Examining the Relationship Between Spatial and Social Proximity in First Nation Food Sharing

Spatial Knowledge and Information Canada, 2019, 7(5), 4

ANA-MARIA BODGAN1, MENG LI2, DAVID NATCHER3, ABIGAEL RICE3, RONG SHEN2, WEIPING ZENG2, JASON DISANO1, SCOTT BELL4

1Social Sciences Research Laboratories, [email protected], 2The Spatial Initiative, [email protected], 3Agricultural and Resource Economics, [email protected], 4Geography and Planning, [email protected], University of Saskatchewan, Canada

ishes as spatial distances increase (Hare, ABSTRACT 1973; Latané et al., 1995). This study links social network analysis Verdery et al. (2012) examined the relation- (SNA) with GIS in the examination of First ship between kinship and spatial proximity. Nation food sharing. Introducing spatial in- They used spatially referenced kinship net- formation into conventional SNA offers new works and found a positive correlation be- perspectives and facilitates a better under- tween closeness of kin and households spa- standing of network data. Multiple GIS vis- tial proximity, attributed to close kin co- ualizations were used to complement the habitation. understanding of the network's multiple dimensions. Spatial statistics were carried Most studies use spatial proximity as the out to test key hypotheses about the rela- main factor contributing to tie formation. tionship between spatial proximity and so- Verderey et al. (2012) draw attention to the cial proximity in the food-sharing network. fact that in the case of kinship networks, the Results show both distance and kinship are closeness of relationships between family important variables in explaining food shar- members also influences where members of ing patterns. these communities decide to live. Social Network Analysis (SNA) is widely 1. Introduction used to answer questions related to individ- uals’ patterns of interaction, cohesion, social Within the social sciences literature, nu- influence, and proximity. However, it is not merous studies examined the role spatial spatially explicit. Mapping spatially refer- proximity plays on social tie formation. This enced social interactions and activities al- research found that people are more likely lows us to uncover spatial patterns, associa- to be friends if they are geographically close tions, and ask new questions of network da- (Festinger et al., 1950; Feld and Carter, 1998; ta (Logan, 2012). This motivated the current Mouw and Entwisle, 2006), whereas those study. Specifically this study set out to ex- who live further from a community “core” amine the spatial parameters that influence tend to be socially isolated (Festinger et al., food sharing between members of the 1950). The physical location of one’s place of Saulteau First Nations (SFN). residence further increases the likelihood of strong social tie formation with others in The SFN is located in northeast British Co- close proximity (Coombs 1973) and dimin- lumbia. The population of SFN is 380, living in 125 on-reserve households (Statistics 2 Examining the Relationship Between Spatial and Social Proximity in First Nation Food Sharing

Canada, 2016). SFN economy is a mixed 2. Methods and Data economy, which includes wage earning and wildlife harvesting activities. 2.1 Data Within the SFN community, sharing har- Data used in this study comes from a larger vested wild foods is prominent. Food shar- research project focused on the Assessment ing serves numerous functions, including of First Nations Environmental Livelihoods the alleviation of household food insecuri- in Northeast British Columbia (Lu et. al., ties, continuance of a cultural tradition, and 2019). social cohesion. In general, the SFN main- Household surveys conducted in the SFN tains strong family bonds. Kin continue to were used to assess how environmental depend on each other for their social and change might affect their harvest and sub- economic wellbeing. sequent food sharing. The first part of the What requires more attention is the rela- survey focused on wildlife harvesting and tionship between social ties and physical the second part of the questionnaire collect- proximity. In this research, we measured ed data on food sharing. Responses were physical proximity using the Euclidean dis- compiled into several spreadsheets for anal- tance, which was used to explore its rela- ysis. Due to lack of spatial information for tionship with social proximity (i.e. kin/non- 13 households, 154 out of the 179 food ex- kin, nuclear/distant kin) by answering the change records were used in the final analy- following: sis. 1. Do kin live closer1 to each other than A modified 10x10 km grid map of the T8TA non-kin? territory was developed with GIS. This al- 2. Do nuclear family members live closer lows us to visualize the concentration of re- than extended kin does? source harvesting, in terms of household 3. Within kin, is there an association be- land use, food weight, by species, travel dis- tween living close to one another and tance, and ecological values including land food sharing? cover and habitat fragmentation. 4. Are households located further from the Households’ locations came in PDF format community’s core less engaged in food from Peace River Regional District (dated sharing? September 30th, 2014) showing land surface 5. Do households located further from the features as well as the location of 167 recog- community’s core receive less food? nized households on Moberly Lake. House- 6. Are households located closer to the hold locations were georeferenced and ex- 2 community core engaging in more food tracted. Certain households were not located sharing? because (1) the reference layer was outdated; The first part of our analysis involves visual- or, (2) survey data collection errors. izing the network data. This elucidated the Locations of regional communities were de- spatial distribution of SFN’s local and re- termined using Google’s geolocator service gional food sharing. In the second part of followed by manual revisions for quality this paper, we focus on testing multiple hy- control. There were in total 77 households potheses related to spatial and social prox- involved in the final analysis. imity. To ensure the integrity of the network da-

