Skip to contents

Bootstrap Confidence Intervals for 2-HT Model Parameters

Source: R/winter_2ht_boot.R
boot_winter_2ht.Rd

Computes bootstrap confidence intervals for the parameters of the 2-HT model by resampling the observed data with replacement.

Usage

boot_winter_2ht(
  x,
  nboot = 1000,
  conf_level = 0.95,
  lineup_size = 6,
  method = c("percentile", "bca"),
  parallel = FALSE,
  ncpus = 2,
  seed = NULL,
  ...
)

Arguments

x

A winter_2ht object from fit_winter_2ht(), OR count data/data frame as in fit_winter_2ht().

nboot

Number of bootstrap samples. Default is 1000.

conf_level

Confidence level for intervals. Default is 0.95.

lineup_size

Lineup size (only needed if x is count data, not a winter_2ht object).

method

Type of bootstrap CI: "percentile" (default) or "bca" (bias-corrected and accelerated).

parallel

Logical indicating whether to use parallel processing. Default is FALSE.

ncpus

Number of CPUs to use if parallel = TRUE. Default is 2.

seed

Random seed for reproducibility. Default is NULL.

...

Additional arguments passed to fit_winter_2ht().

Value

An object of class "winter_2ht_boot" containing:

boot_estimates

Matrix of bootstrap parameter estimates (nboot x 4)

ci

Matrix of confidence intervals for each parameter

original_fit

The original model fit

nboot

Number of bootstrap samples

conf_level

Confidence level used

Details

The bootstrap procedure resamples within TP and TA strata with replacement to keep both experimental totals fixed. For each bootstrap sample, the 2-HT model is re-fitted, and the distribution of parameter estimates across bootstrap samples is used to construct confidence intervals.

Two types of confidence intervals are available:

  • percentile: Uses quantiles of the bootstrap distribution

  • bca: Bias-corrected and accelerated intervals (Efron, 1987). The bias correction counts bootstrap ties with the estimate as half, and the acceleration comes from the jackknife over individual observations, centered and scaled within TP/TA strata, using at most six distinct leave-one-out response-count fits.

References

Efron, B. (1987). Better bootstrap confidence intervals. Journal of the American Statistical Association, 82(397), 171-185.

Examples

# \donttest{
counts <- c(
  n_tp_suspect = 147, n_tp_filler = 94, n_tp_reject = 141,
  n_ta_suspect = 38, n_ta_filler = 138, n_ta_reject = 206
)
fit <- fit_winter_2ht(counts, lineup_size = 6)
#> Warning: 2-HT estimate is on a parameter boundary; regular Hessian standard errors are unavailable. Bootstrap intervals also need boundary qualification.

# Bootstrap CIs
boot_fit <- boot_winter_2ht(fit, nboot = 200, seed = 123)
#> Bootstrap iteration 100/200
#> Bootstrap iteration 200/200
print(boot_fit)
#> 
#> Bootstrap Confidence Intervals for 2-HT Model
#> ==============================================
#> 
#> Number of bootstrap samples: 200
#> Confidence level: 95.0%
#> CI method: percentile
#> 
#> Parameter Estimates with Bootstrap CIs:
#>    Estimate Boot_Lower Boot_Upper Boot_SE
#> dP   0.3169     0.2580     0.3675  0.0304
#> dA   0.0000     0.0000     0.1888  0.0541
#> b    0.0273     0.0012     0.0677  0.0176
#> g    0.4452     0.4122     0.5182  0.0275
plot(boot_fit)
#> Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
#> ℹ Please use `linewidth` instead.
#> ℹ The deprecated feature was likely used in the r4lineups package.
#>   Please report the issue at <https://github.com/CGTZA2/r4lineups/issues>.

# }