To reveal more about the community of influential relationships surrounding farmers, pile sorts and semistructured interviews were conducted with 10 dairy farmers, 11 agricultural bankers, and 10 agricultural insurers from the central Wisconsin region. The region included a 7 county area (Adams, Clark, Jackson, Langlade, Marathon, Portage, and Wood counties). Bankers and insurance agents likely serviced farms outside these counties. Agricultural bankers and agricultural insurance agents were defined as those whose portfolio was at least half agriculture. These professionals were chosen as the focus given their financial investment in the success of their clients’ farms. However, as will be shown, a number of different individuals and institutions can be influential of farmers’ decision-making. The study was reviewed and approved by the Marshfield Clinic Research Institution’s Institutional Review Board.

Prior to research activities, one to three advisors representative of each group were recruited to assist in the development of the pile-sorting and interview instruments. Advisors were not included as research participants. These advisors reviewed possible cards and interview questions. Advisors also assisted in participant recruitment conducted through snowball sampling.

Three of the interview questions were framed as pile-sorting prompts, asking the participants to sort 32 index cards. On each card was written an individual or group that was thought to possibly affect farmers’ decision-making, both generally and about health and safety. Three of the pile-sorting prompts included 5-point Likert-like scales. The pre-established scales eased some of the burden for the participants and provided researchers with additional quantitative data. All Likert-like responses were assigned scores of 1–5, 5 being favorable in terms of possible influence that person or institution may exert on farmers’ decision-making whether it be measured by time, knowledge, or trust. It is important to note that farmers, agricultural bankers, and insurance agents were the study participants as well as individual index cards. In other words, they were asked to rank themselves during pile-sorting activities. The initial analysis (presented here) focuses in on how the three participant groups (farmers, agricultural bankers, and insurance agents) rated themselves and each other. Future manuscripts will focus on the application of the methods to developing public health interventions and the full results of all 32 cards. Interviews ranged from 45 to 80 minutes; time variability was largely dictated by the amount of detail research participants shared as they placed cards into piles. Results (photographs of piles of cards quantified into spreadsheets, fieldnotes, MP3 audio recordings, and interview transcripts) were analyzed with SAS (SAS Institute Inc., Cary, NC) and NVivo. Nvivo is a qualitative analysis software program that allows users to identify common themes and code data.

The research team consisted of the principal investigator, two research associates, two MPH student interns, and a biostatistician. Research activities were conducted from summer of 2014 to fall of 2016. Three research team members with a background in farming initially formulated a draft of the interview questions and 60 possible cards. Three consults were conducted with farmers to field test the questions and cards. Farmer-advisors suggested changes to wording and helped reduce the number of cards to 32, a process that included consolidating, omitting, and adding cards (see Figure 3). Per most pile-sorting instruments, this process was essentially the free-listing stage.

The interview instrument used with farmers began with three demographic questions regarding the size of their dairy operation, how many people worked on the farm, and how they utilized their acreage. These questions were useful for easing the farmer into discussing their operations and revealed features of the farm that might explain why some people are more influential than others. For example, more or less milking cows per acre used for crops might explain why a veterinarian is more or less influential than an agronomist. Bankers and insurance agents were asked demographic questions regarding the amount and kinds of agricultural clients, their years of experience, and whether farm injuries and deaths had impacted their business.

Following the opening line of questions, the pile-sorting prompts began for all three groups. Farmers were first asked, “How often do you talk with these people?” The Likert-like scale to assist participants’ sorting included categories of: daily, weekly, monthly, yearly, and hardly ever. After all the cards were sorted, follow-up questions were asked regarding why some people and groups were spoken to so often or seldom. Agricultural bankers and agricultural insurance agents were asked to sort the cards according to “role they play in the success of a dairy farm.” This was occasionally clarified to mean the financial success of the dairy farm. The Likert-like scale to assist both groups with this question consisted of the categories: critical to success, advantageous, neither critical nor impeding, can be harmful, and impedes success. Participants were asked to think aloud while they placed cards into categories. Thus, the method produced more in-depth data than if they had independently completed a survey. Digital photographs were taken of the sorted cards. Cards were then collected and reshuffled after each question.

Follow-up questions for the initial pile-sorting prompts of both bankers and agents probed why some people and groups were considered so important to a successful farm and why others impede success. The questions allowed researchers to investigate the role of other parties in dairy farmers’ daily lives. These questions also provided more depth around how substantive each actor's role was in the financial and general success of a farm. The question for dairy farmers generated lists of actors most involved in the daily operations on a farm.

