Figure 4.

Results of partial least squares structural equation model

Statistically significant paths links to total per capita burden of all classifiable directly transmitted (eg, via handshake or coughing) versus environmentally mediated infectious diseases globally are shown, with symbols representing the relevant latent variables (definitions, sample sizes, and measurement indicators for each latent variable are given in appendix pp 38–39). Red lines represent negative associations, and black lines positive associations, among the variables linked by those lines. Numbers along paths (and also path thickness) correspond to the weighted correlation coefficients which signify the strength of the association between two linked variables; total effects can be estimated by multiplying path coefficients along one or more segments, and summing across all possible paths. Total significant effects on disease burdens are summarised in appendix p 40; paths with p>0·1 were removed from the full model to produce the final model shown here (see appendix pp 42, 44). Artwork credit: N Nova.