Step-down Cochran-Armitage trend test with consistent scoring

Usage

stepDownTrendTestBinom(
  successes,
  totals,
  doses,
  scoring = c("doses", "ranks", "log_doses", "equal_spaced"),
  alternative = c("two.sided", "greater", "less"),
  rao_scott = FALSE,
  phi = NULL,
  phi_method = c("simple", "trend_adjusted"),
  tarone_results = NULL
)

Arguments

successes: Vector of success counts
totals: Vector of total trials
doses: Vector of dose levels
scoring: Method for defining trend: "doses", "ranks", "log_doses", "equal_spaced"
alternative: Direction of test
rao_scott: Whether to apply Rao-Scott correction
phi: Overdispersion parameter (if NULL, estimated from Tarone test)
phi_method: simple or trend_adjusted description
tarone_results: Optional results from Tarone.trend.test for phi estimation

Value

stepDownTrendBinom object with consistent scoring A list with class "stepDownTrendBinom" containing:

method: Description of the test performed
data.name: Description of the data
p.value: Matrix of p-values for each step
statistic: Matrix of test statistics for each step
alternative: The alternative hypothesis
doses: The dose levels used
noec: The No Observed Effect Concentration (highest dose with no significant effect)
loec: The Lowest Observed Effect Concentration (lowest dose with significant effect)

Details

The step-down procedure starts with all dose groups and sequentially removes the highest dose until no significant trend is detected or only two groups remain. This helps identify the lowest dose at which a significant effect occurs.

The Cochran-Armitage test is appropriate for detecting trends in binomial proportions across ordered groups. The Rao-Scott correction adjusts for potential overdispersion in binomial data, which is common in toxicological studies.

Examples

# Example with simulated data
successes <- c(20, 18, 15, 10, 5)
totals <- rep(20, 5)
doses <- c(0, 1, 2, 5, 10)

# Run step-down test with Cochran-Armitage
result <- stepDownTrendTestBinom(successes, totals, doses)
print(result)
#> 
#> Step-down Cochran-Armitage test for trend (doses scoring) (doses scoring) 
#> 
#> data: successes out of totals at doses 0, 1, 2, 5, 10 with doses scoring 
#> 
#> Step-down results:
#>        Doses_Included Statistic      P_Value
#> Step 1 0, 1, 2, 5, 10 -5.746549 9.108323e-09
#> Step 2     0, 1, 2, 5 -4.090480 4.304814e-05
#> Step 3        0, 1, 2 -2.462659 1.379110e-02
#> Step 4           0, 1 -1.450953 1.467931e-01
#> 
#> Alternative hypothesis: two.sided 
#> NOEC: 1 
#> LOEC: 2 

# Run with Rao-Scott correction
result_rs <- stepDownTrendTestBinom(successes, totals, doses, rao_scott = TRUE)
#> Estimated phi = 8.571 using simple method with doses scoring
print(result_rs)
#> 
#> Step-down Rao-Scott corrected Cochran-Armitage test for trend (doses scoring) (doses scoring) 
#> 
#> data: successes out of totals at doses 0, 1, 2, 5, 10 with doses scoring 
#> 
#> Step-down results:
#>        Doses_Included  Statistic    P_Value
#> Step 1 0, 1, 2, 5, 10 -1.9628953 0.04965834
#> Step 2     0, 1, 2, 5 -1.3972184 0.16234791
#> Step 3        0, 1, 2 -0.8411904 0.40024125
#> Step 4           0, 1 -0.4956136 0.62016705
#> 
#> Alternative hypothesis: two.sided 
#> NOEC: 5 
#> LOEC: 10