The second pile-sorting question for all three groups queried who was thought to have knowledge about farm safety and who would be capable collaborators in designing a farm safety program. Emphasis was placed on the design of a farm safety program and not its implementation. Likert-like scales provided participants with categories of responses that included: know the danger and how to prevent it; know the danger and how to be cautious; know to be cautious but are unaware of specific dangers; no knowledge of danger or caution or; reckless.

The third pile-sort questions prompted participants to sort cards according to which person or group would do best at delivering and implementing a farm safety program because of the trust established between them and the farm community. The initial question was edited after the first four participants to improve clarity. The phrase, “In terms of how quickly you would make changes,” was added to the question “how much do you trust these individuals/groups?” The addition reflected necessary explanations provided to the initial four participants and are not expected to differentially bias the data of the dairy farmers. Likert-like scales for this pile-sorting question included: change as quickly as feasible; change eventually; change after further consultation with other people; not likely even with other consults or; skeptical about the truth and will not make a change.

Numerical values were assigned to the categories on the five-point scale where a low score (1) indicated lesser elements of trust, knowledge, or involvement on the farm and a high score (5) indicated greater elements of trust, knowledge, or involvement on the farm. Kruskal–Wallis tests were performed to evaluate the overall differences in the three groups across all items and for select individual cards/items. If the Kruskal–Wallis test achieved at least marginal significance for a particular card/item, pairwise comparisons between groups were undertaken by performing exact Jonckheere–Terpstra tests.²⁸ These tests revealed when agricultural bankers, insurance agents, and dairy farmers were agreeing (or disagreeing) about the relationship and possible role a person or an institution could play in the health and safety of a dairy farm. In addition, qualitative data collected during the interview and pile sorting activities provided further depth and detail as to the content and nature of the relationship between the individuals and institutions.

A Kruskal–Wallis test indicated that differences between the three groups of participants were present in their composite responses (i.e., sum of Likert responses for the 32 cards) on designing a farm safety program (p = 0.07) (Figure 4). Pairwise comparisons revealed a significant difference between dairy farmers and bankers (p = 0.02, exact Jonckheere–Terpstra test) and a marginally significant difference between bankers and insurance agents (p = 0.10). No differences in composite responses were detected between the three groups with respect to implementing a farm safety program (p = 0.52).

Bankers, insurers, and farmers believe that bankers are not critical or harmful to the design of farm safety program (see Table 1). Insurers were more confident in their own ability to assist in the design of farm safety programs (mode = 5, necessary to the design of a farm safety program) compared to bankers gauging their own ability (mode = 4, helpful).

Among the three groups of participants, agricultural bankers and insurance agents most closely resembled each other regarding the potential role of agricultural bankers in designing and implementing a farm safety program. Dairy farmers were slightly less confident in the role of a banker on a farm safety program. Ultimately, the three participant groups did not differ significantly (p = 0.27 for design and p = 0.25 for implementation of a farm safety program).

Qualitative data support and deepen the quantitative results. Bankers reported being encumbered by regulation, restricting their ability to design or implement a safety program. One banker explained,

Bankers saw their job as only secondarily or tertiarily related to safety.

Attitudes toward involving insurance agents in designing a farm safety program differed significantly in a Kruskal–Wallis test (p < 0.05). Farmers constituted most of the difference as their pairwise comparison in Jonckheere–Terpstra tests revealed farmers to differ significantly from bankers (p < 0.05) and insurance agents (p < 0.05), but bankers and insurance agents were in general agreement with each other (p = 0.35). Similar results were obtained regarding the implementation of a farm safety program, although the comparison of bankers and insurance agents yielded a p-value of 0.98.

Insurance agents were more confident in their ability to implement a safety program (mode = 5, farmer would make the change as quickly as possible) than bankers had in their own ability (mode = 4, farmer would make the change eventually, see Table 2).

Similar trends can be observed in the qualitative data. While bankers were reluctant to see a role for themselves in designing and implementing a safety program, bankers lauded the potential of insurance agents to do the same.

Insurance agents echoed bankers’ confidence regarding the role insurance agents can play in designing a safety program. In response to why he had put insurance agents as “necessary” to this work, one agent replied,

Insurance agents were believed to be less encumbered by regulation, more motivated to achieve safer farms, and more confident in their own ability to do so.