taset and account for unknown or missing the label “unknown location” was used. 1 In this study, the term “close(r)”means physical However, with missing data, there is less proximity certainty over the reliability of findings, which is an important limitation to recog- 2 Community “core” defined as the geographic center nize. In the future development of this re- of the community. search we plan to incorporate various impu- Examining the Relationship Between Spatial and Social Proximity in First Nation Food Sharing 3 tation techniques, already established in the SNA domain, to mitigate against missing data related errors.

2.2 Software 1. Python and R were used in this study to process the original data for the initial ex- ploratory analysis and preparation for sub- sequent analysis. 2. UCINET was used for statistical testing and SNA descriptive statistics. Sociograms were created with NetDraw. 3. ArcGIS was used in this project to man- age geodatabases as well as data manipula- tion, visualization and spatial analysis. 4. Interactive web-based SNA sociograms Figure 1. Overall methodology of this paper were programed as custom single page web applications. The backend was powered by Apache and PHP. HTML, JavaScript and 2.3.2 Survey data reformatting CSS were used for the front-end develop- A structured survey was conducted. Original ment. paper results were scanned into PDF format 2.3 Methods from correspondents in SFN; some responses were handwritten. A spreadsheet tem- 2.3.1 Overall methodology plate was used to manually transcribe each The workflow started with social data collec- survey, resulting in 150 individual spread- tion through a structured survey (Figure 1). sheets for all households in SFN. Custom Surveys were cleaned and processed before Python and R scripts were developed to analysis. Analysis was done in three steps. reformat these spreadsheets into formats suitable for data visualization and analysis. 1. Conventional SNA, which helped to explore the dataset through visualization and 2.3.3 Data visualization performing non-spatial statistics as well as Several visualizations were used to explore hypotheses testing. the food sharing network dataset. A socio- 2. Visualization and spatial analysis using gram is an effective way of visualizing social desktop GIS and WebGIS that facilitate ex- network data. It depicts relationships ploratory analysis, interpretation and pre- among specific groups, with the aim of dis- pare distances for investigating of SFN’s covering underlying relationships (Figure 2). food sharing network. A desktop SNA software and web-based so- 3. Based on the spatial data and findings lutions were used to prepare the socio- from the previous two steps, testing hypoth- grams3. The latter was specifically designed eses on the interaction of spatial proximity to streamline the procedure from raw data and social proximity.

3 The term sociogram only partially depicts what these diagrams are. We are considering using an al- ternative term – sociomap, to better label these fig- ures. 4 Examining the Relationship Between Spatial and Social Proximity in First Nation Food Sharing to final presentation and dissemination to between in-degrees and out-degrees (Free- stakeholders. man, 1979). Conventional sociograms are not spatially explicit. Desktop GIS and WebGIS were 3. Results used to process data for various geographic 3.1 Integration of SNA and desktop GIS visualizations. However, visualizing social network data presents unique challenges, The main objective of this study is to bridge including multiple scales (e.g. regional and SNA and GIS to reveal hidden patterns. household), multiple dimensions (e.g. dif- Such link could be easily implemented ferent food categories), directions of ties (e.g. through the loose coupling of SNA and GIS giving and receiving in food sharing), multi- via a thematic map. Figure 2 shows how the ple weights of nodes and ties, as well as sociogram was usually rendered inside SNA overlapping ties that interfere with each software. In this case, the food-sharing net- other and impede visualization. A custom- work inside SFN is visualized at the house- built online solution was developed. hold level. At the regional level, with all communities’ locations defined, a GIS can 2.4 Data analysis display a thematic map, or a geographic so- To determine the role of spatial proximity in ciogram (Figure 3). Moreover, the previous SFN’s food sharing network, student-t test household-scale sociogram is embedded to was conducted to compare the food sharing offer a complete view of the food-sharing (measured by food quantity, mean Euclide- network for SFN. an distance) categorized by kinship type. Permutation based t-tests and other SNA descriptive statistics were conducted in UCINET. To measure spatial autocorrelation we used both Moran’s I and Geary’s C. Moran's I ranges from +1 - strong positive spatial autocorrelation, to -1 - strong negative spatial autocorrelation, with 0 indicating a random pattern. For Geary’s C, a value of 1 denotes no association, less than 1 a negative autocorrelation, whereas a value greater than 1 denotes positive autocorrelation. Households’ activity levels within the food- sharing network were measured using in- degree centrality scores in order to assess how much they received. Out-degree cen- Figure 2. SFN Food Sharing Sociogram NetDraw trality scores were calculated to assess how much a household shared with others. The number of vertices4 adjacent to a given vertex in a symmetric graph is the degree of that vertex, also referred to as Freeman degree centrality. In a directed or non- symmetrical graph, we further distinguish

