Symbolize a drmTMB fit (Gaussian and bivariate Gaussian, v0.1)

Builds a symbolized_model from a drmTMB fit. v0.1 covers the Gaussian location-scale fixed-effects path (with optional (1 | group) random intercepts on mu) and the bivariate Gaussian (biv_gaussian()) location-scale-correlation path. Other families and components return capability errors via capability_check().

Usage

# S3 method for class 'drmTMB'
symbolize(
  fit,
  symbols = NULL,
  units = NULL,
  context = NULL,
  ci_method = "wald",
  ...
)

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".

ci_method

Confidence-interval method passed to drmTMB::confint. One of "wald" (default, fast), "profile" (slower, more honest), or "bootstrap".

...

Reserved for method-specific extra arguments.

Value

A symbolized_model object.

Confidence intervals

The returned fixed_effects and interpretation tibbles carry a confidence band per coefficient: confint_low, confint_high, excludes_zero, plus a ci_method column recording which method produced them. The default ci_method = "wald" is fast and is what drmTMB::confint(fit, method = "wald") returns by default. Wald intervals can be too narrow when group counts are small (finite-df situations). For more honest intervals, pass ci_method = "profile" — slower but profile-likelihood-based. Satterthwaite / Kenward-Roger corrections are not implemented; when Wald looks suspicious the recommended alternative is "profile".

References

Nakagawa, S. (forthcoming). drmTMB: Distributional regression in TMB. https://itchyshin.github.io/drmTMB/

Kristensen, K., Nielsen, A., Berg, C. W., Skaug, H., & Bell, B. M. (2016). TMB: Automatic Differentiation and Laplace Approximation. Journal of Statistical Software, 70(5), 1-21.