Trend estimate for linear and logit models

The Wald test and CI
Likelihood ratio test
The Pearson goodness-of-fit test
Likelihood ratio (deviance) goodness-of-fit test

Described in Chapter 5 "The Ordered rx2 Table"

Trend_estimate_CI_tests_rx2(
  n,
  a = seq_len(nrow(n)),
  linkfunction = "logit",
  alpha = 0.05
)

Arguments

n: the observed counts (an rx2 matrix)
a: scores assigned to the rows
linkfunction: Link function for the binomial distribution see ?family for more details
alpha: the nominal level, e.g. 0.05 for 95% CIs

Value

An object of the contingencytables_result class, basically a subclass of base::list(). Use the utils::str() function to see the specific elements returned.

Examples

# Alcohol consumption and malformations (Mills and Graubard, 1987)
Trend_estimate_CI_tests_rx2(mills_graubard_1987, 1:5)
#> Wald test:                    P = 0.17600, T = 1.353
#> Likelihood ratio test:        P = 0.18529, T = 1.755 (df = 1)
#> Pearson goodness-of-fit test: P = 0.12812, T = 5.682 (df = 3)
#> LR (deviance) test:           P = 0.21704, T = 4.447 (df = 3)
#> Trend estimate and Wald CI:   betahat = 0.2278 (95% CI -0.1022 to 0.5578)

# levated troponin T levels in stroke patients (Indredavik et al., 2008)
Trend_estimate_CI_tests_rx2(indredavik_2008, 1:5)
#> Wald test:                    P = 0.07549, T = -1.777
#> Likelihood ratio test:        P = 0.07715, T = 3.124 (df = 1)
#> Pearson goodness-of-fit test: P = 0.00777, T = 11.889 (df = 3)
#> LR (deviance) test:           P = 0.00930, T = 11.501 (df = 3)
#> Trend estimate and Wald CI:   betahat = -0.1828 (95% CI -0.3844 to 0.0188)