Compares indices from any fitted models supported by [influ()] or from pre-computed [influ_diag] objects. Models are reduced to the same index schema before plotting.
Usage
plot_compare(
fits,
labels = NULL,
year = NULL,
probs = c(0.25, 0.75),
show_probs = TRUE,
rescale = "raw",
rescale_series = NULL,
...
)Arguments
- fits
A fitted model, an [influ_diag], or a list of either. Alternatively, supply calculated [cpue_index()] objects, without mixing input types.
- labels
Optional unique, non-empty model labels. Repeated automatically generated labels are disambiguated with numeric suffixes.
- year
Optional focus-variable name. It is inferred when omitted.
- probs
Interval probabilities used when diagnostics must be calculated.
- show_probs
Show uncertainty ribbons.
- rescale
`"raw"`, or a positive numeric geometric mean.
- rescale_series
Optional series number supplying the common scale over overlapping focus levels.
- ...
Arguments passed to [influ()] when `fits` contains models.
Details
Compared indices must be on the same scale: response ratios, response differences, or link-scale contrasts. Ratio plots start at zero; difference and link-scale plots retain negative values. Inputs should describe comparable responses and focus effects; matching scales alone does not establish that the fitted models answer the same question. Fitted-model inputs retain their existing year-effect-contrast meaning. To compare expected-response indices, supply `cpue_index()` results with comparable reference populations and response definitions. Their stored intervals are used; set rescaling during calculation, not during plotting.