Multiple Correspondence Analysis of Veterinary Treatments for Lameness in NZ Dairy Cows

Descriptive analysis can provide a valuable overview of a dataset; however it is limited in the number of associations that can be detected and enumerated. Correspondence analysis can be used to summarise the associations among a set of categorical variables in a small number of dimensions, and give a low-dimensional (often two-dimensional) graphical representation of these associations. This paper describes the use of such an analysis to investigate the associations between factors in a dataset based on lameness diagnoses and treatments made a veterinarian in New Zealand.
Only the three most common lameness diagnoses were used in this analysis. These were white line disease (WLD) (935 cases), sole injury (sole) (599 cases) and axial wall lesions (Axial) (242 cases). These outcome variables were then analysed with the predictor variables: foot (front or hind), breed (Jersey or Friesian) and claw (lateral or medial). The correspondence analysis (Figure 1) showed that white line disease was closely associated ...