4 Households are the graphs’ vertices, and food- sharing interactions are ties. Examining the Relationship Between Spatial and Social Proximity in First Nation Food Sharing 5

Figure 3: SFN local and regional sociogram To understand the pressure of harvesting made as a thematic map using ArcGIS activities on the landscape, a grid-based web sociogram was created (Figure 6 and 7). 3.2 Web-based geographic sociogram This web application offers a graph view and Though sociograms could be made using a map view. The graph view displays the desktop software, web-based visualizations food-sharing network for any chosen re- were also investigated. Moreover, web- source category. It also differentiates based visualizations offer interactive fea- households by their selected roles. The map tures that cannot be achieved through a view links the social network to the land- loose coupling of desktop SNA and GIS. scape. As a result, the researcher can trace the food extracted within a 10x10km2 grid Figure 4 and 5 show how the WebGIS can and how it was shared within the social streamline and extend non-spatial methods. network.

Figure 4: Regional food sharing network for SFN Figure 6: Grid-based online sociogram for the and another community. Arrows indicate food food-sharing network in SFN: the graph view. flow directions.

Figure 5: Regional food sharing network for SFN Figure 7: Grid-based online sociogram for the and another community. Resource types can be food-sharing network in SFN: the grid view. This filtered by the dropdown list. The network is also view allows researchers to query resource cate- visualized with a Sankey diagram to portray the gories harvested within a selected grid and dis- volume of exchanges for selected food types. plays its food-sharing network. 6 Examining the Relationship Between Spatial and Social Proximity in First Nation Food Sharing

3.3 Statistical Analysis 3.3.1 Social Proximity vs. Spatial Proximity One of the main objectives of this paper is to understand if there is a statistically significant difference attributable to physical distances between kin and non-kin households, and between nuclear and extended kin. Re- sults show that within the SFN community, family members live closer to one another than non-kin. However, this difference is not statistically significant. Table 1. Significance test of distances between Figure 9: Distances by closeness of kin kin and non-kin groups We were interested in whether kin that live KIN NON-KIN close to one another share more food. We Distance 1.68±1.3 km 2.02±0.89 km calculated Geary and Moran’s I statistics to Difference -0.35 km p=0.40 answer the question. Results show a weak in Means negative tendency between degree centrality and distance to kin (G=1.43, p=0.1; MI=- When compared to members of the extend- 0.24, p=0.08), however this autocorrelation ed family, members of the nuclear family holds at p=0.1. In other words, there seems live closer together. This difference was not to be a tendency for kin who live close to one statistically significant either. another to exchange more, if we accept a p- value of 0.1. Furthermore, given that this Table 2. Significance test of distances between result was obtained using an incomplete nuclear and extended groups kinship dataset, further research is required NUCLEAR EXTENDED to consolidate this finding. Distance 1.54±1.23 km 1.80±1.18 km When we look at the entire dataset, irre- Difference -0.25 km p=0.51 spective of whether households exchanging in Means food are related, households living close to

each other tend to share more food (G=2.08, p=0.01). This should be interpreted with caution, since we restricted our food- sharing network to exchanges taking place only within the SFN community. 3.3.2 Social Engagement vs. Spatial Proxim- ity Similar analysis was performed to answer the remaining three questions. In the following section, distances represent the distance of households from the community’s geographic center (‘core’). Figure 8: Distances by kin group Are households located further from the community’s core less engaged in food sharing? Levels of engagement in the food-sharing network were measured using Freeman’s Degree Centrality scores. Households with Examining the Relationship Between Spatial and Social Proximity in First Nation Food Sharing 7 at least four connections were considered pared to non-high givers (out-degree cen- highly engaged in the food-sharing network. trality < 3). However, the difference in mean Results show that less engaged households distance from the core between these two live, on average, closer to the community’s groups was not statistically significant (p = geographic core. However, there was no sta- 0.97). tistical difference (p = 0.49). This result Table 5. Distances between high givers and non- needs to be further refined – spatial cluster- high givers ing in the data means that there are multiple High Non-High community foci. Givers Givers Table 3. Significance test of distances between Distance from 1.47±0.68 km 1.46±0.91 km highly engaged and non-highly engaged groups SFN core Highly Non-Highly

