Symbolize a metafor rma.uni fit (meta-analysis / meta-regression, v0.13 first slice)
Source:R/symbolize-metafor.R
symbolize.rma.uni.RdBuilds a symbolized_model from an
rma.uni fit. v0.13 covers random / mixed-effects meta-analysis
with optional moderators. The fitted object's yi and vi slots
are treated as known (sampling variances are not parameters), and
the between-study heterogeneity tau^2 appears in the
variance_components tibble.
Usage
# S3 method for class 'rma.uni'
symbolize(fit, symbols = NULL, units = NULL, context = NULL, ...)Arguments
- fit
A fitted statistical model object.
- symbols
Optional named character vector mapping variable names to user-supplied LaTeX symbols, e.g.
c(body_mass = "W_i", temperature = "T_i").- units
Optional named character vector mapping variable names to units, e.g.
c(body_mass = "g", temperature = "C").- context
Optional short character description of the model, e.g.
"avian body-size location-scale model".- ...
Reserved for method-specific extra arguments.
Location-scale (rma.ls)
When the fit was created with rma(..., scale = ~ z),
symbolize.rma.uni() adds a second submodel tau2 to the
returned object. metafor parameterises the scale model as
log(tau^2_i) = alpha_0 + alpha_1 z_{1i} + ... + alpha_q z_{qi}
– log of the variance, not the SD – with coefficient family
alpha. This contrasts with the SD parameterisation that brms,
glmmTMB, and drmTMB use for their distributional location-scale
models (log(sigma_i) = gamma_0 + gamma_k z_ki). The natural-scale
reading on a metafor scale slope is therefore "tau^2_i changes
multiplicatively by exp(alpha_k) per unit of z_k" – a change in
the variance, not the SD. The two scales are related by
alpha_k ~ 2 * gamma_k (since log(tau^2) = 2 * log(tau) + const).
References
Viechtbauer, W., & López-López, J. A. (2022). Location-scale models for meta-analysis. Research Synthesis Methods, 13(6), 697-715.
Nakagawa, S., Mizuno, A., Morrison, K., Ricolfi, L., Williams, C., Drobniak, S. M., Lagisz, M., & Yang, Y. (2025). Location-scale meta-analysis and meta-regression as a tool to capture large-scale changes in biological and methodological heterogeneity: A spotlight on heteroscedasticity. Global Change Biology, 31, e70204.