The primary foci for this project were the relationships between central Wisconsin dairy farmers and agricultural bankers and agricultural insurance agents. These are the only three groups that the data can reveal mutual consensus about one another. However, the other cards in the pile-sorting questions emerged as important possible agents of change or influence, often with agreement across all three groups. These results are currently under further analysis and testing. The preliminary results may lead to important next steps. First, as expected, fellow farmers, spouses, and immediate family were highly ranked in terms of their knowledge and even more so for the level of trust that might enable the implementation of a safety program. Previous researchers have explored these individuals, but not necessarily in any large intervention endeavor.^21-25 Second, all three groups held veterinarians and agronomists in high regard for their ability to convince farmers to make changes in their operation. Third, agricultural health and safety experts were held in high regard toward the design of a safety program, but often fell in standing when it came to implementation. Last, results for extension agents were somewhat scattered. Interview data revealed that this assessment is highly dependent on the particular extension agent in the county where the participant resides, there being no choice as with insurance agents and bankers.

The snowball recruiting process often led to questions of how many and of what quality agricultural insurance agents the research team was seeking. This led to three important and unanticipated findings. First, research participants felt that 10–11 agricultural bankers and agricultural insurers represented at least half of those in the professions in the region. If true, this bodes well for intervention design, as small clusters of the dominant agricultural insurers and bankers could be recruited and organized much more easily than farmers. Banks and insurance companies with regional focus and coverage may be particularly viable, though they may lack the financial leverage of larger companies. Given the confidence each group had in agricultural insurers’ abilities to design and implement safety programs, including themselves, they seem to be a promising first target.

Second, the research team learned that most participants from all three groups believed there to be a shortage of high-quality agricultural insurance agents and bankers when compared to decades past. “High quality” was often characterized as “understanding farming,” i.e., appreciative of the vagaries of income and amount of risk involved. This was often further contextualized with explanations that most older agricultural bankers and insurers were once farmers themselves and had sought the profession after retiring from farming, giving them native knowledge and status in the profession. It was believed that with the shrinking amount of farmers and the prolific migration of young people away from farming, this line of farm-reared and seasoned professionals is dwindling. This was part and parcel to the third unanticipated result: research participants in all three professions believe there is a shortage of young, eager people interested to replace them in the future.

The second and third unanticipated results have both positive and negative implications for safety interventions looking to leverage agricultural bankers’ and insurance agents’ relationships with farmers. As the number of perceived high-quality agricultural bankers and insurance agents continues to stagnate or dwindle, their status in the eyes of farmers may do the same. The farmers’ beliefs that these professionals can understand both farming and their profession may erode and any possibility of increasing safety’s role in their relationship may do so as well. As young and beginning agricultural insurance agents and bankers do enter the field, they will likely not have a farm background and will need mentorship in all things agriculture, which presents an additional avenue of intervention. Agricultural health and safety experts should be keen on developing materials for and collaborating with those mentors, making health and safety integral to how future agents and bankers calculate and finance risk.

Generally, imagining the agricultural community around farmers through the lens of the SEM (spheres of influence) appears to be a fruitful direction. Moreover, applying what can be learned about these personal relationships can flush out more cues to action as suggested in the HBM. Applying agricultural health and safety knowledge through preexisting, imbedded relationships and cues to action therein, however, will require further research. Given the variability across agricultural communities, local knowledge will prove essential in determining which agents of change and cues to action are worth attempting.

As a pilot project, the research team was very satisfied with the quality and quantity of data. The primary limitations given the size and scope of this project were the limited sample sizes and the amount of time necessary to accumulate the data. While 10 dairy farmers is not be representative of all central Wisconsin dairy farmers, it is interesting to note however, that participating agricultural bankers and insurance agents commonly believed that 10–11 in each profession represented at least half of central Wisconsin's population. Future studies should investigate the veracity of this claim. If true, these professionals represent a much more manageable intervention population than farmers themselves. These limitations are typical of a primarily qualitative approach based on one-on-one interactions across a large geographic region. Reaching more farmers with this approach would take a large amount of resources or advancement in the technology to capture similar kinds of data. Digital pile-sorting programs are emerging and could enable this wider catch. In terms of agricultural bankers and agricultural insurance agents, while small samples are likely representative of entire regions, statistical analysis can be difficult. Again, digital modes of pile sorting might resolve this. However, differences in states’ and regions’ agricultural production and insurance or banking systems may be a deterrent to consolidating these professions across wider regions.