Plot Chi-Square Tests for Categorical Associations (optionally stratified by covariates)

Conducts Chi-square tests between sets of categorical variables and visualizes the results. NOTE: Chi-square tests do not natively "adjust" for covariates. If covars are provided, this function can (optionally) run tests within strata (each combination of covariate levels), and combine p-values across strata (Fisher's method) for a single summary p-value per pair. If you need true covariate adjustment, use regression-based models (logistic/multinomial).

Usage

PlotChiSqCovar(
  data,
  predictor_vars,
  outcome_vars,
  covariates = NULL,
  Relabel = TRUE,
  Ordinal = TRUE,
  min_n = 4,
  Data = lifecycle::deprecated(),
  xVars = lifecycle::deprecated(),
  yVars = lifecycle::deprecated(),
  fdr_scope = c("matrix", "per_outcome", "per_predictor"),
  covars = lifecycle::deprecated()
)

Arguments

data

A data.frame containing the dataset.

predictor_vars

Character vector of x-axis categorical variables.

outcome_vars

Character vector of y-axis categorical variables. If NULL, uses xVars.

covariates

Optional character vector of covariate variables used for stratification (not adjustment).

Relabel

Logical; whether to use variable labels (sjlabelled) in the plot.

Ordinal

Logical; included for backward compatibility (currently unused here).

min_n

Minimum number of complete observations required for a tested association.

Data

Deprecated (since 19.15.0). Use data instead.

xVars

Deprecated (since 19.15.0). Use predictor_vars instead.

yVars

Deprecated (since 19.15.0). Use outcome_vars instead.

fdr_scope

Either "matrix" (default) or "per_outcome", passed to ApplyFDRCorrection(). "matrix" corrects across all p-values at once (historical behavior). "per_outcome" corrects separately within each outcome: outcomes are the y-axis variables (outcome_vars).

covars

Deprecated (since 19.15.0). Use covariates instead.

Value

A list with:

p

ggplot for unadjusted p-values

pvaltable

wide table of unadjusted p-values

p_FDR

ggplot for FDR-adjusted p-values

pvaltable_FDR

wide table of FDR-adjusted p-values

details

long table with diagnostics (n, warnings, strata info)

Examples

data(SampleData)
data(SampleVariableTypes)

Labelled <- RevalueData(SampleData, SampleVariableTypes)$RevaluedData

# Derive a few more categorical variables so the matrix has off-diagonal
# structure to read; self-associations are dropped.
Labelled$APOE4 <- ifelse(
  grepl("E4", as.character(Labelled$Genotype)), "Carrier", "Non-carrier")
Labelled$AgeGroup <- cut(
  Labelled$age, breaks = c(-Inf, 65, 80, Inf),
  labels = c("<65", "65-79", "80+"))
Labelled$TauTertile <- cut(
  Labelled$tau,
  breaks = stats::quantile(Labelled$tau, c(0, 1 / 3, 2 / 3, 1), na.rm = TRUE),
  labels = c("Low", "Middle", "High"), include.lowest = TRUE)

result <- PlotChiSqCovar(
  Labelled,
  predictor_vars = c("Diagnosis", "sex", "APOE4"),
  outcome_vars = c("Genotype", "AgeGroup", "TauTertile")
)
#> Warning: There were 3 warnings in `dplyr::summarise()`.
#> The first warning was:
#> ℹ In argument: `test = list(tryCatch(stats::chisq.test(XVal, YVal), error =
#>   function(e) NULL))`.
#> ℹ In group 2: `XVar = "APOE4"`, `YVar = "Genotype"`.
#> Caused by warning in `stats::chisq.test()`:
#> ! Chi-squared approximation may be incorrect
#> ℹ Run `dplyr::last_dplyr_warnings()` to see the 2 remaining warnings.

# Raw p-value associations
result$p


# FDR-adjusted associations
result$p_FDR