Engaged Engaged Distance 1.62±0.81 km 1.43±0.88 km 4. Conclusion from SFN core Difference in By incorporating spatial information into 0.18 km p=0.49 Means social network analysis, this study investigated visualization options and discussed the contribution of spatial variables within Do households located further from the social network. the community’s core receive less food? Spatially explicit statistics were used to test key hypotheses about the role of spatial In-degree centrality scores quantify the total proximity and social proximity in the local number of incoming ties for each household. food-sharing network. Both geographic dis- In the SFN food-sharing network, this score tances and kinship are important variables represents the number of times a household in explaining food-sharing patterns. received food. Based on these scores, we created a dummy variable for high receivers Further research is essential, especially in and low receivers (receiving food above or the context of kinship networks where social below three times). Results show that proximity can reinforce spatial proximity households receiving less food are located, and vice versa. Moreover, impacts of differ- on average, closer to the community’s core ent measures of spatial distances need to be when compared to high receivers. This dif- evaluated. ference was not statistically significant Furthermore, this line of inquiry needs fur- (p=0.79). ther advancement, by building on anthropo- Table 4. Significance test of distances between logical and social studies realized within the high receivers and non-high receivers context of Canadian indigenous communi- High Non-High ties.

Receivers Receivers Distance from Acknowledgements 1.48±1.39 km 0.88±0.81 km SFN core The British Columbia Ministry of Forests, Lands and Natural Resource Operations provided fund- Do households located in the core of ing for this research. We are grateful to them for the community give more food? their support as well as for the support and col- laboration of the SFN. The authors are thankful Households located closer to the core of the to The Spatial Initiative and the Social Network community are not giving a lot of food to Lab for their commitment in this project. other community members. On the contrary, households giving more food were located slightly further from the core, when com- 8 Examining the Relationship Between Spatial and Social Proximity in First Nation Food Sharing

References Lu X., Natcher D., Rice A., Bodgan A., Zeng W., Ha T., Wang J., Azizi M., Abandankeshi Coombs, G. (1973). Networks and exchange: The F. Z., Doan T., Disano J., Bell S. (2019). role of social relationships in a small Data collection and Spatial Analysis for voluntary association. Journal of An- Value Assessment of First Nations Envi- thropological Research, 29(2), 96-112. ronmental Livelihoods in Northeast British Columbia. Spatial Knowledge Feld, S. & Carter, W. (1998). When Desegrega- and Information Canada paper proceed- tion Reduces Interracial Contact: A ings, 9. In Press. Class Size Paradox for Weak Ties. Amer- ican Journal of Sociology - American Mouw, T. & Entwisle, B. (2006). Residential seg- Journal of Sociology. 103. 1165-1186. regation and interracial friendship in 10.1086/231350. schools. American Journal of Sociology, 112(2), 394-441. Festinger, L., Schachter, S. and Back, K. (1950) https://doi.org/10.1086/506415. “Social pressures in informal group: a study of human factors in housing” Statistics Canada. (2016). East Moberly Lake 169, Stanford University Press, Stanford, Cal- Indian reserve [Census subdivision], ifornia. British Columbia and Peace River, Re- gional district [Census division], British Freeman, L. C. (1979). “Centrality in Social Columbia. Retrieved from Networks: Conceptual clarification”, So- https://www12.statcan.gc.ca/census- cial Networks 1, 215-239. recensement/2016/dp- Hare, A. P. (1973). Theories of Group Develop- pd/prof/details/page.cfm?Lang=E&Geo ment and Categories for Interaction 1=CSD&Code1=5955801&Geo2=CD&Co Analysis. Small Group Research (Behav- de2=5955&Data=Count&SearchText=m ior). SAGE Publications. ober- ly%20lake&SearchType=Contains&Sear Latané, Bibb, James H. Liu, Andrzej Nowak, Mi- chPR=01&B1=All&TABID=1. chael Bonevento, and Long Zheng. "Dis- tance matters: Physical space and social Verdery, A. M., Entwisle, B., Faust, K., & Rind- impact." Personality and Social Psy- fuss, R. R. (2012). Social and spatial chology Bulletin 21, no. 8 (1995): 795- networks: Kinship distance and dwelling 805. Logan, J. R. (2012). Making a place unit proximity in rural Thailand. Social for space: Spatial thinking in social sci- networks, 34(1), 112-127. ence. Annual review of sociology, 38, 507-524.