Create a Hypothesis Test Object

Constructs a hypothesis test object that implements the hypothesize API. This is the base constructor used by specific test functions like lrt(), wald_test(), and z_test().

hypothesis_test(stat, p.value, dof, superclasses = NULL, ...)

Arguments

stat

Numeric. The test statistic.

p.value

Numeric. The p-value (probability of observing a test statistic as extreme as stat under the null hypothesis).

dof

Numeric. Degrees of freedom. Use Inf for tests based on the normal distribution.

superclasses

Character vector. Additional S3 classes to prepend, creating a subclass of hypothesis_test.

...

Additional named arguments stored in the object for introspection (e.g., input data, null hypothesis value).

Value

An S3 object of class hypothesis_test (and any superclasses), which is a list containing at least stat, p.value, dof, plus any additional arguments passed via ....

Details

The hypothesis_test object is the fundamental data abstraction in this package. It represents the result of a statistical hypothesis test and provides a consistent interface for extracting results.

This design follows the principle of data abstraction: the internal representation (a list) is hidden behind accessor functions (pval(), test_stat(), dof(), is_significant_at()).

Extending the Package

To create a new type of hypothesis test:

1. Create a constructor function that computes the test statistic and p-value.

2. Call hypothesis_test() with appropriate superclasses.

3. The new test automatically inherits all generic methods.

Example:

my_test <- function(data, null_value) {
  stat <- compute_statistic(data, null_value)
  p.value <- compute_pvalue(stat)
  hypothesis_test(
    stat = stat, p.value = p.value, dof = length(data) - 1,
    superclasses = "my_test",
    data = data, null_value = null_value
  )
}

See also

lrt(), wald_test(), z_test() for specific test constructors; pval(), test_stat(), dof(), is_significant_at() for accessors

Examples

# Direct construction (usually use specific constructors instead)
test <- hypothesis_test(stat = 1.96, p.value = 0.05, dof = 1)
test
#> Hypothesis test ( hypothesis_test )
#> -----------------------------
#> Test statistic:  1.96 
#> P-value:  0.05 
#> Degrees of freedom:  1 
#> Significant at 5% level:  FALSE 

# Extract components using the API
pval(test)
#> [1] 0.05
test_stat(test)
#> [1] 1.96
dof(test)
#> [1] 1
is_significant_at(test, 0.05)
#> [1] FALSE

# Create a custom test type
custom <- hypothesis_test(
  stat = 2.5, p.value = 0.01, dof = 10,
  superclasses = "custom_test",
  method = "bootstrap", n_replicates = 1000
)
class(custom)  # c("custom_test", "hypothesis_test")
#> [1] "custom_test"     "hypothesis_test"
custom$method  # "bootstrap"
#> [1] "bootstrap"