Analyze nonlinear selection gradients (gamma) — analyze_nonlinear

Estimates quadratic and correlational selection gradients by OLS on relative fitness.

Usage

analyze_nonlinear_selection(
  data,
  fitness_col,
  trait_cols,
  fitness_type,
  binary_response_col = NULL,
  group = NULL,
  se_type = c("ols", "hc3")
)

Arguments

data: A data frame containing fitness and trait measurements.
fitness_col: A string specifying the response column for the OLS gradient model (relative fitness).
trait_cols: A character vector of trait column names.
fitness_type: A string indicating the fitness type: "binary", "continuous", "count", or "proportion".
binary_response_col: Optional string naming the raw fitness column (0/1 for binary, counts for count fitness) used for the GLM that supplies p-values. If NULL, fitness_col is treated as the raw outcome and relativised internally.
group: Optional grouping column, used as in analyze_linear_selection(): a separate intercept for each group in the GLM that supplies the p-values.
se_type: Standard errors of the least-squares gradients: "ols" (the default) or "hc3", heteroscedasticity-consistent; see selection_coefficients().

Value

A list containing the fitted nonlinear models, summaries, ANOVA tables, and VIFs.

Examples

prep <- prepare_selection_data(bumpus, "survival", c("total_length", "weight"))
fit <- analyze_nonlinear_selection(prep, "survival", c("total_length", "weight"), "binary")
#> there are higher-order terms (interactions) in this model
#> consider setting type = 'predictor'; see ?vif
#> Warning: Collinear traits (VIF above 5) may inflate the standard errors
extract_quadratic_coefficients(c("total_length", "weight"), fit)
#>            Term      Type Beta_Coefficient Standard_Error   P_Value
#> 1 total_length² Quadratic       -0.2334206      0.2111590 0.2440227
#> 2       weight² Quadratic        0.0209649      0.1628337 0.9387342
extract_interaction_coefficients(c("total_length", "weight"), fit)
#>                    Term          Type Beta_Coefficient Standard_Error   P_Value
#> 1 total_length × weight Correlational      -0.06949785      0.1585041 0.5723638