Estimation Methods

Do not rely on asymptotic approximation alone. Licklider's Statistical Audit runs Bootstrap CI, Permutation Test, and Bayes Factor in parallel with the primary analysis — and gates export only after confirming that conclusions hold regardless of which method is used.

What Are Estimation Methods?

Confidence intervals and p-values returned by conventional statistical software depend on assumptions about sample size and distribution shape (asymptotic approximation). When n is small, the distribution is skewed, or outliers are present, those assumptions may not hold — and the reported 95% CI may in practice cover less than 90% of the true parameter.

Licklider's Estimation Methods automatically run multiple estimation approaches in parallel with the primary analysis and visualize whether they agree. When divergence exceeds the acceptable threshold, the system blocks or warns — giving the researcher the chance to address the issue before a reviewer requests it.

Four Parallel Estimation Tools

4.3.1 Bootstrap CI

Runs Bootstrap CI alongside parametric CI, flags divergence risk for small n (< 30) or skewed distributions, and warns when Bootstrap CI should be the primary estimate.

4.3.2 Permutation Test

Runs a distribution-free permutation test alongside the conventional test, displaying agreement between both p-values and flagging when they diverge.

4.3.3 Bayes Factor Supplement

Calculates BF₁₀ alongside frequentist results, quantifying evidence as strong support for H₁, evidence for H₀, or uninformative data.

4.3.4 Assumption & Robustness Guard

Runs QQ plot diagnostics, outlier influence analysis, and heteroscedasticity checks — then automatically suggests robust alternative methods when assumptions are violated.