This vignette demonstrates Hierarchical Linear Regression (HLR) using
the report_hlr function from the rwf package.
HLR builds regression models in sequential steps (blocks), adding
predictors at each step. This lets you assess how much additional
variance each new set of predictors explains beyond what earlier
predictors already account for — measured as the change in R² (ΔR²).
Installation instructions for rwf can be found on the rwf GitHub repository
report_hlr fits a series of hierarchical regression
models and returns a structured report with model fit statistics (R²,
adjusted R², F-change, ΔR²), standardized and unstandardized regression
coefficients, and significance tests at each step.
The example uses the built-in infert dataset, which
records infertility after induced and spontaneous abortions. The
arguments specify:
df = infert: the dataset.corlist = 8: column index of the continuous
predictor(s) to include.factorlist = 1: column index of categorical
predictor(s) treated as factors.predictor = "case": the outcome variable (case/control
status).random_effect = "case": the variable used as a random
effect in the mixed model comparison.The ΔR² column in the output tells you whether each block of predictors significantly improved model fit — a key quantity when building theoretically motivated regression models step by step.
result<-report_hlr(df=infert,corlist=8,factorlist=1,predictor="case",random_effect="case")
result## dv model fixed random call Model df AIC BIC logLik Test L.Ratio p.value ## 1 pooled.stratum base pooled.stratum ~ 1nlme::gls(model = formula(fbaseline), data = temp, method = "ML") 1 2 2119.941 2126.968 -1057.971 NA NA ## 2 pooled.stratum random_intercept pooled.stratum ~ 1 ~1 | case lme.formula(fixed = formula(fbaseline), data = temp, random = frandom_intercept, method = "ML") 2 3 2121.941 2132.482 -1057.971 1 vs 2 3.060691e-07 0.9995586 ## 3 pooled.stratum random_intercept_predictor pooled.stratum ~ case ~1 | case lme.formula(fixed = fpredictor, data = temp, random = frandom_intercept, method = "ML") 3 4 2123.936 2137.989 -1057.968 2 vs 3 5.861154e-03 0.9389750 ## 4 pooled.stratum random_intercept_slope pooled.stratum ~ case ~case | case lme.formula(fixed = fpredictor, data = temp, random = frandom_slope, method = "ML") 4 6 2127.936 2149.016 -1057.968 3 vs 4 4.047206e-08 1.0000000