You are hereMental Models of Sustainable Agriculture
Mental Models of Sustainable Agriculture
Michael Levy, Neil McRoberts, and Mark Lubell
January 15, 2016
While interest in sustainable agriculture is widespread and increasing, precisely what is meant by “sustainable agriculture” is often ambiguous. Furthermore, the number, diversity, and interdependence of related factors (agronomic, social, economic, environmental, political, etc.) make it difficult for stakeholders to agree on how to develop programs and policies for supporting sustainable agriculture. To address these problems we assembled panels of agriculture experts from the public, private, and non-profit sectors and elicited their “mental models” of sustainable agriculture—detailed cause-and-effect representations of how sustainable agriculture works. Mental models allow us to examine how the understanding of sustainable agriculture is shared across multiple individuals and regions. Here, we group the individual mental models into a statewide model that shows the relationships among 37 key concepts related to sustainable agriculture.
Because most of our participants are already well-versed in sustainability thinking, the most central concept in the statewide mental is the catch-all term “sustainable agriculture”, which captures a broad range of concepts related to the traditional social, economic, and environmental goals of sustainability. Hence, it is more informative to look the goals and strategies that are linked to the central idea.
The other central goals represent more specific economic goals like product prices and environmental goals like resource management. However, the various terms related to science, knowledge, and education are also highly ranked concepts. Social aspects of sustainability receive less emphasis on average. This mirrors previous research on farmers where the economic goals of sustainability receive high priority, but also reflects the professional role of extension professionals in providing outreach, education, and policy engagement.
Interestingly, many of the central concepts are related to contextual factors that influence individual farm decision-making. These include markets and prices, water policy, and regulations and incentives. Some of the specific on-farm management issues, such as irrigation efficiency and IPM, are more peripheral in the statewide model.
While many central goals are shared across individuals and regions, there is much more diversity across individuals and regions in the strategies needed to achieve those goals. Within the overall architecture of the mental model, the shared goals tend to be “outputs” that are affected by a variety of other concepts, rather than inputs into the system.
There is a relatively high level of agreement that economic viability and environmental responsibility are important goals for sustainability, but less emphasis on social goals. Funding programs and policy should be developed to help elevate social goals to be on more equal footing with the other two traditional aspects of sustainability.
Science, education, and research are important inputs into the system, which means not only understanding how on-farm and food system practices influence the goals, but also analyzing how to better deliver knowledge to a diverse range of stakeholders. Importantly, at least as much research effort should be allocated to understanding how broader agro-ecological and social factors influence on-farm decisions, as effort allocated towards the consequences of specific on-farm practices.
For shared goals like economic profitability, research should not be devoted to clarifying the goals but rather identifying strategies for achieving those goals. The appropriate strategies to achieve those goals are likely to vary across regions of the state depending on agro-ecological, economic, and social factors. There may also be important variation across different agricultural operations depending on the specific configuration of agro-ecological and social factors at the farm level.
Participants in 4 workshops created 86 different mental models, which contained 1,474 different concepts. To aggregate these concepts, we used a semantic text analysis that calculates the similarity between two concepts based on how many shared words they have in their Google search result. For example “excessive chemical inputs” and “fertilizers” both return Google results including words like “chemical”, “soil”, and “nitrogen”, so they would have a high similarity score. The resulting similarity matrix provides the basis for a hierarchical clustering algorithm, which groups the 1,474 unique concepts into 37 aggregated categories. Since each category contains multiple individual concepts, we use expert judgement to assign a single label to each category. Continuing with the example, the clustering algorithm grouped “excessive chemical inputs” and “fertilizers” together into a concept that we labeled “best management practices”.
To measure the relationships between concepts, each participant was asked to draw ties between the concepts in their individual mental model. The ties represent cause-and-effect relationships, where an arrow originating from one concept is perceived to have a causal influence on the concept to which it is linked. To aggregate these individual maps, we add together the ties between concepts in all the maps. In the above example, if the participant who mentioned “excessive chemical inputs” drew a tie from that to “poor water quality”, and the participant who mentioned “fertilizers” drew a tie from that to “surface water contamination”, the statewide map would have a tie with the value of two from “best management practices” to “water management” (the aggregate categories). In the above map, arrow thickness and arrowhead size are proportional to the number of causal ties between the two concepts, and this information is also taken into account in calculating the centrality of each concept.
The causal status of a concept reflects whether it has more outgoing ties to other concepts or receives more incoming ties from other concepts. Specifically, it is the logarithm of the ratio of outgoing to incoming ties, so that it is positive for concepts with more outgoing ties (causes) and negative for concepts with more incoming ties (effects).
Figure 1 shows the overall structure of the mental model network among all 37 aggregated concepts. Arrow width is sized to the number of ties between the two concepts (the strength of the causal relationship between them), and concepts are sized to their total number of ties (their centrality in the model). Clicking on a concept highlights that concept’s ties and neighbors.
Figure 2 ranks the centrality of each of the 37 aggregate concepts in the overall statewide model, in terms of the number of overall causal relationships centered on that concept. In network analysis, this is called “degree centrality” and represents the sum of the strength of connections attributable to a concept. Concept centrality is normalized to the number of original concepts that were grouped into the aggregate concepts. This prevents a concept from appearing highly central simply because the aggregation algorithm grouped many concepts into one.
Each point in Figure 2 is also colored according to the balance of incoming versus outgoing ties associated with the concept. The more “red” concepts are “ends” or “goals” with a higher ratio of incoming to outgoing ties, while the blue concepts are “means” or “strategies” with more outgoing than incoming ties. The core finding from Figure 2 is that the more central concepts tend to be goals that are shared across individual mental models, while less central concepts are “means” that tend to have more diversity across individuals. Different people see a variety of ways to promote the core, shared goals, and these are more specific to each individual. This is consistent with our previous research, which found farmers from different regions tended to share sustainability goals but differ in the perceived practices that would promote them.
One notable exception to this pattern is the concept for research. Research appears quite central in the model and is perceived as more of a cause than effect – participants widely agree on the importance of research to promote other aspects of the sustainable agriculture system. The perception of research as an input to the system reflects the professional norms and activities of the outreach and extension stakeholders who participated in the workshops.
Future Research Questions
The analysis here is preliminary and includes data from workshops in Yolo, Merced, and Ventura Counties as well as a workshop at UC ANR’s 2015 Joint Strategic Initiatives Conference. Additional workshops are scheduled in early 2016 in Plumas, San Diego, Riverside, and Sonoma Counties. The algorithm used to group concepts is a novel semantic analysis tool that will likely undergo further refinement, and the labels applied to each aggregate concept need to be validated for consistency by a panel of experts. Future analyses will examine how the centrality of concepts varies across regions to provide a better picture of how understanding of sustainable agriculture differs across California.
The next stage of this research will involve a statewide survey of sustainable agriculture stakeholders. Survey respondents will provide a short written definition of sustainable agriculture, which can be compared to the mental models. The survey will also measure the personal communication networks of respondents, attitudes towards sustainability, and participation in sustainability programs and projects. We can then assess the correlation between the mental models of individuals, and their position in communication networks.