Use Cases › Overview Group Comparison & ANOVA ↳ Two-Group Comparison ↳ One-Way ANOVA ↳ Two-Way ANOVA ↳ Repeated Measures Regression & Curve Fitting ↳ Linear Regression ↳ Logistic Regression & AUC ↳ Dose-Response & IC50 ↳ Mixed Models (GLMM) Survival Analysis ↳ Kaplan-Meier Workflow ↳ Cox Regression Workflow Normality & Power ↳ Pre-Study Power Calculation ↳ Normality Testing Categorical & Association ↳ Chi-Square Test ↳ Fisher's Exact Test ↳ Odds Ratio & Risk Life Sciences Applications ↳ Flow Cytometry ↳ Gene Expression / Omics ↳ Animal Study Workflow ↳ Clinical Trial Endpoints

Regression & Curve Fitting

Regression models the relationship between variables, predicts outcomes, and characterizes biological dose-response curves. Licklider handles assumption diagnostics, model validation, curve parameter reporting, and causal language guards — so your regression outputs meet journal requirements without manual post-processing.

Section covers: OLS Linear Regression · Logistic Regression · AUC-ROC · 4-Parameter Logistic (4PL) dose-response · Non-linear regression · IC50/EC50 · Linear Mixed Model (LMM) · Generalized Linear Mixed Model (GLMM) · Regression diagnostics (residuals, Cook's D, VIF)

When to Use This Section

Use Regression & Curve Fitting workflows when:

  • You want to model the relationship between a continuous predictor and a continuous outcome (linear regression)
  • Your outcome is binary (logistic regression) or count-based (Poisson / negative binomial)
  • You need to fit a biological dose-response curve and extract IC50/EC50 and Hill slope
  • Your data have a clustered or hierarchical structure requiring random effects (LMM/GLMM)

Available Scenarios

LR
Linear Regression
OLS with full diagnostics — epidemiologist modeling age–biomarker relationship (n=120). Residual plots, Cook's D, VIF reported.
LG
Logistic Regression & AUC
Binary outcome prediction with ROC — clinical researcher developing a diagnostic score with OR, 95% CI, and AUC-ROC.
DR
Dose-Response & IC50
4PL curve fitting — pharmacologist calculating drug IC50 with 95% CI and asymptote disclosure.
MM
Mixed Models (GLMM)
Clustered data with random effects — multi-site study with patient-level random intercept and site-level random slope.

Model Selection Guide

  • Continuous outcome, one predictor: OLS linear regression (check assumptions: linearity, normality of residuals, homoscedasticity, independence)
  • Binary outcome (0/1): Logistic regression → Odds Ratio, 95% CI, AUC-ROC
  • Count outcome: Poisson regression (or negative binomial if overdispersed)
  • Proportion outcome (between 0 and 1): Beta regression — do NOT use OLS (Proportion OLS Prevention guard will block this)
  • Dose-response sigmoidal curve: 4PL non-linear regression → IC50, EC50, Hill slope
  • Clustered / hierarchical data: LMM (continuous) or GLMM (non-continuous) with random effects