collected from 10 villages and over 400 households. This set of data includes (1) the ties of major social relationships between households (namely, network data), household level demographic characteristics (namely, demographic data), and the year in which each household adopted a new crop (namely, diffusion data). Table 1 presents basic statistics of these villages. Fig. 1 shows how the new crop was diffused in these villages (which constitute an administrative village named GRV) and how households' income changed accordingly.

Sample selection
The households in each village are selected as follows. We begin with a name list of households that signed a land contract with the administrative village committee (AVC) of GRV. The numbers of households with a contract are presented in the second row of Table 3. The total number over all villages is 463. Subsequently, we excluded (1) the households that were not an independent household before signing the contracts, (2) the households whose member(s) retired from farming before 2000we use the remaining households to create the social networks in these villages, they thus called network sample (fifth row of Table 3) -(3) Emigrated before 2000, (4) emigrated after 2000 and had not adopted before the time they ceased to practice farming, (5) with no or limited ability to practice farming, and (6) households doing non-farming job. The remaining households constitute the sample that can be used to study the diffusion of the new crop. They thus called diffusion sample (tenth row of Table 2).

Kinship data
We collected the family trees of all 63 lineages and all the intermarriages occurring between 1950 and 2000 in the ten villages. Using these data, we created the kinship ties (including blood ties and affinity ties) between the households. In addition, we calculated the weights of the ties according to  the closeness in blood. We set the weight of the closest tie, i.e., the one between siblings and the one between an individual and his parents, to be 1, that of the second closest tie to be 0.5, and so forth. Refer to [2] for an elaboration on the methods of calculating weights of blood ties and weights of affinity ties.

House neighborhood relationships
The data of house neighborhood were collected using Google Earth map, on which the houses in the villages can be clearly identified. With the aid of the tool ArcGIS, we calculated the distances between houses. We then took those within a distance smaller than 15 m as first-degree neighbors of each other and set the weight to be 1, and those houses that are located in the distance 15-30 m as the second-degree neighbors and set the weight to be 0.5.

Land plot neighborhood relationships
The households generally have 5-8 plots of farming land. A household's plots are often adjacent to other households'. Plot neighborhood is an important social relationship in the villages. The information of plot neighborhood is well documented in each household's land contracts. We can use this information to define the plot neighborhood between households. Plots smaller than 0.5 mu is classified as small plots, those between 0.5 and 1 mu are classified as medium-sized plots, and those larger than 1 mu are large plots. According to this classification, we set the weights small, mediumsized and large plot as 1/4, 1/2 and 1, respectively.

Political relationship data
Political relationship in the villages is composed of ties between village leaders, including AVC members and the heads of village, and party membership ties. Specifically, there are six AVC members and ten heads of village (each village has one), and they work together at least once per week. There have been 38 party members by 2014 in GRV, and they work together about once per month. We collected the names of all village leaders and party member in GRV from 2000 to 2014. To differentiate the strength of the two different kinds ties, we set the weight of a tie between village leaders to be 1, whereas the weight of a tie between party members to be 0.5. Table 3 presents the network characteristics of the social network consisting of all types of ties in GRV.

Diffusion data
We conducted a questionnaire survey over all households in the villages. In the questionnaires, we asked the farmers (usually the heads of the households) the acreage of the land that they used to farm the AS in each years, especially the years in which they farmed more than 1 mu. The questionnaire results were verified by two pieces of documentary evidence: the historical imagery on the Google Earth map, from which the land plots that were planted AS in 2003 can be identified, and the record

Demographic data
We obtained detailed demographic information of each person that is registered to GRV from the Department of Residency Registration in the town and AVC. The information covers acreage of land and number of farming workers of households, and surnames, year of birth and number of schooling years of the heads of household, in each village from 2000 to 2014.