Education and Economic Factors Shape Clusters of Biosecurity Beliefs and Practices: Insights from an Exploratory Survey of Midwest U.S. Swine Producers

Abstract

Despite existing biosecurity frameworks, there is limited empirical evidence on how US swine producers' beliefs, behaviors, and risk perceptions influence enhanced biosecurity implementation. We conducted an online survey among US swine producers to understand their biosecurity beliefs, behaviors, and practices. We used descriptive, unsupervised machine learning, and Factor Analysis for Mixed Data (FAMD). Of fifty-four respondents, 48.1% reported implementing some biosecurity measures, and 72.2% valued having enhanced biosecurity protocols. Majority (53.7%) considered their veterinarian's biosecurity opinion most important, and 37% were not concerned about African swine fever. Almost all (90.7%) felt confident they could contain an outbreak on their farms. However, none practiced enhanced biosecurity. The cluster analysis identified four distinct producer profiles (K = 4). Cluster A had young, inexperienced producers operating breeding facilities, with moderate biosecurity adoption. Cluster B included young, small-farm producers with variable biosecurity practices and low mortality rates. Cluster C comprised farms with moderate experience, higher mortality rates, and the lowest biosecurity adoption. Cluster D was composed of older, experienced, educated producers with the highest biosecurity standards and lowest mortality rates. FAMD revealed clustering along human capital and resource availability dimensions. Regular biosecurity assessments, tailored recommendations, and training would improve biosecurity in the swine industry.

Keywords: biosecurity; cluster analysis; disease prevention; machine learning; producers’ behavior; swine farm.

Figure 1

Swine Producer Biosecurity Perceptions and Practices. (A) shows perceptions toward implementing enhanced biosecurity measures on farms, including motivations, perceived benefits, and conditions under which they would adopt them; (B) shows the influence of veterinarians, producers, and external pressures on farmers’ decisions to implement enhanced biosecurity measures.; (C) shows farmers’ perceptions of disease prevention, knowledge of herd health status, and views on the feasibility of implementing biosecurity measures.

Figure 2

Biosecurity Measures Implementation Level by Cluster (ordered from Least adopted to Most). The Y-axis represents the percentage of respondents within each cluster reporting adoption of a given biosecurity measure (0–100%). The X-axis lists the specific biosecurity practices assessed. Cluster A shows moderate but consistent adoption across most measures, with strengths in line of separation and rendering. Cluster B exhibits the most variable adoption. Cluster C shows consistently low adoption across most measures. Cluster D demonstrates the highest overall adoption, particularly in resource-intensive activities such as “Feed Truck Protocols” and “Daily Inspection.”.

Figure 3

The distribution of beliefs about biosecurity measures across clusters (K = 4). The Y-axis represents the biosecurity-related belief questions (variables) posed to respondents within each cluster. The X-axis reports the proportion of farmers in each cluster adopting a given biosecurity belief (0–100%). Cluster A shows moderately positive beliefs across most measures. Cluster B shows variable belief patterns (some measures receiving firm support while others become weaker endorsement). Cluster C demonstrates weaker beliefs in most biosecurity measures. Cluster D demonstrates consistently strong beliefs across most biosecurity measures (mostly scoring above 60–80%).

Figure 4

A scatterplot (the FAMD factor map) generated by plotting the factor scores for Component 1 (x-axis) versus Component 2 (y-axis). In this plot, each point on this plot represents a farm, and points are color-coded by their K = 4 cluster labels. Farms with similar profiles (in terms of age, production experience, education, farm size, and operation type) are located close together.